Being the front runner for AI at your company is a terrible job. Do it anyway.

You volunteered. Or maybe you just knew more about AI than anyone else in the building, and that was enough. Someone said “you should lead this” and suddenly the job was yours. No title change. No budget. No team. Just you.

At a 200-person company, there is no AI team. There is you.

Research published this month in HBR found that people in high-AI-oversight roles experience 12% greater mental fatigue, 33% more decision fatigue, and 34% higher intent to quit. That last number hit me hard. The people who care most about AI at your company are the ones most likely to walk out the door.

Tomas Kazragis, VP of Engineering at Omnisend (about 200 employees), described the problem clearly: “We asked people to move, and they did, without a clearly defined objective or measurable result.” That pattern keeps showing up. Someone gets excited, starts pushing, and realizes the organization hasn’t decided what it’s trying to accomplish.

Your reward for caring more is burning out first.

This isn’t a how-to guide for building a champion program. I wrote one of those already on structuring champion networks. This is about what it feels like to be the person inside a mid-size company who carries AI forward, mostly alone, while the rest of the organization alternates between ignoring you and resisting you. And why, despite all of that, the role matters more than almost anything else happening at your company right now.

The early adopter tax nobody told you about

The promise was that AI would save you time. A TechCrunch investigation found the opposite: early AI adopters are working longer hours, not shorter ones. The tools open up new possibilities, and each possibility becomes another task on a list that was already too long.

One in seven workers now report fatigue from juggling AI tools. That number will feel low if you’re the person everyone comes to when their prompt doesn’t work, when the AI gives a weird answer, or when they want permission to use something they found online. You didn’t sign up to be a help desk. You became one anyway.

Then there’s the failure rate. 95% of generative AI pilots don’t make it to production. Each failed pilot costs you political capital you can’t earn back. The third time you pitch something that doesn’t deliver, people don’t argue with you. They just stop showing up.

Eoin Hinchy, CEO of Tines (about 300 employees, workflow automation), described this grind in practice: “There had to be a lot of pep talks, dialogue, and reassurance with the engineers, product team, and our sales folks saying all this blood, sweat, and tears up front in this unglamorous work will be worth it in the end.” He also admitted there were moments they thought they’d cracked it, “only for us to realize, actually, no, we need to go back to the drawing board.”

42% of companies abandoned most of their AI initiatives in 2025. Up from 17% the year before. That’s not a blip. That’s organizations collectively deciding the whole thing isn’t worth it.

I call this the “frozen champion” pattern. It happens the same way nearly every time. Enthusiasm. A few small wins that feel like momentum. Then a wall of organizational inertia that the wins can’t punch through. Management asks for ROI numbers you don’t have yet. A pilot gets shelved because another priority ate the budget. The colleague who was excited last month stops showing up to your meetings. Slowly, the champion goes quiet. Not because they stopped caring, but because caring started costing too much. The emotional toll of this role at your company is something most organizations never account for. In conversations I’ve had with people in exactly this role, the pattern repeats with almost depressing reliability.

Maya Mikhailov, CEO of fintech AI startup SAVVI AI, nailed the root cause: “We need to buy ChatGPT and figure out what to do with it later is not a business strategy.” But that’s exactly the strategy most mid-size companies hand to their champion. Figure it out. Report back. Good luck.

Everyone is working against you

You are not paranoid. The data says you are correct.

39% of managers either prohibit or don’t encourage AI use on their teams. Not 39% of companies. Individual managers. So even if the CEO gave an inspiring speech about AI being the future, nearly two out of five managers are either blocking it or passively letting it die in their departments.

Then there’s outright sabotage. A Fast Company investigation found 31% of employees admit to actively sabotaging their company’s AI strategy. Some ignore mandates. Others withhold feedback. Most just revert to old tools the moment nobody’s watching. Eric Vaughan, CEO of IgniteTech, talked about getting “flat-out, ‘Yeah, I’m not going to do this’ resistance.” And he’s the CEO. Imagine being the champion with no authority at all.

Meanwhile, over 80% of workers are using AI tools the company never approved. The tension between shadow AI and sanctioned champion-led adoption creates an absurd situation. You’re trying to get people to use approved tools while they’re already using ChatGPT on their phones under the desk. You’re fighting for something people already have in unauthorized form. How do you even sell that? I probably shouldn’t admit it, but some days the whole thing feels ridiculous.

