Many founders I work with have already settled on an answer before they’ve finished thinking through the question. The head of digital, maybe. The COO who has always been good with systems. Or, for the more technically confident founder, whoever has the most credibility with AI tools. The instinct makes sense on the surface. AI feels like a technology problem, and technology problems have always gone to technical people. The trouble is the instinct misreads the actual job, and that misread tends to surface at the worst possible time, typically around six months in, when adoption has stalled and the work finds its way back to you.
What does “owning AI” in your business actually mean?
Being the AI lead means being accountable for turning the technology from something you’ve licensed into something your team actually uses. That covers choosing where to start, piloting the right tools, running the change work when adoption stalls, and reporting to you on what works and what does not. The role sits much closer to programme management than software engineering. Technical fluency helps. Operational standing is the thing you cannot do without.
The distinction matters because the two qualities attract different people and sit in different parts of the organisation. A technically strong hire brings tool knowledge, model intuition, and the credibility to push back on a vendor’s claims. An operationally experienced operator brings team authority, change management experience, and the willingness to surface problems before they compound. Both are valuable. But in a business of 30 to 150 people, where the team needs to trust the person asking them to change how they work, only one of those qualities is load-bearing from day one.
Why does who holds the mandate shape whether AI sticks?
MIT’s 2025 research on generative AI found that roughly 95% of pilots stall before delivering measurable results. The root cause was a gap in how AI was integrated into the way people actually work, not a problem with the model or the data. That is fundamentally a people-and-operations failure. The person running your programme sets the terms for how seriously the team takes it.
BCG’s analysis of adoption patterns found roughly half of organisations stuck in stagnating or emerging stages, unable to convert a promising pilot into a consistent working habit. The companies that moved through that ceiling generally had one thing others did not, genuine change authority behind the programme, with someone who kept driving adoption after the early enthusiasm had faded.
Sponsorship matters alongside delegation. Spencer Stuart’s research on executive AI engagement found that progress came from leaders who stayed close enough to understand what was working and why, while giving their operational lead clear authority over day-to-day decisions. Delegating and disappearing produced weaker results, consistently. The founder’s visibility is not the same as the founder’s interference, and the line between them is worth knowing.
Where does the technical-person instinct go wrong?
Research on AI readiness has identified what some executive search firms have called the AI readiness paradox. Firms commonly hand the AI mandate to strong operators who have the authority to drive change but lack AI-specific preparation. Alternatively, they appoint a technical specialist with real tool knowledge but no operational standing with the team that needs to adopt it. Both choices carry a real failure risk, and the question is which gap closes faster.
The operator’s failure mode shows up in knowledge gaps, choosing the wrong tools, accepting vendor claims without being able to evaluate them, or underestimating what the technology can genuinely do for the business. These are real limitations. They are also addressable. An operator can be supported, upskilled deliberately, or given access to technical advisers who fill the gaps as the programme progresses.
The technical specialist’s failure mode runs in a different direction. The programme stalls because the team does not take the person seriously when they ask for change. Technical credibility does not transfer into operational authority, and without authority, the best implementation plan in the world will not shift behaviour. Building that kind of standing from scratch in a new organisation takes longer than many AI programmes can afford before the founder starts asking why adoption numbers are flat.
When do you genuinely need technical depth?
Technical expertise matters in two specific situations. The first is when you are building something bespoke, such as a custom AI system or a product feature requiring model training. The second is when a vendor is pitching a complex solution and you lack the means to evaluate the claims. Both are genuine needs, and neither requires that expertise to sit inside the AI lead role.
For the typical owner-managed business in the 30 to 150 people range, the AI programme in the first two years does not require bespoke development. The work is getting the right tools into consistent use across the team, measuring what changes, and building the habit across different parts of the business. An experienced operator can cover that ground with a trusted technical adviser on retainer, or through a structured engagement that brings specialist knowledge in without needing it to own the programme. The technical layer belongs in the support structure, not the mandate.
Russell Reynolds’ research on AI and executive leadership reinforces this. The mandate belongs with someone who can drive change across the business. Technical depth comes in where the work calls for it, at the tool selection stage, during vendor evaluation, or when a bespoke build is genuinely on the table. Buying that depth in as needed keeps the hiring requirement realistic and keeps the AI lead focused on the thing that determines outcomes.
What should you look for in the person you pick?
The person who owns your AI programme needs three things in reasonable combination. Enough AI fluency to know what good adoption looks like and to recognise an oversell when a vendor is pitching. Operational standing to be taken seriously when they ask the team to change how they work. And a disposition toward honest reporting, so you hear about problems before they become yours to fix.
That combination sits with experienced operators more often than with technical hires. Technical people can develop the operational dimension over time, but in a business that needs results within a year, the safer starting position is the person who already holds the harder quality.
If your instinct is to hand this to your most technically capable person, test it against the real job description. Running an AI programme in a founder-led business is change management with a technology wrapper. The change work is where programmes fail, often gradually and without anyone drawing attention to it, long before you would formally describe the programme as struggling. Technical ability does not fix a team that does not trust the person asking them to work differently.
The gap between a strong operator and a capable AI programme lead is closable with the right support. The gap between a technical specialist with no operational standing and the lead this role needs is not, and it tends to widen rather than narrow as time passes. Start with the person best suited to the harder part of the job. The technical knowledge follows.
If you’re weighing this decision and want a second read on who you’re considering, book a conversation and we can talk it through.
