×
Evaluate Your Team's AI Readiness

Take our official assessment and explore training paths to build secure, compliant AI workflows.

Access Portal

AI at Work Podcast

Ep #4 AI Agents, Chaos Workflows & Why Sora Is Ruining Our Lives (In the Best Way)

AI at Work, AI in Business

Audio episode

Listen to this episode

Summary

In this episode of AI at Work, Elijah and Kevin dive into the shifting terrain of automation and AI-powered agents. They highlight how traditional “if this, then that” workflows are giving way to dynamic systems driven by large language models, platforms like n8n enable you to describe what you want done, not just how. They share two real-world use cases, one clutter-clearing example (categorizing home-repair receipts) and one high-value business trigger (automated meeting-prep briefs). The conversation underlines the big leap: moving from rule-based automations to agent architectures that can decide, adapt, and act.

Takeaways
• Automation still matters, but the magic happens when an LLM steps into the workflow and makes decisions.
• Platforms like n8n now bridge no-code visual workflows with AI agentic functionality. n8n+1
• Start small: find one tedious, repetitive task you do and automate it today. Build confidence before scaling.
• Be aware of guardrails: data access, security, and scope still matter when your agent can act autonomously.
• The shift is real: business value lies not just in moving data, but in orchestrating flows where parts of the system think.

Chapters
00:00 Hectic mornings, Google Docs & teleoperated Sora robots
02:30 Playing with Sora & falling into the AI video rabbit hole
04:40 Bengal cat chaos & the dark side of AI gen videos
07:10 AI-native kids & shifting expectations about tech
09:30 Smart speakers, home automation hell & the orchestration gap
13:45 The Midwit meme & the $16 wired earbud breakthrough
17:20 Agents vs automations: what’s the actual difference?
24:30 Picking a platform, connectors, APIs & pricing models
45:20 Real-world agents: sales prep & everyday automations
01:01:45 Agentic browsers & letting AI work inside your apps

 

 

 

Explore practical AI implementation: https://assessment.ascendlabs.ai/

Book a conversation with Kevin: tidycal.com/kevinwilliams

Show Notes & Links: https://www.spark6.com/podcast

Submit listener questions:

[email protected]

[email protected]

Check out Kevin’s stuff:

AscendAI: https://ascendlabs.ai/

Follow Kevin on LinkedIn: https://www.linkedin.com/in/kevinguywilliams/

Check out Eli’s Stuff:

SPARK6 Agency: https://www.spark6.com/

Sign up for FREE AI Framework Friday Newsletter: https://www.spark6.com/newsletter

Follow Elijah on LinkedIn: https://www.linkedin.com/in/elijahszasz/

Transcript and citation details

Transcript

00:00 Play from here Welcome to AI at Work.

00:06 Play from here I'm Elijah Saas.

00:07 Play from here I'm a serial entrepreneur with 28 years of experience in tech, and today I run an agency

00:12 Play from here that builds AI solutions.

00:13 Play from here Hey, I'm Kevin Williams.

00:15 Play from here I spend most of my time helping companies figure out how to actually use AI, not in

00:20 Play from here theory, but in the work itself.

00:27 Play from here What's up, Kevin?

00:28 Play from here Hey, how's it going this morning?

00:29 Play from here It's going okay, just the usual hectic morning of running around the kids to school and all

00:34 Play from here that stuff.

00:35 Play from here And yeah, and then I love opening up the Google document that we have to just keep track of

00:41 Play from here all of our ideas all week, and by all week, I mean 30 minutes before we actually record

00:45 Play from here there's all the new notes.

00:47 Play from here That's just how it works, right?

00:50 Play from here Yeah, I helpfully marked all of the things that you're supposed to say in orange.

00:55 Play from here So lest you think that Eli is just super clever, I'm actually scripting him as if he's

00:59 Play from here my own sock puppet robot.

01:01 Play from here Totally.

01:02 Play from here Now this is your cue to ask me about this.

01:05 Play from here You know what I think is even better is we could just go through our text thread from

01:08 Play from here the week, and that's the notes.

01:11 Play from here We could just talk about you struggling with Sora.

01:14 Play from here That could be the whole show.

01:18 Play from here What were you trying to do again with Sora?

01:22 Play from here Like what awful thing were you trying to make it do that it refused to do?

01:25 Play from here Well, the thing that actually this happened after our recording last week, I just got

01:30 Play from here kind of obsessed with this idea of teller operated humanoid robots and just sort of

01:36 Play from here the bizarre circumstances.

01:37 Play from here As one does.

01:38 Play from here As one does, you know, on a Tuesday afternoon, and just sort of the bizarre circumstance

01:44 Play from here of the fact that you're going to have these like offshore and onshore teams that are controlling

01:48 Play from here these things, doing the most mundane things in people's houses.

01:53 Play from here But people are weird.

01:55 Play from here And you know that they're going to ask these robots to like do things that, you know, maybe

02:00 Play from here are a little bit kooky.

02:01 Play from here And there's going to be like some middle aged guy in, let's just say, the Philippines who's

02:07 Play from here operating this thing.

02:09 Play from here And yeah, I had to I had to visualize that with Sora at one point this week.

02:14 Play from here Okay, wait a minute, translate that into the visualization with Sora.

02:17 Play from here What does that mean?

02:18 Play from here What were you trying to get Sora to do?

02:19 Play from here I was trying.

02:20 Play from here I created a quick Sora video that had two humanoids having this really bucolic home

02:28 Play from here scene where they were hugging each other.

02:30 Play from here And then it zoomed out.

02:32 Play from here And there were two guys with big mustaches who sort of looked at each other and then

02:35 Play from here shrugged and were like, with VR headsets on, with VR headsets, like I showed it to some

02:41 Play from here people.

02:42 Play from here It was good.

02:44 Play from here It is the only thing that I've actually posted on Sora.

02:47 Play from here So by all means, follow me at Ascend Labs and you too can see my one Sora post.

02:52 Play from here Have you spent quite a bit of time on Sora?

02:55 Play from here What are your feelings about it?

02:56 Play from here Oh, I think it's amazing.

02:58 Play from here I'm not a big like social media tool person, but the cameo feature where it takes your

03:05 Play from here own face or it can take like your pet or whatever it is and put them into a superhero costume

03:11 Play from here is just fun.

03:13 Play from here It's a little bit of Uncanny Valley, but it's, it's really, really good.

03:18 Play from here It's real.

03:19 Play from here However, it's so much more real than the other stuff that to date.

03:23 Play from here And there's such continuity in it.

03:25 Play from here It is only like 10 seconds long, but my own use, and I've probably created a hundred Sora

03:30 Play from here videos of like hangs out in shame type thing is that I download them and really share them

03:36 Play from here with my family.

03:37 Play from here And my teenagers are just completely disgusted with me at this point because I keep sending

03:42 Play from here them like cheesy dad type Sora videos, you know, I've intentionally tried to stay off

03:47 Play from here of it.

03:48 Play from here I think that, you know, the people who always say, Oh, I know the tricks that you're playing

03:53 Play from here on me end up being the ones most susceptible to those tricks.

03:56 Play from here And I think with all of the dirty social media algorithms, I'm one of those people.

04:00 Play from here I know exactly what you're doing to me.

04:02 Play from here And now I'm just going to scroll through Tik TOK for the next four hours of my life, knowing

04:06 Play from here exactly what you're doing to my brain and how you're writing it.

04:09 Play from here And with Sora basically layering on all of that evil short form, I'm just going to hijack

04:16 Play from here your attention for hours and hours type thing with the shock and awe of AI and how real

04:23 Play from here it looks and the novelty, man, it just gets in my brain.

04:26 Play from here I really have intentionally just try to stay away from it because I will just watch hours

04:31 Play from here of AI slop like it ain't nobody's business.

04:33 Play from here I'll just keep on going down that path.

04:36 Play from here I'll admit I'm pretty resilient.

04:37 Play from here I will scroll a little bit, but there are only so many times you can watch the Queen

04:41 Play from here of England like doing a rap or eating Jamaican jerk chicken.

04:44 Play from here I'm just not following enough people.

04:45 Play from here Something like that.

04:46 Play from here Well, I'm not using it as a social network.

04:49 Play from here I'm actually just playing with the thing.

04:51 Play from here And from that perspective, it's fun.

04:53 Play from here Yeah, it is.

04:54 Play from here It is so fun with the kids like and I wonder when the novelty of this is going to wear

04:58 Play from here off.

04:59 Play from here I guess.

05:00 Play from here Cool.

05:01 Play from here Another AI Gen video.

05:02 Play from here But so last night we have this Bengal cat that the family loves and it looks like a

05:07 Play from here mini leopard.

05:08 Play from here You've seen this little kitty.

05:10 Play from here Kids absolutely love it and it may or may not kill lots of birds even when confined

05:14 Play from here to the backyard.

05:15 Play from here It's just a little hunter and I'm like, that'd be funny to make a Sora video of Neo taking

05:20 Play from here down like a bald eagle that gets too close to the backyard or something.

05:24 Play from here And so I gen it and it looks so good and the kids are just cracking up and it looks amazing.

05:31 Play from here And he also does this thing where he'll camp out around our house, wait for me to come

05:35 Play from here by or anybody to come by and then just attack them like a wild cat, but in a nice way.

05:40 Play from here Not in a I'm going to maul you kind of way, but with his claws retracted, batting at your

05:44 Play from here legs.

05:45 Play from here And I said, oh, that'd be such a good one if I came home and then the cat came and jumped

05:51 Play from here on me, but it knocked me over this time.

05:53 Play from here And so I put the prompt into Sora, but this time instead of Bengal cat, it just assumed

05:59 Play from here I was talking about a Bengal tiger.

06:02 Play from here And so I gen the video, the kids are waiting and I walk in the door and it looks so real.

06:06 Play from here And then this tiger just attacks me and starts mauling me and my daughter is freaking out.

06:16 Play from here The dark side of generative AI videos.

06:19 Play from here I don't know.

06:20 Play from here You went pretty quickly from I couldn't possibly use it to I'm making cat videos with it.

06:24 Play from here This is why Kevin, this is why it's a slippery slope.

06:29 Play from here It's either I'm all on the gas pedal or I don't touch it.

06:32 Play from here And so that's why I stay off reels and tick tock.

06:36 Play from here And when Sora was introduced, the light bulb went off in my head.

06:39 Play from here At this point, I think that most people have seen it.

06:42 Play from here But if you haven't, you really do need to see it.

06:45 Play from here It lends itself to all kinds of larger questions about sort of our perception of media and

06:52 Play from here just how good it's getting.

06:55 Play from here And we'll say this almost every week.

06:58 Play from here This is the worst it's ever going to be.

07:00 Play from here And this is astonishing already.

07:03 Play from here When you just roll back the clock by a couple of years and see what the early video generators

07:07 Play from here look like versus this, I'm sure it can be a little bit frustrating now and again.

07:13 Play from here But wow, it is super cool from a tech perspective.

07:17 Play from here And no wonder Hollywood's worried.

07:18 Play from here Oh, for sure.

07:20 Play from here It's funny, too, because showing this to a five-year-old and what her experience with

07:25 Play from here tech at large has been or her expectations after I give it the prompt and she's waiting

07:31 Play from here on the edge of her seat to see what this video looks like, it's like, isn't it done?

07:36 Play from here Isn't it done?

07:37 Play from here Where is it?