Anjali Arora, CTO of Perforce, said it directly: “Organizations have to ask a very basic question: Do we even have these people today?” Most mid-size companies don’t. They have one person.

The contrast with enterprise is painful.

Citi has over 4,000 internal AI champions across 230,000 employees. DBS Bank in Singapore built a 700-person Data Chapter embedded throughout the organization. JPMorgan Chase has entire teams dedicated to AI adoption. You have yourself, maybe one curious colleague, and whatever time you can steal from your actual job.

What I keep seeing is that the champion often needs someone who can walk into a room with no history, no political debts, and both the technical ability and the business language to translate between the two sides. That combination is rare inside most companies. It’s worth finding externally.

What separates the 8% who get there

Only 8% of companies qualify as AI front-runners in a survey of 2,000 companies. Other research puts it at roughly 6%. Either way, the vast majority are stuck in permanent pilot mode.

The common thread among the 8%? They treat the champion role as a real job. Not volunteer work.

Front-runner companies show 4x higher talent maturity than companies still experimenting. They invest in the people, not just the technology. And C-suite sponsorship creates a 2.4x multiplier on AI success when it’s aligned with business impact. Those aren’t soft findings. That’s the difference between programs that scale and ones that evaporate.

Look at what the companies getting this right actually do. JPMorgan Chase took the approach of democratizing access while mandating nothing; 200,000 employees onboarded to their internal AI platform within eight months. Schneider Electric built mandatory four-tier training from the production floor to the C-suite. Novartis created an AI Fellows program that trained high-potential people intensively and sent them back to business units as resident experts. These companies invested in the person carrying the flag. Not just the technology the flag represents.

Bill McLaughlin, CEO of managed IT firm Thrive, acknowledged the gap directly: “Many businesses recognize the necessity of adopting AI to remain competitive, yet numerous mid-sized companies struggle with where to begin or lack the resources to implement this technology strategically and securely.” Josh Withers of True Platform went further: “The real shortage is for adaptable leaders who can guide an organization through significant technological and cultural change.”

The pattern that keeps showing up when I work alongside internal champions is this: the technical knowledge isn’t the bottleneck. It’s the ability to connect that knowledge to business outcomes in language the CEO understands. Building Tallyfy taught me that technology is 20% of the problem. Convincing people is the other 80%. Someone who’s built products, run a company, and understands the technical stack can compress that translation work dramatically.

Even the best companies have only scaled 34% of their strategic AI bets. That’s the front-runners. I don’t see the gap closing on its own.

Start here Monday morning

Get executive air cover in writing. Not a verbal nod in a hallway. An email, a Slack message, something you can point to when a department head questions why you’re spending time on “that AI stuff” instead of your regular work. HBR’s research on organizational barriers is unambiguous: AI programs without explicit executive sponsorship dissolve. Every time.

Set a kill threshold before starting any pilot. Before you build anything, agree with your sponsor on what failure looks like. “If we don’t see X result in Y weeks, we stop and redirect.” This protects your reputation. Each pilot that fades without a clear outcome costs credibility whether it was a genuine failure or not.

Build a small network. Two or three curious people in different departments. They don’t need to be technical. They need to be trusted by their teams. I wrote a full guide on how to structure champion networks if you want the mechanics. Even a tiny network reduces the isolation that kills most champions.

Find someone who bridges both worlds. The biggest unlock I’ve seen for internal champions is pairing with someone external who speaks both languages fluently. Not a consultant who brings frameworks and leaves. Not a technologist who can’t explain ROI to the CFO. Someone who’s built things, shipped things, and can sit in a technical architecture meeting at 10am and a board strategy session at 2pm. That person compresses your learning curve from years to months. If the resistance feels overwhelming, that outside perspective becomes even more valuable.

Nobody promoted you into this role. Nobody will promote you out of it. The work is mostly thankless, frequently frustrating, and occasionally feels pointless. But the companies that figure out AI in the next two years will look back and point to one person who refused to let it die.

That person is probably you.

Worth discussing for your situation? Reach out.