You Already Have the People. You Just Haven’t Given Them the Structure.
The most common thing I hear from mid-market leaders about AI goes something like this: “We want to move on this, but we don’t have the technical staff to pull it off.” It sounds reasonable. It’s also, in most cases, the wrong diagnosis.
Wyatt Barnett is Head of AI at the NCTA the Internet and Television Association in Washington, D.C. Eighty people. No engineering staff. No CTO. And he’s built more functional, working AI capacity than most organizations twice his size. When I talked with him recently on AI at Work, the thing that kept standing out wasn’t the tools he’s using or the projects he’s shipped. It was the organizational model underneath all of it.
He calls it the Tiger Team. And if you run a company between 40 and 200 people, it’s worth understanding.
Listen to the full episode:
The Tiger Team Model: AI Capacity Without Engineering Titles
The Tiger Team isn’t a department. It’s not a committee. It’s ten people drawn from across the organization not from IT, not from the executive floor who serve as what Wyatt calls forward-deployed problem-catchers.
The profile matters: typically senior coordinators, one from each major department, who understand the work deeply enough to recognize when something could be done differently. Not the department head, and not the most junior person on the team. The person who knows where the friction actually lives.
Their job is to be in the room when a problem surfaces before it travels up to a department head, becomes a formal proposal, and lands on the desk of someone who wasn’t part of the original conversation. That journey, Wyatt noted, is where most good ideas go to die in organizations this size.
The Tiger Team short-circuits that process. Someone on the team notices a problem, brings it to the group informally, and they start working. No requisition. No weeks-long approval cycle. Just: here’s what we’re seeing, here’s what we tried, here’s what happened.
To make this work, Wyatt gave the team two things that most organizations withhold: dedicated time and a small discretionary budget. The budget piece is underrated. Most tools worth testing have a one-seat option under $50 a month. The Tiger Team can buy a seat, run it for a month, see if it does what it claims, and kill it if it doesn’t without a procurement process, without a vendor demo, and without anyone writing a business case for a tool that might not even work. As Wyatt put it: buying cheap seats beats going through enterprise sales every time.
What “Forward-Deployed” Actually Means for a Non-Technical Org
The phrase “forward-deployed engineer” comes from the world of high-growth software companies it usually describes a technical person embedded with a client to solve problems in real time. Wyatt borrowed the concept and stripped out the engineering requirement.
In his model, forward-deployed means proximity. It means being in the department, in the meetings, in the daily workflow where the real friction lives. The insight is straightforward: the people best positioned to identify AI opportunities aren’t the ones in the technology function. They’re the ones doing the work.
This is where most AI initiatives miss. Organizations build centralized AI teams or, more commonly, assign AI responsibility to IT and then wonder why adoption is slow and solutions don’t stick. The problem isn’t the technology. It’s that the people building the solutions aren’t close enough to the problems to build the right ones.
Wyatt’s Tiger Team members aren’t responsible for coding everything themselves. When someone hits an architecture question where does the data come from, what happens to it, how does it get from A to B, Wyatt steps in. But the goal is explicit: training wheels, not a tandem bike. Build the team up until they don’t need the scaffold. That’s an AI strategy. The other thing where the technical team does everything and everyone else waits is a dependency.
For context on what that looks like in practice: one Tiger Team member spent weekends building and vibe-coding on his own time. He’s since moved on to a more technical role elsewhere. Another member uses Notion AI to analyze the last four Tiger Team meeting transcripts before each session and surface agenda items Wyatt might have missed. These aren’t engineers. They’re people who were given time, tools, and permission.
The Confidence Problem Nobody Talks About
When I asked Wyatt how he builds up the team’s capabilities, he said something I’ve heard versions of from every organization that’s moved well on AI: most of this is a confidence exercise.
People who’ve spent years doing their jobs well are being asked to be beginners again. That’s not a training problem. It’s an identity problem. And no amount of AI tooling solves it.
Wyatt’s response to this is one of the more transferable things in the conversation. He runs what he calls AI lab days: find a small problem, write a paragraph about what you’re trying to solve, work on it using any tool available, then spend ten minutes demoing it at a group lunch. No high stakes. No formal presentation. Just: here’s a thing I built, here’s what I learned.
The effect isn’t primarily skill development. It’s proof of concept proof to the person building it that they’re capable of building something useful. Once someone has that reference point, their relationship with the technology shifts. They stop waiting to be told what to do with AI and start noticing opportunities on their own.
This is the organizational dynamic that most AI strategies skip. They invest in tool access and training hours and then measure adoption rates and wonder why the numbers are low. The gap isn’t knowledge. It’s confidence. And confidence doesn’t come from a training module — it comes from building something real, even if it’s small.
As Wyatt described it, a lot of these people now have things on their resume they wouldn’t have had otherwise. At an organization where there’s limited upward movement, that matters. The Tiger Team became a retention tool almost by accident. People who might have left for bigger organizations stayed because the work got more interesting.
What This Looks Like in Practice: Tools, Costs, and the Real Risks
For organizations considering something like this, a few practical observations from Wyatt’s setup are worth noting.
On tools: the winners at an 80-person org aren’t the enterprise platforms. They’re the ones that work at human scale. Notion for meeting notes and connected databases. Airtable as a data orchestrator that doesn’t require a data engineer to maintain. Claude and ChatGPT connected to existing workflows in Word, Outlook, and wherever the organization’s actual work lives. The tool question matters less than most people think. The organizational structure to use the tools matters more.
On token costs: Wyatt is one of the more honest voices I’ve spoken with on this. The real risk isn’t the current cost it’s the price of success. They built a tool that reads FCC filings, large PDFs, to make sure nothing gets missed. When they turned it on at scale, the cost was several hundred dollars a day. The value was real. But nobody had run that math before flipping the switch. The lesson isn’t to avoid building it’s to build the business case before you deploy, not after. What is this worth? What does it replace? Is there a cheaper way to handle the parts that don’t require the LLM?
On org structure: Wyatt’s view is that whoever owns the AI function needs to be at the leadership table not reporting through finance and risk management, where the instinct is to slow things down rather than move them forward. That doesn’t mean ignoring risk. It means making sure the people thinking about what’s possible have as much organizational weight as the people thinking about what could go wrong.
The Thing Your AI Strategy Is Actually Missing
The reason Wyatt’s model works isn’t the Tiger Team itself it’s what the Tiger Team represents. It’s an organization that decided to develop the people already closest to the work rather than waiting for the right hire.
Most mid-market companies treat AI readiness as a staffing problem. When you don’t have technical staff, you wait until you do. But the domain knowledge the understanding of the business, the customers, the processes is expensive to replicate and already sitting inside the organization. The technical capability is learnable. The business knowledge isn’t.
Wyatt’s framing: you can tech people up pretty easily. But you can’t quickly create someone who understands how the organization works. So start with the people who understand the organization and build the technical capability around them.
If you run a company between 40 and 200 people and you’re waiting for the budget to hire an AI team, the Tiger Team model is worth a hard look. The people you need are probably already there. The missing ingredient usually isn’t talent it’s structure, time, and permission to try.
If you want to know where your organization actually stands before you start building, the assessment at https://assessment.ascendlabs.ai/ will give you a clear picture of what’s ready and what isn’t.
And if you want to think through what this structure could look like in your specific organization, I set aside time every week for exactly these conversations: tidycal.com/kevinwilliams