07:38 Play from here Where is the video?

07:41 Play from here My wife's in the room, too, looking at this whole thing unfold.

07:44 Play from here And I'm trying to explain how many hundreds of hours it would have taken some person with

07:51 Play from here a camera and a video editor and a special effects person to make this tiny little clip

07:57 Play from here that we're waiting a full maybe two minutes for it to generate.

08:01 Play from here And yeah, I just went right over her head.

08:04 Play from here She doesn't.

08:05 Play from here She doesn't know.

08:06 Play from here She doesn't care.

08:07 Play from here So it's so crazy to think about just the perspective and the expectations of these generations

08:12 Play from here that are coming up that are going to be AI first generations, right?

08:17 Play from here Oh, absolutely.

08:18 Play from here You know, it's the boiling frog phenomenon.

08:21 Play from here Like here we are, even you and I, we're living it.

08:25 Play from here And it's a far cry from those early days with GPT-3-5 and how neat it was and how it can

08:34 Play from here write a sonnet to, why aren't you outputting my 80-page document immediately?

08:39 Play from here Totally.

08:40 Play from here Totally.

08:41 Play from here I mean, we've had this conversation and the number of angry comments I've shouted into

08:46 Play from here my microphone at these different models when it doesn't give me exactly what I want on

08:51 Play from here the first prompt.

08:53 Play from here And you know, I don't know if this is true, but you're like, oh, I'm going to be very

08:57 Play from here nice to my future robot overlords.

08:59 Play from here I will never curse at them or yell at them.

09:03 Play from here Maybe I have some anger management issues, but yeah, yeah, I've already come into that.

09:07 Play from here I expect you to do exactly what I said the first time and we're just not there yet.

09:15 Play from here It is listening.

09:16 Play from here It definitely is.

09:17 Play from here I don't know if you have the new generation of Alexa in your household, but it now will

09:22 Play from here make opinions about like your antacid that you've added to your shopping list.

09:27 Play from here Hope you feel better, Kevin.

09:28 Play from here I'm like, ah, no, that just got filed away into Amazon somewhere.

09:34 Play from here This is going to be a whole episode because I've been going down the home automation rabbit

09:38 Play from here hole and everybody is in that in-between stage right now where it's a partial rollout or

09:43 Play from here you can have this part of Gemini for Google Assistant, but not this part yet.

09:47 Play from here But this is where you can sign up for early access.

09:50 Play from here And so it's all in this weird, either it's the super old school, you need to know the

09:55 Play from here exact phrase that you put into your automation in order to make your lights go on and off

10:02 Play from here in the right room, or you have these half integrated generative AI models in there.

10:10 Play from here So none of it is working well right now.

10:12 Play from here I've almost had to just give up and put the whole thing on pause.

10:16 Play from here I really do want to do a home automation thing with AI at some point.

10:19 Play from here I have to laugh a little bit because I have various home automation hubs and boxes.

10:23 Play from here I was doing like some cleanup over the weekend and I was like, wow, I have all of these smart

10:27 Play from here outlets and all of these things.

10:29 Play from here I had ambition to do all of these automations and just sort of threw up my hands.

10:34 Play from here And at this point, it's sort of knowing that over the hill, I'm not quite sure how far

10:39 Play from here over the hill, we're going to have more orchestration layers to a lot of this.

10:45 Play from here That's going to be, it's going to start to sort of normalize those things.

10:49 Play from here You might be using some janky Chinese switch, which would totally be my MO.

10:56 Play from here And it's almost impossible to use.

10:57 Play from here Kasa is a big one.

10:58 Play from here That's really janky.

10:59 Play from here Kasa, I have some Kasa.

11:00 Play from here I have some other stuff.

11:01 Play from here TP-Link.

11:02 Play from here Yes.

11:03 Play from here But at the end of the day, that's a standard, right?

11:06 Play from here So the switch is using a standard.

11:09 Play from here And even if the standard is a little bit janky, an AI system can map to it such that you should

11:16 Play from here pretty soon.

11:17 Play from here I mean, I imagine this is going to be a future in Alexa or Google Home.

11:21 Play from here You should be able to say, hey, find that random device that is on my network and let's

11:26 Play from here talk through how I want to set it up, as opposed to the app and the third-party app and enabling

11:32 Play from here the skill.

11:33 Play from here And it's just, it's a mess right now.

11:34 Play from here All of these things.

11:35 Play from here And I mean, some of it might turn out to be garbage, but I really think that that just

11:41 Play from here idea of orchestration is going to creep up on us in a lot of different ways that people

11:47 Play from here aren't necessarily thinking about.

11:49 Play from here And you know, my whole thing about auditory communication is also going to play in.

11:54 Play from here These things are so finicky to, as you were saying, download the app, configure this,

11:59 Play from here whatever it is.

12:00 Play from here But if you could verbally troubleshoot your way through, the experience just gets so much

12:06 Play from here better for everyone.

12:08 Play from here And that's just on the consumer front.

12:10 Play from here Think of how many, how much like industrial machinery and things like that are out there

12:14 Play from here that if you could basically have a conversation with it, as far as calibration, everything

12:21 Play from here just gets easier and better.

12:23 Play from here So there are good things heading our way in that respect, for sure.

12:26 Play from here And you're just talking about setup, right?

12:28 Play from here Then there's the actual day-to-day use where if you don't remember the specific command

12:32 Play from here with Google Home or Alexa, you just got to keep guessing, wait, were those called the

12:37 Play from here lower floor lights or the basement lights?

12:40 Play from here And if you don't get the name right, it just says, sorry, I can't help you with that.

12:43 Play from here Or I don't know what you're talking about.

12:45 Play from here With generative AI powering this, you could just say that, hey, it's kind of dark in the

12:49 Play from here basement and it'll turn on the lights.

12:52 Play from here That's the way it's working or promised to work right now.

12:55 Play from here And you can do this with the open source Home Assistant platform and you can power it with

13:00 Play from here GPT-5, for example.

13:03 Play from here But it requires running a little server in your house and all sorts of things most people

13:08 Play from here who aren't complete nerds like me won't do.

13:10 Play from here Last count, I had 85 IP-enabled devices in my house that are running on this automation

13:17 Play from here network.

13:18 Play from here So making the change is a big lift.

13:20 Play from here That's why I'm sitting it out right now.

13:22 Play from here I started to play with some stuff over the weekend and then realized where this was going,

13:26 Play from here kind of the direction of Sora, where this is just going to consume my life.

13:29 Play from here So I had to stop.

13:30 Play from here So there's going to be at least one person in our audience who's going to be like, hey,

13:35 Play from here I'd love to just sit in front of your house and see if I can mess with your lights with

13:39 Play from here like 80 different IP addresses.

13:41 Play from here I'm sure you have that fully locked down from a security perspective.

13:44 Play from here I actually really do.

13:45 Play from here I really do.

13:46 Play from here You might.

13:47 Play from here Most consumers won't.

13:48 Play from here No.

13:49 Play from here What's your Wi-Fi password?

13:50 Play from here Password.

13:51 Play from here Password.

13:52 Play from here One.

13:53 Play from here Well, speaking of this conversational AI, I also wanted to thank you for a good recommendation.

14:04 Play from here And this one is, have you ever seen the midwit meme?

14:07 Play from here Do you know what that is?

14:09 Play from here The midwit?

14:10 Play from here Okay.

14:11 Play from here So it's these three guys.

14:12 Play from here And on the left, you have the dunce, the novice, the guy who just doesn't really know what

14:17 Play from here he's doing.

14:18 Play from here He's looking for the simplest solution.

14:19 Play from here In the middle, you have the guy who is just going to overcomplicate the crap out of the

14:24 Play from here entire process, try to optimize it to the ninth, on and on and on.

14:28 Play from here And then on the far right, you have the guy that looks like the one on the far left, but

14:32 Play from here he's the Jedi.

14:33 Play from here He's the total master who just has this whole thing nailed, and he's the best of the best,

14:39 Play from here but he's doing the

14:39 Play from here Same thing as the novice, right?

14:42 Play from here So that's the midwit meme,

14:43 Play from here is the midwit is the one who gets caught up

14:45 Play from here in the complexity.

14:47 Play from here This is a really good meme to know

14:48 Play from here as you kind of go on this episode

14:49 Play from here and talk about agents and automation workflows,

14:52 Play from here because you can definitely fall into the midwit meme.

14:54 Play from here But our last episode,

14:55 Play from here we were talking about this conversational AI

14:57 Play from here and how one of my biggest complaints was,

14:59 Play from here if I'm not at my home set up

15:01 Play from here with all my fancy stuff and monitors and microphone,

15:03 Play from here and maybe in a noisy coffee shop,

15:05 Play from here I find myself either crouched down

15:07 Play from here onto my laptop microphone,

15:09 Play from here trying to whisper into it,

15:12 Play from here or using one of these wireless earbuds

15:15 Play from here that pick up the noise of the entire cafe

15:17 Play from here or airplane or wherever I am,

15:19 Play from here and it just doesn't work.

15:20 Play from here Also, when you go wireless,

15:22 Play from here there's this awful latency

15:23 Play from here where it ends up missing the first five words that you say,

15:27 Play from here then you have to go back and type it in.

15:28 Play from here Point being is,

15:30 Play from here if you listen to the last episode,

15:31 Play from here when you get used to a 400% increase

15:35 Play from here in your data input into your machine,

15:38 Play from here it's really tough to go back.

15:40 Play from here It's really tough to go back.

15:41 Play from here So I was telling you, Kevin,

15:43 Play from here about how, oh, I've got the one AirPod

15:45 Play from here I'm putting in my mouth trying to talk,

15:48 Play from here but it's not really working.

15:49 Play from here And on my last business trip,

15:51 Play from here I tried bringing my fancy wireless microphone

15:53 Play from here with a receiver that plugs in,

15:55 Play from here the latency was awful.

15:57 Play from here And you said, why don't you just get a pair

15:59 Play from here of those cheap Apple wired earphones

16:02 Play from here that have the microphone

16:03 Play from here that you can just hold up and talk into.

16:06 Play from here And they're like 16 bucks on Amazon.

16:08 Play from here And I thought, huh, that's kind of a pretty good idea.

16:12 Play from here They came yesterday.

16:13 Play from here I went to a coffee shop.

16:14 Play from here Dude, it worked amazing.

16:15 Play from here It was perfect.

16:16 Play from here I was like, it was such a perfect,

16:17 Play from here like this is the midwhip meme, right?

16:19 Play from here Like sometimes you just get the old janky version

16:25 Play from here of the thing that you know it works

16:27 Play from here and it's super solid.

16:29 Play from here And yeah, I've had some workflow automations

16:33 Play from here and agents that have done the same thing.

16:35 Play from here I think you just called me a midwhip, but.

16:38 Play from here I was a midwhip.

16:39 Play from here I was the one with the seven wireless microphones

16:42 Play from here trying to find this perfect setup.

16:43 Play from here And you said, why don't you just buy the ones

16:46 Play from here that have been out for like 20 years and use those?

16:49 Play from here I'm like, oh, yeah, that works.

16:51 Play from here A quick reminder that we called out WhisperFlow

16:54 Play from here on the last couple of episodes.

16:56 Play from here Super Whisper is another one that helps amplify

17:00 Play from here how you're interacting with anything

17:02 Play from here just by adding LLM transcription.

17:04 Play from here And it's still just one of my favorite tools for sure.

17:07 Play from here Every day.

17:08 Play from here Yep.

17:09 Play from here No affiliation, by the way.

17:10 Play from here There's no affiliation.

17:11 Play from here It's just, you know, big hug to cool technology.

17:14 Play from here Yeah, yeah.

17:15 Play from here Well, speaking of big, cool technology,

17:17 Play from here we made a promise to disclose a favorite

17:21 Play from here workflow automation or agent

17:23 Play from here or probably a good place to start.

17:25 Play from here What the hell is the difference, Kevin?

17:27 Play from here These words get thrown around all the time

17:28 Play from here and kind of like AGI or super intelligence,

17:31 Play from here everybody seems to have their own definition.

17:34 Play from here So how would you separate out an agent

17:38 Play from here or agentic AI from a automation or a workflow automation?

17:43 Play from here So you let marketers get their hands on anything

17:45 Play from here and it gets really, really complicated in a hurry.

17:48 Play from here And I would say that a lot of people

17:52 Play from here have been talking about agents like 2025

17:55 Play from here being the year of the agent, but you're right.

17:57 Play from here Like what the heck is an agent?

17:59 Play from here And there are a lot of different definitions out there.

18:01 Play from here And that definition is going to slide over time.

18:05 Play from here But the way that I think about it

18:07 Play from here is that an agent is something that executes a task

18:12 Play from here in some way, shape or form.

18:13 Play from here You basically put your hypothetical dime in

18:16 Play from here on the left side and on the right side,

18:19 Play from here something cool and perhaps useful pops out.

18:22 Play from here And it will accomplish that

18:24 Play from here through perhaps really a bunch of complicated means.

18:27 Play from here But at the end of the day, that's what you're after

18:30 Play from here is an input and an output

18:33 Play from here that might be part of a larger process.

18:35 Play from here Okay, well, how's that different from a workflow?

18:38 Play from here People like us have been doing stuff in Zapier

18:40 Play from here and make for like a decade that essentially,

18:44 Play from here oh, I want to move this file from here to there

18:46 Play from here and run some code on it or whatever it is.

18:49 Play from here I'm putting my coin in on the left

18:51 Play from here and I'm getting something out on the right.

18:53 Play from here The difference what we have now is agents add an LLM layer

18:58 Play from here that can have decision-making in the middle of it somewhere.

19:01 Play from here So it's transforming that data in a way

19:05 Play from here that's often much lighter weight

19:08 Play from here than it would have been in previous generations,

19:11 Play from here such that you can be much, much, much more dynamic

19:14 Play from here with the output on the other side.

19:17 Play from here And then you have to think of it

19:19 Play from here in terms of agents playing a role

19:23 Play from here in multi-agent workflows at the end of the day.

19:27 Play from here So what we were talking about just a few minutes ago

19:30 Play from here as far as the switching,

19:32 Play from here what's still missing is this orchestration layer

19:36 Play from here of an agent that understands all of the agents

19:39 Play from here such it knows which is the right one to use

19:43 Play from here in what circumstance.

19:44 Play from here Oh, they need something off of LinkedIn.

19:47 Play from here So I need to use the agent that knows how to go into LinkedIn

19:50 Play from here and get my direct messages.

19:52 Play from here Oh, it needs the agent that is going to create

19:55 Play from here a rag database that stores a bunch of documents.

19:58 Play from here I need to call that one instead.

20:00 Play from here Instead, at the moment,

20:02 Play from here we're still basically duct taping things together

20:06 Play from here in workflow software.

20:07 Play from here And Zapier and Make are still solid options,

20:10 Play from here but as we've mentioned before,

20:11 Play from here at least my go-to here as a non-coder,

20:14 Play from here and that's important to note, is N8n

20:17 Play from here because I can create these individual agent flows

20:22 Play from here and then stitch them together

20:24 Play from here as far as a larger, essentially self-orchestrated layer

20:28 Play from here such that I can accomplish whatever it is that I'm after.

20:33 Play from here That orchestration layer is happening.

20:35 Play from here It's well-known that that's what people are after,

20:38 Play from here but when people say agent generally

20:41 Play from here from a functional perspective,

20:43 Play from here they're talking about these AI augmented workflows.

20:47 Play from here Yeah, so it might even be fair to say

20:49 Play from here that these automation workflows

20:52 Play from here that have been around forever and ever

20:54 Play from here are working on sets of conditional logic.

20:57 Play from here If this happens, then do this,

20:59 Play from here but if this happens, then do this.

21:01 Play from here And so kind of like N8n

21:04 Play from here and a lot of these other very popular building platforms,

21:07 Play from here you do get this visual interface

21:10 Play from here of what these different nodes look like,

21:12 Play from here and that part's familiar.

21:14 Play from here You have this visual workflow

21:15 Play from here where you see these branches

21:17 Play from here of if these different things could happen,

21:19 Play from here but do you think it's fair to say

21:20 Play from here that the real difference

21:22 Play from here between these automation workflows

21:24 Play from here as we've known them with things like Zapier,

21:27 Play from here depending on how you like to pronounce it,

21:29 Play from here would be that instead of just conditional logic,

21:33 Play from here you have this fuzzy logic that is being powered by an LLM.

21:37 Play from here It's almost as if an LLM was interjected

21:40 Play from here into the workflow automation

21:42 Play from here and it's now making decisions on the fly

21:45 Play from here that the older automations couldn't make before.

21:48 Play from here Is that really the difference

21:50 Play from here between an agent and an automation?

21:53 Play from here Is just the interjection of an LLM

21:55 Play from here somewhere in that process?

21:56 Play from here I'm going to quibble just a little bit

21:58 Play from here in that I think that you're not wrong at all,

22:03 Play from here but it's almost the simplification of it

22:07 Play from here and the accessibility of it

22:09 Play from here that makes it deployable right now.

22:11 Play from here Pretty much everything we've described

22:13 Play from here could have been done

22:15 Play from here through more traditional development previously,

22:18 Play from here but because it was development,

22:21 Play from here it was a whole deal to do things

22:24 Play from here as opposed to almost having a conversational layer to it

22:27 Play from here that kind of, it rounds off the edges

22:29 Play from here the same way we were talking about the switches

22:32 Play from here and that allows you to deploy these things a lot faster.

22:36 Play from here You're not, again, you're not wrong

22:38 Play from here because the app, the applied technology

22:40 Play from here is taking the place of all of that code,

22:43 Play from here but for our listeners,

22:46 Play from here knowing that these things are so much more accessible

22:49 Play from here because they're buildable by AI

22:51 Play from here and they have an AI step in it

22:53 Play from here that just makes it so much easier to use,

22:56 Play from here it opens up this huge world of possibility

22:59 Play from here of things that previously would have just been like,

23:02 Play from here why bother?

23:03 Play from here Like I've got so-and-so on my staff

23:05 Play from here that deals with this horrible thing

23:07 Play from here and that's why we pay them the big bucks

23:09 Play from here to deal with this horrible thing.

23:11 Play from here And now you can totally reassess that

23:15 Play from here in a way that you don't have to call on a dev group to solve.

23:18 Play from here I think that's a really important point

23:19 Play from here is the fact that previously,

23:23 Play from here let's use Zapier as an example.

23:25 Play from here You could set up all these visual nodes

23:27 Play from here and build out your workflow automation,

23:29 Play from here but with many of these other platforms like N8n or Lindy,

23:34 Play from here you can just put in a prompt

23:36 Play from here of the thing that you're trying to accomplish

23:38 Play from here and then the platform builds it for you,

23:41 Play from here which sometimes works, sometimes doesn't,

23:43 Play from here we'll get into that.

23:44 Play from here But that is a huge differentiator right there,

23:47 Play from here just like we were talking with that home automation.

23:49 Play from here The big unlock will be when you can just describe

23:52 Play from here what it is that you want done

23:53 Play from here without having to either learn about the code

23:57 Play from here or learn about that platform specific interface

23:59 Play from here and how to use it.

24:01 Play from here You can just describe what it is you want

24:03 Play from here and it should just build it.

24:04 Play from here Like that is the Holy Grail, right?

24:06 Play from here That's where so much of this is heading.

24:08 Play from here That's going to be an unlock for anybody

24:10 Play from here who isn't a developer,

24:11 Play from here but wants all the cool tools

24:12 Play from here that developers can build at their whim.

24:14 Play from here No, absolutely.

24:15 Play from here And we're close.

24:18 Play from here N8n has its AI workflow builder,

24:22 Play from here which works reasonably well.

24:25 Play from here You can't pay him for it.

24:26 Play from here It must be pretty good, right?

24:28 Play from here Yeah, it works well.

24:29 Play from here It really does work well.

24:31 Play from here It's not seamless though.

24:32 Play from here So we're getting closer.

24:36 Play from here And then we mentioned Cloud Skills last time,

24:39 Play from here installing a Cloud Skill into Cloud,

24:42 Play from here which probably we should attach that to this episode as well,

24:46 Play from here such that folks can download the Cloud Skill.

24:50 Play from here You can develop flows that way.

24:52 Play from here This is what I want to accomplish.

24:54 Play from here This is the coin I'm putting in at the front end,

24:57 Play from here and this is what I want out the backend.

24:59 Play from here Because I don't want anybody to lose track

25:03 Play from here of the importance of scope

25:05 Play from here and understanding what you're trying to do.

25:08 Play from here Because if you have a poor grasp of what's going in

25:11 Play from here and what's coming out, guess what?

25:13 Play from here You're going to get bad stuff out from bad stuff in.

25:17 Play from here Garbage in, garbage out at the end of the day,

25:19 Play from here and you're probably going to be frustrated.

25:20 Play from here So there is some education and some acclimatization

25:25 Play from here to how the technologies work.

25:27 Play from here And there's also just some experience needed

25:30 Play from here in how to define scope.

25:32 Play from here So you can't lose track of that.

25:36 Play from here You'll have better results if you think harder

25:38 Play from here about what you're trying to accomplish,

25:40 Play from here at least in those two points.

25:42 Play from here Let the system figure out some of the hows in between,

25:47 Play from here and then, but focus on what it's supposed to look like

25:51 Play from here in the output.

25:51 Play from here Well, you mentioned Zapier 2 being one of these

25:54 Play from here more old school workflow automation tools

25:57 Play from here that have been around forever.

25:58 Play from here And even these legacy platforms are adding in the ability

26:03 Play from here to hook up different models to the workflows, right?

26:07 Play from here So it then becomes a matter very often

26:10 Play from here of the more you're willing to pay for the subscription,

26:14 Play from here the easier it is going to be to get off the ground

26:17 Play from here and get something going, right?

26:19 Play from here And so you mentioning that you're not super technical,

26:22 Play from here but N8n is your platform of choice.

26:25 Play from here They've done a lot of things on that product iteration

26:29 Play from here to make it a lot more user-friendly

26:31 Play from here and let people get started faster.

26:33 Play from here Whereas before a lot of people would open that

26:34 Play from here and they take one look at the menu,

26:36 Play from here they try to hook up just one of these tools.

26:39 Play from here It doesn't work the first time they shrug their shoulders.

26:41 Play from here They're like, I don't, yeah, this is way too much.

26:44 Play from here And it has gotten so much better.

26:47 Play from here Tend to go an inch deep on 20 of these tools

26:51 Play from here because I just want to try something and experience them.

26:55 Play from here And then I circle back and say, okay,

26:57 Play from here this is the poison I'm going to pick.

27:00 Play from here I'm going to go all in on this

27:01 Play from here because as we talk about building these different workflows

27:05 Play from here or these different agents,

27:06 Play from here whatever you want to do or whatever you want to call it

27:08 Play from here at that point, if you're in a fragmented ecosystem

27:14 Play from here and you are running a business

27:15 Play from here or you're running a division in a business,

27:17 Play from here it does start to become a little complex

27:20 Play from here if something goes wrong, if you have to edit something,

27:24 Play from here wait, which one was that on?

27:26 Play from here You better have good SOPs, good documentation to go back to

27:29 Play from here because the more spread out it is,

27:31 Play from here and we've talked about this before, Kevin,

27:34 Play from here it's better to pick a platform

27:36 Play from here and really get good at that platform

27:38 Play from here than to do what I do as a techno nerd and pick all of them.

27:43 Play from here Try one thing, although I do eventually go back.

27:46 Play from here And so you've landed on N8N as your pick.

27:51 Play from here Oh, a hundred percent.

27:52 Play from here And it has a couple of different reasons.

27:55 Play from here And boy, there's some major business implications for this

27:57 Play from here as far as the ecosystem.

27:59 Play from here First is just from a cost perspective

28:02 Play from here as currently configured, it's impossible to beat.

28:05 Play from here That you can self-host it

28:06 Play from here and you're basically paying for process

28:08 Play from here as opposed to paying for actions.

28:10 Play from here And that's essentially broken

28:12 Play from here the workflow SaaS pricing model

28:18 Play from here because Zapier or Make would charge you per interaction.

28:22 Play from here And yes, there is that model in place in N8N,

28:25 Play from here but if you want to self-host it

28:26 Play from here and do a gazillion different processes,

28:29 Play from here you can and just pay for that process

28:32 Play from here as opposed to paying for the interactions.

28:34 Play from here So it's another-

28:35 Play from here Break that down a little bit, that interaction,

28:37 Play from here just so people understand,

28:38 Play from here just break that down a little bit, that interaction.

28:40 Play from here When you're paying for that interaction,

28:42 Play from here what exactly does that mean?

28:43 Play from here What are you being charged for?

28:44 Play from here So generally, if there's a trigger response

28:46 Play from here that something happens in your email,

28:48 Play from here that's your coin in,

28:50 Play from here and something happens on the other side,

28:52 Play from here as in you've moved the file to Google Drive,

28:54 Play from here whatever it is, that interaction is charging you,

29:00 Play from here you get a certain quota a month.

29:02 Play from here It's usually a few thousand in the basic accounts

29:04 Play from here and it goes up from there.

29:05 Play from here But if you're doing anything at any sort of scale,

29:08 Play from here say you're a D2C company

29:09 Play from here and you have tens of thousands of orders a month

29:12 Play from here and you're trying to do something clever with it,

29:14 Play from here you'll tap that out in a huge hurry.

29:16 Play from here It's important to point out even your own inbox, right?

29:19 Play from here You know what's too good to do, right?

29:21 Play from here If you have some automation that's looking at every email

29:24 Play from here that comes in and you're burning through one of those credits

29:26 Play from here or one of those tokens every time,

29:28 Play from here that adds up faster than you can imagine, right?

29:30 Play from here So you'll be out of space or like the old days of cell phones

29:35 Play from here when you'd get over the amount of minutes

29:36 Play from here that you were allotted, all of a sudden

29:38 Play from here you're getting this exponential upcharge as well.

29:41 Play from here It can get expensive really fast.

29:42 Play from here So the other piece is the shareability

29:44 Play from here and the portability of this stuff.

29:46 Play from here And you and I talk about this a lot,

29:49 Play from here that we build all of these cool things.

29:51 Play from here And because at the end of the day,

29:53 Play from here a workflow is essentially JSON,

29:55 Play from here the ability to download it and share it

29:57 Play from here in a frictionless environment is incredibly powerful.

30:02 Play from here Think about if you being the dev,

30:05 Play from here had developed a cool app that does X, Y, and Z

30:08 Play from here as far as like lead enrichment or whatever.

30:10 Play from here And yeah, you might be able to give your buddy Kevin

30:13 Play from here a login or something to it.

30:15 Play from here But if you wanted to give it to me in a way

30:17 Play from here that I was going to deploy it,

30:19 Play from here all of a sudden I'm gonna be spinning up Vercel

30:22 Play from here and I'm gonna be like trying to figure out

30:24 Play from here how to host this thing.

30:25 Play from here And I'm just gonna throw up my hands.

30:27 Play from here But it's literally click, click, click simple

30:31 Play from here to take somebody else's workflow and upload it into N8n.

30:36 Play from here And then yeah, you have to connect the endpoints

30:38 Play from here because they had it connected to their email

30:40 Play from here and you need to connect it to your email.

30:42 Play from here They had it connected to Salesforce

30:44 Play from here and you need it connected to HubSpot.

30:47 Play from here So there's some sort of adapting

30:50 Play from here that you have to do around the edge.

30:52 Play from here But this is gonna be a giant unlock,

30:54 Play from here particularly in the SMB community

30:57 Play from here where founders and leaders are gonna be like,

31:00 Play from here hey, yeah, cool.

31:01 Play from here I knocked together this fun little thing.

31:03 Play from here It solved a bunch of my problems.

31:05 Play from here Here you go, bud.

31:06 Play from here And off you go to the races

31:08 Play from here because I'm not really in the SaaS game.

31:11 Play from here That's not the thing that I'm too interested in.

31:14 Play from here I'm interested in solving problems.

31:17 Play from here The sorts of stuff that both of us

31:19 Play from here have sort of built casually in the last couple of months

31:22 Play from here or three, four months,

31:23 Play from here and we'll describe some of that going forward,

31:25 Play from here these would have been full-fledged SaaS companies

31:27 Play from here just a few years ago.

31:29 Play from here And we're like, meh, why would I do that?

31:31 Play from here Selling SaaS is really hard.

31:33 Play from here And since this wasn't very hard for me to build,

31:36 Play from here why on earth would I invest all the time and dollars

31:38 Play from here into trying to build some sort of brand around this thing?

31:41 Play from here And that might be wrong.

31:42 Play from here And there are plenty of people out there

31:44 Play from here that are taking these workflows

31:45 Play from here and they're putting a fancy logo on them

31:47 Play from here and they're getting out there

31:48 Play from here because they have some sort of a distribution channel.

31:51 Play from here And they're gonna get $50 a month for six months

31:55 Play from here until people realize, oh, wow,

31:57 Play from here you could have just downloaded the JSON.

31:59 Play from here I would say probably 90% of the 20-something kids

32:03 Play from here that you see on TikTok and Instagram

32:04 Play from here who are the new AI experts

32:07 Play from here who have the clickbait thumbnail on their video that says,

32:10 Play from here here's how you can make a million dollars using AI

32:13 Play from here starting your own business this year.

32:14 Play from here I'd say 90% of it is basically the thought

32:18 Play from here of reselling N8N workflows

32:20 Play from here that they didn't build to begin with

32:21 Play from here that they probably just got on the internet

32:24 Play from here as a JSON file that they imported in

32:27 Play from here and then are trying to resell them as a business.

32:29 Play from here I don't think that business model really works.

32:33 Play from here I don't know, maybe it does.

32:35 Play from here Maybe we're in the wrong business, Kevin.

32:36 Play from here Well, I mean, you could say the same about websites.

32:40 Play from here circa 2001 to 2025,

32:45 Play from here there are a lot of people out there

32:46 Play from here that make a very solid living.

32:48 Play from here In fact, you at various times

32:49 Play from here have made a very solid living off of doing this,

32:53 Play from here which is configuring and customizing essentially a website.

32:58 Play from here There are parts of that

32:59 Play from here that you're going to have to hard code

33:01 Play from here and there are parts of it that you're going to be using

33:03 Play from here like off-the-shelf sort of WYSIWYG software.

33:05 Play from here So I don't begrudge this at all.

33:07 Play from here And frankly, I think it's a huge opportunity

33:10 Play from here Service versus product though, right?

33:12 Play from here Yeah, service versus product.

33:13 Play from here Like reselling prompts or JSON files for NAD workflows

33:17 Play from here versus I'm going to help you solve this problem

33:20 Play from here and these are the tools I'm going to use.

33:22 Play from here Very different.

33:23 Play from here I think a lot of the TikTok stars that are AI experts,

33:27 Play from here little more of the latter,

33:28 Play from here a little more of the, you know,

33:30 Play from here trying to resell basic stuff

33:31 Play from here without a fundamental knowledge of it.

33:33 Play from here But who knows, who knows.

33:35 Play from here There are also rights issues and commons issues

33:38 Play from here and like all of these things that are surrounding it,

33:40 Play from here but hustlers are going to hustle.

33:42 Play from here Yeah.

33:43 Play from here That's the way of it.

33:44 Play from here And as a result in their wake,

33:46 Play from here people are creating just amazing things

33:48 Play from here that are just out there and you can grab them.

33:52 Play from here You know, you need some sort of a vector rag system

33:57 Play from here that's ingesting documents.

33:59 Play from here Okay, click, click, click, find 10 of them,

34:01 Play from here find one that's going to work for you

34:03 Play from here and adapt it to your purposes.

34:05 Play from here Like this, it's a remarkable era

34:07 Play from here of being able to just cycle skip and get the right pieces.

34:11 Play from here And, you know, internally we're building these things

34:13 Play from here that the other day it was something like,

34:17 Play from here oh, we don't really know how to figure out the cost for this

34:19 Play from here because of, you know, we're accessing this model

34:22 Play from here and that model and the other one.

34:24 Play from here So we built a token counter

34:25 Play from here of how much is this going to cost us to run?

34:28 Play from here And now we have the token counter.

34:29 Play from here So if we need that for something else,

34:31 Play from here we just add it to whatever project

34:33 Play from here and then it creates just a simple,

34:37 Play from here I think it's a G sheet and just calculates

34:40 Play from here what's the estimated cost per run

34:42 Play from here such that you know how much your fun little tool is burning

34:46 Play from here in terms of cost and, you know, resources.

34:49 Play from here That's so cool.

34:50 Play from here I mean, that would have been

34:51 Play from here a relatively complex piece of software

34:52 Play from here you would have had to build just a couple of years ago

34:54 Play from here that you can now no code, slap together,

34:57 Play from here find somebody who's already done it,

34:59 Play from here which is even better, import that JSON file in

35:01 Play from here and you are up and running.

35:03 Play from here I mean, that's worth mentioning

35:04 Play from here is that some of these N8n workflows

35:07 Play from here can be incredibly complex.

35:09 Play from here You're looking at this labyrinth of conditional logic

35:12 Play from here and all these nodes and connecting lines.

35:15 Play from here I mean, even though it's amazing

35:18 Play from here that you can do it without code

35:19 Play from here and you can have this drag and drop interface

35:21 Play from here to put all this stuff together,

35:22 Play from here it doesn't mean that it doesn't take time.

35:24 Play from here So to have that unlock of being able

35:27 Play from here to just import a code file in there,

35:29 Play from here have it all built,

35:30 Play from here and then you had mentioned putting in the connectors

35:32 Play from here and I want to talk about that.

35:34 Play from here That is just such a huge unlock.

35:37 Play from here So these connectors, okay.

35:39 Play from here I want to say the first time I used N8n

35:41 Play from here and I was benchmarking it

35:42 Play from here against some of these other automation or agent tools,

35:45 Play from here you know, Lindy is one of them.

35:47 Play from here The connectors in Lindy were so seamless, right?

35:50 Play from here They had these templates to get you started.

35:54 Play from here I think one of these templates

35:55 Play from here turned into a better version of something you built on N8n.

35:57 Play from here We're probably going to be talking about this,

35:59 Play from here but I remember the connectors were so simple.

36:01 Play from here And then I thought, huh,

36:03 Play from here I'm going to go over to N8n

36:04 Play from here and I'm going to replicate this same thing

36:06 Play from here that was a template in Lindy

36:07 Play from here and try to get it set up from scratch,

36:09 Play from here putting the nodes together and the connectors in N8n.

36:12 Play from here And I had an awful time with the connections.

36:14 Play from here They just didn't hook up as seamlessly as they did in Lindy.

36:18 Play from here And we've both been playing with these for a long time.

36:21 Play from here So have you seen or have you had trouble

36:24 Play from here with the connectors specific to N8n

36:26 Play from here or have you seen an improvement over time

36:28 Play from here as to how seamless they are to get going?

36:31 Play from here Definitely improvement over time.

36:33 Play from here In 24, we basically threw up our hands

36:35 Play from here and we're like, this is tough to do right now.

36:39 Play from here And now they're adding who knows how many hundreds

36:42 Play from here of connectors a day, a week or whatever that are seamless.

36:45 Play from here And we're not getting technical here,

36:47 Play from here but the ability to do basically anything

36:50 Play from here you can imagine through a webhook,

36:51 Play from here like if it has an online implication,

36:54 Play from here you can use a webhook to grab the information

36:57 Play from here and do something with it.

36:59 Play from here It is more advanced, but you can get around

37:02 Play from here some of those limitations.

37:05 Play from here Certain platforms are more locked down than others

37:08 Play from here as far as their native connection functionality.

37:12 Play from here So again, beginner, intermediate, advanced.

37:17 Play from here In the intermediate advanced territory,

37:20 Play from here APIs are still a thing.

37:22 Play from here So an API is making a data call to another platform

37:26 Play from here and you can usually do a lot more with an API,

37:30 Play from here which you've probably used Cloud or GPT

37:32 Play from here or something else to configure

37:34 Play from here than you can with the straight connectors.

37:38 Play from here Partially because it's a business model barrier

37:42 Play from here for a lot of the big SaaS platforms out there

37:46 Play from here that they don't really want you doing leaps and bounds

37:50 Play from here ahead of their own product development.

37:53 Play from here They have to grudgingly give you some access

37:55 Play from here because they recognize that they'll lose

37:57 Play from here a lot of people as customers

37:59 Play from here if they can't do the things that they want to do,

38:01 Play from here but nor are they just going to open up

38:05 Play from here their chest of treasures and be like,

38:09 Play from here oh yeah, do all of the things that we can't do for you

38:13 Play from here and make sure that you don't pay us

38:15 Play from here for our next iteration of our product.

38:18 Play from here Right, so APIs aside

38:20 Play from here and these more complex webhook configurations aside,

38:23 Play from here give people a very brief non-technical description

38:26 Play from here of what a connector is

38:27 Play from here as we get into building these workflows.

38:30 Play from here Connectors are simply the,

38:32 Play from here well, it's the connection between one data source

38:36 Play from here and another data source.

38:37 Play from here And often-

38:38 Play from here And it's supposed to be easy.

38:39 Play from here And often these are defined

38:42 Play from here if you were sort of thinking in terms of left and right,

38:44 Play from here like what's the coin that's going into the machine

38:47 Play from here and what's coming out of the machine.

38:48 Play from here On the left, it's going to be your triggers.

38:51 Play from here So, so many things in our daily life

38:55 Play from here are triggered by things that happen in email.

38:58 Play from here Like it's a solicitation, it's an invoice,

39:01 Play from here it's a bill, it's whatever it is.

39:03 Play from here And then what do you do?

39:04 Play from here You do something with it.

39:06 Play from here You forward it to the accounting team

39:08 Play from here or you forward it this way

39:10 Play from here and they do a transformation on top of it, right?

39:13 Play from here So one connector might be the source of information.

39:18 Play from here So something happens in a database,

39:20 Play from here something happens in email,

39:21 Play from here something happens in Slack,

39:22 Play from here something happens in HubSpot, wherever it is.

39:24 Play from here And then you go through the transformation in between

39:27 Play from here and there may be a connector in there

39:29 Play from here to say a data repository.

39:31 Play from here It might be a database.

39:34 Play from here It might be a Google Sheet.

39:35 Play from here And then the destination output on the other side.

39:39 Play from here Again, that could be a data repository.

39:41 Play from here It could be Airtable.

39:42 Play from here It could be G Suite Sheets.

39:44 Play from here It could be any number of things.

39:46 Play from here And then some sort of like a notification connector.

39:50 Play from here In my world, my nerve center is Slack.

39:53 Play from here So I like to have Slack embedded throughout my process

39:58 Play from here such that I'm not having to bounce around

40:00 Play from here between different environments.

40:01 Play from here So I connect Slack to my workflow

40:05 Play from here such that things are happening in Slack.

40:08 Play from here So just a super simple example is a transcript

40:13 Play from here that's turned into, a call transcript

40:16 Play from here that's turned into an outbound follow-up email.

40:19 Play from here There's a lot of stuff that's happening behind the scenes

40:22 Play from here but it's connecting to Slack

40:24 Play from here such that I can accept or edit right in Slack

40:29 Play from here as opposed to having to go into some Google document

40:31 Play from here and do the edits and things like that.

40:33 Play from here And then it goes back into its automation

40:36 Play from here out the other direction.

40:37 Play from here I think that's good.

40:38 Play from here And I think it's worth noting that,

40:40 Play from here so you described all of these different things

40:42 Play from here in the ecosystem.

40:43 Play from here It could be your email client.

40:45 Play from here It could be Slack.

40:46 Play from here It could be your Google Drive.

40:48 Play from here It could be the Google suite of tools like Sheets and Docs.

40:51 Play from here All of those things have log creds for your account, right?

40:55 Play from here And so the best case scenario with these connectors

40:58 Play from here is that you just put in your log creds

41:00 Play from here and now you're hooked up, right?

41:02 Play from here Once it has access,

41:03 Play from here as far as how robust these connectors are,

41:05 Play from here that's where the options come in.

41:07 Play from here That's where how far open did they peel back the kimono,

41:12 Play from here right?

41:12 Play from here Can I look at all of my documents

41:15 Play from here or just the ones in this folder?

41:17 Play from here Can I have access to all these different things

41:19 Play from here in my email client?

41:20 Play from here Or am I just going to use the trigger

41:22 Play from here of when a new email comes in, right?

41:24 Play from here So when we talk about how robust these connectors are,

41:27 Play from here it's the ease of use of setting it up.

41:29 Play from here Can I just put in my log creds and I'm in?

41:31 Play from here And then how many options,

41:33 Play from here once I'm connected to that tool,

41:34 Play from here do I have to do things with in whatever the workflow

41:38 Play from here or automation is that I want to build?

41:39 Play from here Okay, so that is a really good point.

41:41 Play from here And this is something that everybody listening

41:43 Play from here should file away,

41:44 Play from here that if you do something like create a workflow

41:49 Play from here within an N8n account that accesses, say,

41:53 Play from here your personal email or your primary business email

41:56 Play from here that does something nifty

41:58 Play from here and other team members have access to it,

42:01 Play from here they can spin up their own N8n instance

42:05 Play from here that does something as simple as every time

42:08 Play from here new email message that mentions my name,

42:11 Play from here send me a digest of whatever it is.

42:14 Play from here So you just handed your team the keys to your email.

42:19 Play from here So-

42:20 Play from here Oh, back to the guardrails.

42:21 Play from here Back to the guardrails.

42:22 Play from here Oh, back to the guardrails.

42:23 Play from here The fun killer stuff.

42:24 Play from here The fun killer.

42:25 Play from here I was waiting for him to pop up.

42:27 Play from here Here he is.

42:27 Play from here Yes, Mr. Fun Killer.

42:29 Play from here It's really easy to set up.

42:30 Play from here Oh, shake a butt in privacy.

42:32 Play from here Yes, yes.

42:33 Play from here So forwarding rules and things like that.

42:36 Play from here And yeah, there are some basic ways to do it.

42:38 Play from here You mentioned get new as a request, an email, for example.

42:44 Play from here If I have a tag,

42:45 Play from here and this is one of the examples we're going to give

42:47 Play from here of an invoice in an email,

42:50 Play from here the only thing that that connection can connect to

42:54 Play from here is new email tagged invoice.

42:58 Play from here As long as other people on the team

43:00 Play from here don't have access to change that connection,

43:03 Play from here you're pretty locked down in that case.

43:06 Play from here But the more sensitive the data that you're working with,

43:08 Play from here the more sophisticated the organization,

43:11 Play from here the more guardrails that you're going to need

43:13 Play from here to have in place.

43:15 Play from here You know, if you're a team of two or three

43:17 Play from here and you're just moving super fast,

43:19 Play from here like there's going to be a lot of informality

43:22 Play from here with this stuff.

43:23 Play from here But I think it's pretty easy to think about the ways

43:27 Play from here that this could go sideways

43:29 Play from here in terms of disclosure, data leakage, et cetera.

43:32 Play from here Well, as you mentioned,

43:34 Play from here we've been playing with this for a long time

43:36 Play from here and we have dozens of these that,

43:38 Play from here it's to the point now where

43:40 Play from here if these got taken out of my business,

43:42 Play from here I'd be really unhappy.

43:43 Play from here Like it would be bordering on catastrophic, to be honest.

43:48 Play from here Oh, we should talk about some of these workflow automations.

43:50 Play from here I don't even know where to start.

43:51 Play from here I know that you have a recent one that you really like.

43:55 Play from here I want you to talk about it.

43:56 Play from here And instead of me trying to pick my favorite baby,

43:59 Play from here I am.

43:59 Play from here I'm going to then talk about a very simple,

44:02 Play from here agentic use case that was a big unlock for me.

44:06 Play from here And I'm hoping that you're just going to look at me

44:09 Play from here and shake your head and be like, yeah, dummy,

44:10 Play from here of course you should have done that.

44:12 Play from here But we'll see if that's true.

44:13 Play from here I'm going to let you go.

44:14 Play from here Tell me about a workflow automation,

44:16 Play from here how you put it together and the value that it's given you.

44:18 Play from here Okay, I'm going to talk about two.

44:20 Play from here One, there's-

44:21 Play from here Cheating, cheating, it's taking all your time.

44:24 Play from here Too complicated to go into on like super deep,

44:27 Play from here a deep level here,

44:28 Play from here but it's sort of the art of the possible.

44:30 Play from here And again, if you're a SaaS company

44:32 Play from here and you're thinking about like the future of your industry,

44:35 Play from here these are things that you need to keep in mind.

44:38 Play from here And then one that's much more straightforward

44:40 Play from here because we have a bit of a challenge to the audience

44:42 Play from here is sign up for a low tier or a free N8N account

44:47 Play from here and do a basic automation.

44:49 Play from here Whether it's something that's solving something

44:51 Play from here in your personal life that's annoying you

44:53 Play from here or some just basic thing professionally.

44:56 Play from here Obviously, you need to have permissions

44:58 Play from here to be able to do this within your corporate account.

45:01 Play from here So perhaps start on a personal task,

45:03 Play from here but really just want to script out

45:05 Play from here how easy some of these things are to do

45:07 Play from here because they really are.

45:09 Play from here On the more complicated front, the one, okay.

45:12 Play from here And one thing I do have to say is I'm saying I, I, I a lot,

45:15 Play from here but I really have a team of people

45:18 Play from here who have become just ninjas at developing workflows.

45:23 Play from here And I really don't want to steal their thunder

45:26 Play from here because they are having so much fun doing this.

45:30 Play from here I have this great guy named Benson

45:32 Play from here and Benson got on a call with me just the other day.

45:35 Play from here He's a young guy and he's like, I have just,

45:37 Play from here I just love this.

45:38 Play from here I'm having so much fun building these things.

45:42 Play from here Like this is what I want to do.

45:44 Play from here And I was like, done.

45:45 Play from here Like you're great at this.

45:46 Play from here It's almost one of those things, Kevin,

45:47 Play from here that the more you do it,

45:48 Play from here the more potential that you see there is, right?

45:50 Play from here You start off with something simple,

45:52 Play from here you add a little bit of complexity

45:53 Play from here and then all of a sudden you have an idea

45:54 Play from here for an entirely new automation.

45:56 Play from here That's 10X more complex than the last one you did,

45:59 Play from here but you would have never gotten to that unlock

46:00 Play from here if you didn't start playing with it, right?

46:02 Play from here It does have this really fun,

46:03 Play from here addictive, iterative cycle to it.

46:06 Play from here And now we're just ideating left, right, and center.

46:08 Play from here So anyway, so when I say I, I mean we,

46:10 Play from here like this is, I just wish I could spend like all day

46:13 Play from here playing with these things when I just can't.

46:16 Play from here Okay, so the one that just blows me away

46:18 Play from here that Benson built for me is I have a little bit

46:21 Play from here of a problem in that I'm sort of our own sales funnel

46:25 Play from here that like people come in and we meet them

46:30 Play from here through LinkedIn or events or whatever it is

46:32 Play from here and they end up on my calendar.

46:34 Play from here Cool, great, I have an open calendar,

46:35 Play from here but often I have no idea who I'm meeting with.

46:39 Play from here So to solve this problem, you know,

46:41 Play from here you want to look and see if there's a buyable solution

46:43 Play from here out there and we looked at all kinds of solutions

46:45 Play from here and they were crazy expensive

46:48 Play from here and none of them would do exactly what I wanted it to do,

46:52 Play from here but it would have been probably $250 a month

46:54 Play from here to stitch together a few different solutions

46:56 Play from here to do what we did, which is look at the calendar,

47:00 Play from here figure out if this is somebody,

47:01 Play from here somebody that you meet with often.

47:03 Play from here Oh, it's Eli, we meet with Eli all the time.

47:05 Play from here You must know who Eli is.

47:06 Play from here It punts that person out because we know them.

47:09 Play from here If it's somebody who you don't meet with,

47:11 Play from here then use what's called Appify,

47:14 Play from here which is another connector layer

47:16 Play from here that we're going to get into at some point here

47:18 Play from here to look at LinkedIn and look at any direct messages

47:22 Play from here that this person and I have had.

47:25 Play from here So it'll take the last like six weeks of messages.

47:29 Play from here Then it will look at email

47:31 Play from here and it will look at the last six email threads.

47:34 Play from here It will put all of that into a rag database.

47:37 Play from here That's sort of important, not to get complicated,

47:39 Play from here but to lock it in so the information

47:41 Play from here isn't going to drift from there.

47:43 Play from here Then it bounces off into HubSpot

47:45 Play from here and determines whether or not

47:47 Play from here there's any sort of a sales relationship

47:49 Play from here or past history with this person or their organization.

47:52 Play from here And then it puts it all together

47:53 Play from here into this really tight briefing that I get on Slack

47:57 Play from here about four hours before the meeting.

47:59 Play from here So I just have this constant flow of meeting prep.

48:02 Play from here And there are people on this call

48:04 Play from here who at various points in their career

48:06 Play from here have had like high skilled SDRs

48:08 Play from here that spent hours if not weeks doing what I just described.

48:12 Play from here And this costs a fraction of a cent to run this thing.

48:16 Play from here And it's just, I would just curl up

48:19 Play from here if I didn't have it anymore.

48:20 Play from here And I like tweaking it and doing other stuff to it,

48:23 Play from here but super, super powerful.

48:24 Play from here So that's sort of the art of the possible

48:27 Play from here if you wanted to build something that's much more complex,

48:30 Play from here but it's doable.

48:31 Play from here And I don't need the $250 a month subscription.

48:35 Play from here I just have my guy Benson.

48:36 Play from here Can you rattle off?

48:38 Play from here So N8n is essentially the connective tissue

48:41 Play from here for all these different services

48:42 Play from here that are now being turned into an automation.

48:45 Play from here You mentioned Appify, right?

48:47 Play from here There are other tools that you're calling

48:49 Play from here to make this whole process happen

48:51 Play from here so that when the meat is put through the grinder,

48:53 Play from here the sausage that comes out

48:56 Play from here is just either a Slack message

48:57 Play from here or an email you get before your meeting

48:59 Play from here that acts as this dossier of like,

49:01 Play from here who is this person?

49:02 Play from here Have we ever spoken before?

49:03 Play from here If so, what have our conversations been about?

49:05 Play from here Is there any recent news about this person,

49:06 Play from here social media, whatever it might be, right?

49:08 Play from here What are the tools in that workflow

49:10 Play from here that you're talking to to make this happen?

49:12 Play from here That are outside of N8n.

49:13 Play from here N8n itself.

49:14 Play from here And then Appify has what are called actors,

49:18 Play from here which going back to the beginning of our conversation

49:20 Play from here are basically agents.

49:22 Play from here So you want to pull information off of YouTube.

49:25 Play from here You want to pull information off of LinkedIn.

49:28 Play from here Obviously you want to be really careful with access

49:30 Play from here and things like that with these tools,

49:32 Play from here but they generally have a very low per use cost,

49:35 Play from here like a few pennies to go off and do something.

49:39 Play from here But somebody's perfected that particular piece

49:41 Play from here and you're plugging that into your system.

49:44 Play from here Then, I mean, HubSpot is our CRM.

49:48 Play from here Very inexpensive, yeah.

49:49 Play from here Very inexpensive.

49:50 Play from here We are using a low cost Pinecone setup for-

49:54 Play from here Is that what you're using for the RAG?

49:56 Play from here We're using Pinecone for the RAG

49:58 Play from here and the RAG itself is running through GPT-5

50:03 Play from here to do the vector structures.

50:06 Play from here Again, I don't want to be careful with terminology

50:08 Play from here and scare people away and whatever.

50:11 Play from here Again, you need a RAG if you want to lock in a context

50:15 Play from here such that it comes back in a way

50:17 Play from here that's low or lower hallucination

50:20 Play from here such that you really want to secure it.

50:22 Play from here So if you're building a chatbot on your site

50:24 Play from here or something like that,

50:25 Play from here your vendor is going to build the RAG

50:27 Play from here such that it's pulling from that information first.

50:30 Play from here Guardrails.

50:31 Play from here Guardrails.

50:32 Play from here It's guardrails around it.

50:33 Play from here And it's just higher quality output.

50:34 Play from here We will get beyond that at some point,

50:36 Play from here but at this point it's pretty click-click easy

50:39 Play from here to do a lot of this stuff.

50:41 Play from here And can be faster too, right?

50:42 Play from here Can actually pull information a lot faster

50:44 Play from here than just your off-the-shelf LLM.

50:46 Play from here And then Slack.

50:47 Play from here Well, another connector is to, we use Google Workspace.

50:52 Play from here So it's connecting to my Google Calendar.

50:56 Play from here And then Slack is that communications hub in my world.

51:00 Play from here I know that you loathe Slack,

51:01 Play from here but there are other ways to do this downstream

51:04 Play from here if you're using ClickUp or you're using another tool

51:07 Play from here where you want it to be displayed.

51:09 Play from here But for my workflow, that makes a ton of sense.

51:12 Play from here And then the logical extension from it would be,

51:15 Play from here okay, I've had my call.

51:17 Play from here My call recorder of choice is something called Spiky.

51:21 Play from here It does a lot of other cool stuff for me,

51:23 Play from here but I get the transcript.

51:25 Play from here And then the transcript should bounce off

51:27 Play from here of all of the information we've already accumulated

51:30 Play from here about the person,

51:31 Play from here such that if I wanted to do a highly tailored

51:34 Play from here sales follow-up, this is already half built.

51:37 Play from here So that's just an agent on an agent that connects these.

51:41 Play from here Just for the record, I don't loathe Slack.

51:43 Play from here I loathe all real-time chats.

51:45 Play from here I just, I hate the distraction so much.

51:47 Play from here I was just loathing the invoices from Slack

51:49 Play from here after ClickUp offered the same service of Slack,

51:52 Play from here but I was already paying for it.

51:54 Play from here Don't get me wrong.

51:55 Play from here It was just the money.

51:56 Play from here It was just the money.

51:58 Play from here Okay, cool.

51:59 Play from here So listen, to be clear,

52:01 Play from here you don't need to know exactly how all this stuff works

52:05 Play from here under the hood in order to build these.

52:07 Play from here You don't need to get into vector databases and RAG

52:10 Play from here and have an intimate understanding

52:11 Play from here of how all of that works

52:12 Play from here in order to build some pretty kick-ass automations.

52:15 Play from here So if that is your more complex one,

52:17 Play from here what is a simple one that somebody could run out

52:20 Play from here using N8n and just build today on their lunch break

52:24 Play from here instead of with their 15 amazing AI assistants that you have?

52:28 Play from here Yeah, and this is challenging since we're doing this in audio

52:34 Play from here so you're going to have to bear with me just a little bit.

52:37 Play from here So something basic that almost every business

52:40 Play from here later deals with is invoice processing.

52:43 Play from here And I always like to think of terms of ROI

52:48 Play from here and there's a certain self-loathing

52:51 Play from here or activity loathing part of ROI.

52:54 Play from here And I really, really, really don't love the fact

52:58 Play from here that I get to be my own CFO and controller

53:00 Play from here and spend quality Saturday time

53:04 Play from here like looking over invoices

53:06 Play from here and making sure that everything is accurate and whatnot.

53:08 Play from here So I wanted to come up with a way

53:10 Play from here of just attending to the issues.

53:14 Play from here So knowing when an invoice is,

53:18 Play from here it's out of the norm in some way, shape or form.

53:20 Play from here And I also wanted to be able to track things in some place

53:23 Play from here that my bookkeeper would be able to find them,

53:26 Play from here et cetera, et cetera.

53:27 Play from here So super simple flow.

53:29 Play from here All this does is it looks for a tag in any email.

53:33 Play from here In this case, it's a transactional email

53:36 Play from here that's tagged with invoice.

53:37 Play from here So yes, I had to go into Gmail and create a tag

53:40 Play from here that would record or a filter or a label,

53:43 Play from here a label in Gmail that knew that something was an invoice.

53:47 Play from here So, okay, that's outside of this flow.

53:49 Play from here That's still pretty normal.

53:51 Play from here Every time one of those shows up,

53:53 Play from here what I want to happen is I want the PDF

53:56 Play from here that's attached to there

53:57 Play from here to be transformed into something useful.

54:00 Play from here So the trigger is a Gmail connector

54:03 Play from here that has a get message function.

54:06 Play from here And the get message function is looking for label invoice.

54:10 Play from here So every time an email comes in that has that label

54:13 Play from here and it looks once an hour, once a minute, once a day,

54:15 Play from here however you want to set it, it grabs that email.

54:18 Play from here Then the next thing that has to happen

54:20 Play from here is it has to do something with the attachment.

54:23 Play from here And the attachment is usually a PDF

54:25 Play from here and a PDF is sort of suboptimal

54:28 Play from here to deal with from a data perspective.

54:30 Play from here So you need to turn the cool visual PDF

54:33 Play from here into uncool data that you can do something with,

54:36 Play from here generally JSON.

54:37 Play from here So there's a next step in there is to extract data.

54:41 Play from here So this is a node in terms of vocabulary.

54:44 Play from here Each of these steps is called a node in N8N

54:47 Play from here and they do different things.

54:48 Play from here Trigger node is Gmail

54:50 Play from here and then extract information node is the next one.

54:53 Play from here So now all of the information from the invoice is there.

54:56 Play from here And then we need to do something with this.

54:58 Play from here And this is where the AI magic comes in.

55:01 Play from here So now you're going to have an AI node on top of it

55:06 Play from here that is going to process the information

55:09 Play from here that came out of the invoice

55:10 Play from here in some sort of a logical form.

55:12 Play from here In my case, I want things that are segregated in my invoices

55:17 Play from here in a way that I can better understand them

55:19 Play from here such that I understand what the line items are.

55:21 Play from here So the logic of the first AI node

55:24 Play from here is to separate all of the pieces.

55:27 Play from here And then the next node is a recording node.

55:30 Play from here In this case, it's just a G Suite or G Sheet node.

55:33 Play from here And it's taking all the information from that invoice

55:36 Play from here that's been extracted and it's putting it in a Google Sheet.

55:39 Play from here Up until now, you could have done,

55:41 Play from here well, except for the interpreting of the invoice,

55:44 Play from here you could have done all of this in 2002.

55:47 Play from here But then I want some logic on top of it

55:50 Play from here that identifies if there's an issue.

55:54 Play from here So I want an AI node, so AI node number two,

55:59 Play from here that's going to look at the history of data

56:02 Play from here that's in the spreadsheet and let me know

56:06 Play from here if my invoice is, say, 15% over or under

56:11 Play from here what it has been before.

56:13 Play from here It's going to flag those

56:15 Play from here and then the next node is a notification node in Slack.

56:19 Play from here So I have a accounting channel in Slack

56:23 Play from here and it's going to flag it there and say,

56:27 Play from here pay attention to this.

56:28 Play from here This is 22% more than it usually is

56:31 Play from here because those are the things that I want to pay attention to

56:33 Play from here that for some reason, something's gone haywire in billing

56:36 Play from here and I want my attention drawn to those.

56:39 Play from here And that's super simple to do those pieces.

56:43 Play from here If you want something completely like normal,

56:46 Play from here any homeowner knows that you go to Home Depot

56:48 Play from here and they send you an invoice from your receipt

56:53 Play from here and you're probably not doing a good job

56:56 Play from here of keeping those little receipts

56:58 Play from here that are accumulating over years.

56:59 Play from here And if you're not, when you sell your house,

57:02 Play from here you're going to take a hit on basis.

57:03 Play from here So super simple version, personal email, tag Home Depot,

57:08 Play from here extract the invoice, transform it into data,

57:11 Play from here and then use an AI step that categorizes that into,

57:16 Play from here is it a consumable?

57:17 Play from here Like, did you buy garden soil

57:18 Play from here or did you buy something that could be counted

57:21 Play from here against the basis of your house?

57:22 Play from here And-

57:23 Play from here I love that one.

57:24 Play from here I like that better than your accounts payable workflow.

57:25 Play from here Well, I mean, it solves the problem, right?

57:27 Play from here That, you know, over 10 years in my house-

57:30 Play from here Nobody keeps track of that shit.

57:31 Play from here Nobody keeps track of it, right?

57:32 Play from here No, it adds up, it's material.

57:34 Play from here It could be material depending on, you know,

57:37 Play from here where you are in your home value, right?

57:39 Play from here So, but now I never have to think about it again.

57:41 Play from here If I sell this house in 20 years,

57:43 Play from here then I'll have all of that data

57:46 Play from here that's just sort of quietly accumulating in the back.

57:48 Play from here Jesus, why have I not been doing this as much?

57:50 Play from here And I never have to think about it again.

57:52 Play from here That's the sort of thing that I want you guys,

57:54 Play from here the listeners, to think about.

57:56 Play from here Like, what's something just sort of annoying

57:57 Play from here in your day-to-day world that you can make more sense of?

58:00 Play from here Is it like a kid's sporting calendar

58:03 Play from here that you want to be added to your home calendar?

58:07 Play from here You know, that terrible email that parents get

58:10 Play from here where it's like, there are 22 events

58:12 Play from here that are coming up in the next four months.

58:15 Play from here And they're at these dates and these locations

58:17 Play from here and only occur for the left-handed girl athletes,

58:22 Play from here you know, here.

58:23 Play from here You can fix all of those things

58:25 Play from here by adding an AI interpretation layer on top of it.

58:30 Play from here So, a lot of this stuff is super easy.

58:33 Play from here Here we are on one side,

58:34 Play from here we're solving something

58:35 Play from here that could have been its own SaaS company

58:36 Play from here with like that SDR lead prep thing.

58:39 Play from here other, I'm saving myself, you know, I'm saving future me untold pain or untold dollars through

58:45 Play from here this silly Home Depot expense category.

58:49 Play from here I love that one.

58:50 Play from here And I get more inbound from my kids' schools, by the way, than if somebody took my email

58:54 Play from here and wrote it in the bathroom of like a funnel hacking conference or something.

58:58 Play from here It is just insane, the amount of inbound from the schools, text messages, emails, everything.

59:03 Play from here That might be my next automation is to figure that out.

59:05 Play from here I think that's a really good one too, almost as good as Home Depot.

59:07 Play from here So that's our challenge.

59:08 Play from here Our challenge is, is go forth, open up free or again, super cheap, an 8N account.

59:14 Play from here If you choose to do Zapier or Make or whatever, fine, I really, they have, they have very

59:17 Play from here similar tools.

59:18 Play from here I just personally find an 8N to be super easy to use and there's so much good information

59:23 Play from here out there.

59:25 Play from here Perhaps use the CLAWD skill that we'll provide in the show notes to set that up in CLAWD

59:31 Play from here so you can ask it questions about what you're actually trying to accomplish and just build

59:36 Play from here something.

59:37 Play from here Maybe you're building something totally simple and perhaps stupid and you'll see the potential.

59:43 Play from here And then what happens is you look at your job or you look at the way your organization

59:47 Play from here runs and you'll want to deconstruct everything.

59:51 Play from here This is a frame shift, right?

59:53 Play from here And the key is start simple because it is easy to overcomplicate this in the beginning.

59:58 Play from here Kevin and I are famous for this.

01:00:00 Play from here And then all of a sudden you're like, ah, none of this crap works, right?

01:00:03 Play from here Start simple, build confidence, and then just start to escalate into bigger problems

01:00:08 Play from here that you want to solve.

01:00:09 Play from here Yep.

01:00:10 Play from here Yep.

01:00:11 Play from here And I hope this is accessible.

01:00:12 Play from here Again, we do tend to get a little bit technical on this stuff, but it doesn't have to be.

01:00:17 Play from here And if you start with the easy, low risk stuff that's going to save you or your team a couple

01:00:22 Play from here hours here and there, just get your juices flowing a little bit.

01:00:27 Play from here Get some reps.

01:00:28 Play from here Totally.

01:00:29 Play from here I'm not saying I'm non-technical.

01:00:30 Play from here I've got like the least technical thing you can do.

01:00:34 Play from here With such a weird unlock for me, you know how we keep having those moments of we will

01:00:38 Play from here try something with a tool that we haven't tried before.

01:00:40 Play from here And then all of a sudden the light bulb goes off and you're like, oh my goodness, this

01:00:44 Play from here could change everything.

01:00:46 Play from here Why haven't I been doing this before?

01:00:48 Play from here And we're going to keep having those moments with AI as this progresses.

01:00:52 Play from here Doubtedly, it's just going to keep happening.

01:00:54 Play from here So a while back, we probably vaguely referenced agentic browsers, right?

01:01:00 Play from here First Perplexity came out with one with Comet.

01:01:03 Play from here Google added a bunch of stuff in Chrome that just didn't really, it was kind of a nothing

01:01:06 Play from here burger.

01:01:07 Play from here Nothing really happened with it.

01:01:08 Play from here Gemini is now being peppered into different parts of the Google ecosystem.

01:01:12 Play from here It's actually really helpful in things like Sheets.

01:01:14 Play from here I kind of can't believe that it was never there because now I'm just using it all the

01:01:19 Play from here time.

01:01:20 Play from here I don't know how I survived without it.

01:01:21 Play from here And then OpenAI came out with their own browser, Atlas.

01:01:25 Play from here And we talked about it a little bit and Kevin and I were like, oh my goodness, have you

01:01:28 Play from here downloaded Atlas?

01:01:29 Play from here Kevin says, oh, it's my default browser now.

01:01:31 Play from here And then I don't know, what, three days, four days later, Kevin, you're like, oh, it's no

01:01:34 Play from here longer my default browser actually.

01:01:37 Play from here Use all these Chrome extensions and it isn't doing this very well and I don't even know

01:01:40 Play from here why I need AI in my browser to be frank.

01:01:43 Play from here So you've kind of went full circle and very early, let's see what happens, right?

01:01:47 Play from here Let me describe a use case and it's easy to get in our own little bubbles and our own

01:01:53 Play from here little AI workflows and wonder, do other people have these problems?

01:01:58 Play from here Is it just me?

01:01:59 Play from here So here's a common problem.

01:02:01 Play from here Take any piece of relatively convoluted software that you're having to deal with on a day-to-day

01:02:06 Play from here basis and you're trying to figure something out about it, anything like I was having a

01:02:12 Play from here problem with my AI note taker, we were using Fireflies and for some reason it was wreaking

01:02:19 Play from here havoc in our CRM because of some settings, it was creating all these duplicates and random

01:02:24 Play from here stuff.

01:02:25 Play from here Couldn't figure it out.

01:02:26 Play from here And so I just go, hey, here's this problem I'm having, tell me how to fix this.

01:02:30 Play from here What are your recommendations?

01:02:31 Play from here And it says, oh, okay, first go to the settings, now go to this, now go to this.

01:02:36 Play from here And after step two, you realize that all the things it's describing don't exist in the

01:02:40 Play from here UI, right?

01:02:41 Play from here So you just completely made it up and you say, yeah, that's not there.

01:02:45 Play from here That doesn't exist.

01:02:46 Play from here It says, oh, you got me.

01:02:48 Play from here Yeah.

01:02:49 Play from here They, they updated the UI.

01:02:50 Play from here I'm like, no, no, they haven't.

01:02:51 Play from here I've been using this for five years.

01:02:52 Play from here They haven't touched this crusty old product forever.

01:02:55 Play from here You are making this up like these don't exist in the UI.

01:02:58 Play from here So what I invariably do is I grab a screenshot and say, this is what I'm dealing with.

01:03:03 Play from here What's going on?

01:03:04 Play from here So, so I see what the problem is now, click this, and then you'll get to that menu.

01:03:08 Play from here And then I click it and I'm like, no, you're wrong again, still nothing in the UI.

01:03:12 Play from here So I take another screenshot.

01:03:13 Play from here So I end up passing these screenshots into my chat over and over and over again.

01:03:18 Play from here Then I look at my watch and like three hours have gone by where I'm like passing screenshots.

01:03:24 Play from here My desktop is now filled with screenshots and I'm still not that much closer to solving

01:03:28 Play from here my problem.

01:03:30 Play from here Has this happened to you where you're passing these screenshots back and forth?

01:03:32 Play from here Oh, absolutely.

01:03:33 Play from here In fact, it happened to be just last night that I have, I have a DBT project set up for

01:03:39 Play from here major platforms.

01:03:41 Play from here And even so you, you can do some pretty clever things by the way, about like creating a,

01:03:47 Play from here like, oh my gosh, I'm rolling my eyes because I don't want to get technical, but creating

01:03:50 Play from here like a get repository for like a FAQs for a platform and then calling it CloudSkills,

01:03:56 Play from here which have come up several times here will do a better job than just prompting, but who's

01:04:01 Play from here got time for that?

01:04:02 Play from here Back to the midwhip meme, right?

01:04:04 Play from here Right.

01:04:05 Play from here Exactly.

01:04:06 Play from here I can't control this whole complex workflow because I can't use this, this, this, this

01:04:10 Play from here goddamn piece of SAS that just has an awful user experience.

01:04:13 Play from here Right.

01:04:14 Play from here And I can't find the settings that do the thing that I want it to do.

01:04:18 Play from here I have this moment where I'm like, wait a minute.

01:04:21 Play from here So I've, I've got Atlas, GPT is in Atlas.

01:04:24 Play from here It can see my screen.

01:04:25 Play from here Why am I doing screenshots?

01:04:27 Play from here Why am I not just having this conversation in the browser?

01:04:29 Play from here So in this case, it was our CRM, we're using pipe drive.

01:04:33 Play from here I was having a hell of a time building out some of these reports because it uses this

01:04:39 Play from here conditional logic.

01:04:40 Play from here The UI is awful.

01:04:42 Play from here And what GPT kept telling me was to, oh, use this condition.

01:04:46 Play from here I'd go to the dropdown menu.

01:04:47 Play from here That condition doesn't exist.

01:04:48 Play from here Right.

01:04:49 Play from here So I'm like, this is what's available.

01:04:50 Play from here Tell me what to do.

01:04:51 Play from here I put it in the screenshot again.

01:04:53 Play from here This isn't rocket science.

01:04:54 Play from here This isn't like a real complex thing to do.

01:04:58 Play from here It's just one of those time-consuming pain in the ass tasks to work through that just

01:05:03 Play from here takes way more time than you ever think it's going to.

01:05:06 Play from here So I decided to test my idea.

01:05:08 Play from here I opened up the CRM inside of Atlas and I started to chat and I say, hey, this isn't

01:05:13 Play from here working in this report.

01:05:14 Play from here Why?

01:05:15 Play from here Oh, I can see what's going on here.

01:05:17 Play from here What you need to do is actually build an automation that will then give you the clean data for

01:05:22 Play from here this report.

01:05:23 Play from here I'm like, oh, okay.

01:05:24 Play from here I haven't done that before.

01:05:25 Play from here I guess I got to figure that out.

01:05:28 Play from here Let's work through it.

01:05:30 Play from here And it asks, well, do you want me to do it?

01:05:32 Play from here I say, sure, yeah, go ahead and do it.

01:05:37 Play from here And one of the big promises of these agentic browsers or just agentic AI in general is

01:05:45 Play from here that like back with operator, for example, you know, it could take over your screen and

01:05:50 Play from here it could do these tasks.

01:05:52 Play from here But the complaint was, A, it didn't do it very well.

01:05:54 Play from here And B, it was like painfully slow, like painfully slow, like you're watching it do the work

01:05:59 Play from here and you're thinking to yourself, I could just be doing this 10 times faster.

01:06:02 Play from here Like this is a complete waste.

01:06:04 Play from here So what I did is I let it go through a process that I didn't know how to do.

01:06:08 Play from here I would have to learn.

01:06:09 Play from here I'd have to figure out.

01:06:10 Play from here I'd have to have a conversation and it just did it.

01:06:13 Play from here And yeah, it was kind of slow, but it did it and it saved it and it worked.

01:06:17 Play from here And I had a list of these things that I had to do.

01:06:20 Play from here So then I did the next one.

01:06:21 Play from here I was at my daughter's gymnastics class.

01:06:23 Play from here I brought my laptop and I sent Kevin a message and I was, I pointed my computer.

01:06:28 Play from here I'm like, yeah, just me taking my daughter to gymnastics while my AI agent is just building

01:06:33 Play from here these workflows for me in the browser.

01:06:35 Play from here It was one of those moments, Kevin, where I'm like, yeah, it's a little slow right now,

01:06:39 Play from here but it's actually working.

01:06:41 Play from here And what an unlock.

01:06:42 Play from here Who cares how long it takes if I'm just off doing something else and I have five of these

01:06:46 Play from here browser tabs open doing these tasks concurrently.

01:06:50 Play from here Like it was pretty, it was one of those moments for me where I'm like, ah, now I get it now.

01:06:55 Play from here Now I get the promise of this browser.

01:06:57 Play from here Now I see where this is going.

01:06:58 Play from here This is huge.

01:06:59 Play from here So a big topic for us at various times going forward is going to be just this impact of

01:07:04 Play from here the agent layer on the internet at large.

01:07:07 Play from here So if I'm working with a direct to consumer company or a company that does a transaction

01:07:11 Play from here online, be it for a property or an appointment or whatever, there's all of this conversation

01:07:16 Play from here going on about the metadata that sits behind sites and making it easier for the LLMs to

01:07:22 Play from here essentially get their talents into things.

01:07:24 Play from here And what you're describing there to me indicates that we're going to need a better layer of

01:07:30 Play from here how like these skills are placed.

01:07:36 Play from here So if you're a pipe drive, you want to have a skills package.

01:07:40 Play from here This doesn't really exist in this format right now.

01:07:42 Play from here So if pipe drives listening, you know, whatever, but you need something there that's like an

01:07:48 Play from here information packet that's fully up to date such that the model isn't just making assumptions

01:07:52 Play from here because you still have the same problem you did before or when you were doing the screenshots

01:07:56 Play from here that if it doesn't have that, that refined skill there is basically just going off of

01:08:01 Play from here its training and it's likely to get confused.

01:08:04 Play from here But if it instead has an embedded skill that it just basically loads when it goes to the

01:08:08 Play from here page, it's like, oh, okay, well this is how I do a pipe drive automation.

01:08:12 Play from here So if you're a brand, you need to be watching this space because that sort of information

01:08:18 Play from here exchange is going to be part of the future of how humans and agents interact with your

01:08:25 Play from here environment.

01:08:26 Play from here And, you know, if you're a consumer and you're using these sites, you should probably be

01:08:32 Play from here playing with these things as well.

01:08:33 Play from here So that's, you know, there's a lot there.

01:08:36 Play from here There's a lot there as far as as far as sort of the future of how we do these things for

01:08:40 Play from here sure.

01:08:42 Play from here But pretty cool.

01:08:43 Play from here For me, that tipping point is, oh, it worked, right?

01:08:46 Play from here It did the thing I wanted to do and it actually saved me a bunch of time and trouble.

01:08:51 Play from here That was that was pretty telling.

01:08:53 Play from here That was pretty incredible.

01:08:54 Play from here And this might be a little nerdy, but there is this system window that appears when it's

01:08:59 Play from here taking over your browser and acting on your behalf.

01:09:01 Play from here So you can watch it think, right?

01:09:03 Play from here You can you can you can watch what it's going through.

01:09:06 Play from here You know, as Kevin said, it's not so sophisticated that it has the instructions.

01:09:10 Play from here It's just doing what I would have to do without the screenshots and the back and forth with

01:09:14 Play from here the model.

01:09:15 Play from here It's just doing that for me, which is pretty incredible.

01:09:18 Play from here And watching it think it will say things like, well, the user gave us the log cred.

01:09:24 Play from here So we want to be careful about maybe we should ask if we want to save this.

01:09:27 Play from here And it's going through those processes.

01:09:29 Play from here Really, really fascinating to just get a behind the scenes look as to how the model's operating,

01:09:34 Play from here how it's working when it does take over the browser and act on your behalf.

01:09:38 Play from here Really, really cool.

01:09:40 Play from here We're already at time and there was something else that I really wanted to get to and maybe

01:09:45 Play from here I'll just have to stop recording.

01:09:48 Play from here It's such pressing news, but that's OK.

01:09:53 Play from here It is a big conversation, though, so maybe a bit of a cliffhanger there, but super fun

01:09:57 Play from here as usual, buddy.

01:09:59 Play from here And I really hope everybody takes your challenge and just gets a free or low cost account and

01:10:06 Play from here tries something that's been bugging them.

01:10:10 Play from here Try to make it easier.

01:10:11 Play from here Try to make it just happen in the background so you don't have to think about it anymore.

01:10:15 Play from here Get some of that psychic baggage off of your head.

01:10:17 Play from here Yep.

01:10:18 Play from here Give it a go.

01:10:19 Play from here Break things.

01:10:20 Play from here Get a little frustrated.

01:10:21 Play from here Solve a little problem.

01:10:23 Play from here Trust me, it's fun.

01:10:24 Play from here It is fun.

01:10:25 Play from here It is fun.

01:10:26 Play from here Maybe not the first time, but it gets fun.

01:10:27 Play from here All right, Kevin.

01:10:28 Play from here Talk soon, bud.

01:10:29 Play from here Bye.

01:10:30 Play from here Thanks for listening.

01:10:37 Play from here If you'd like to send us a question to answer on the show, email me at Elijah at Spark six

01:10:42 Play from here dot com.

01:10:43 Play from here That is S.P.A.R.K. the number six dot com or Kevin at Kevin at Ascend Labs dot A.I.

01:10:50 Play from here to learn more about A.I. strategy and training for your business.

01:10:54 Play from here Ascend A.I. labs at Ascend Labs dot A.I. and to learn more about design and development

01:11:01 Play from here of custom A.I. solutions, visit my agency at Spark six dot com.

01:11:06 Play from here You can also sign up for my free newsletter A.I. Framework Friday by visiting Spark six

01:11:11 Play from here dot com forward slash newsletter.

01:11:23 Play from here Thank you.

01:11:24 Play from here Bye.

How to cite this episode

Credit the named speaker and AI at Work by AscendAI. Link to the episode or the passage you quoted.

“Ep #4 AI Agents, Chaos Workflows & Why Sora Is Ruining Our Lives (In the Best Way).” AI at Work by AscendAI, 2025-11-18. https://ascendlabs.ai/podcast/ai-agents-chaos-workflows-why-sora-is-ruining-our-lives-in-the-best-way/

Recent episodes

View all episodes
×
AI Solutions Graphic
Free Resource
Get Started Now