How do I know an AI adviser will not carry what they learn to a competitor?

You cannot know it fully, because an NDA is a reactive instrument that gives you a cause of action after the damage is done rather than preventing it. What reduces the risk is structure: industry-vertical exclusivity against direct competitors written into the contract, isolated working environments per client, and a referral pipeline worth more than any single engagement. The residue is trust, and you are right to make an adviser prove it before you sign.

Why do most AI projects fail, and what changes the odds?

Roughly 80% of AI projects fail against intended business value, about twice the rate of ordinary IT projects, and most of that failure traces to nobody owning the outcome. Engagements built around 90-day milestones with kill thresholds defined upfront, where failing projects get cut early rather than invoiced through, improve the odds without guaranteeing them.

Hard questions we hear

Q: Is two days a month enough for real AI leadership?

It is enough when what your company needs is strategic judgment applied at the right moments rather than daily presence, because most of what a full-time Chief AI Officer does all week does not require a Chief AI Officer. If your company genuinely needs daily AI leadership, fractional is the wrong model and an honest adviser will tell you so.

Prospective clients ask us difficult questions. The best ones ask brutal questions, because they have been sold transformation before and are still paying for it. We keep a running record of every objection we hear, and these five are the hardest. Not the most aggressive. The hardest, meaning the ones where our answer has to concede something real. Here they are, with the answers we give in the room.

“An NDA doesn’t un-know things. How do I know you won’t carry what you learn here to your next client?”

You don’t, not fully. NDAs are reactive instruments: they give you a cause of action after the damage is done, and enforcing one means spending two years in court trying to prove a breach. The real risk is that an adviser carries your strategic insight in their head and applies it, even unconsciously, somewhere else. No contract eliminates that.

What reduces it is structure. We maintain industry-vertical exclusivity for direct competitors, written into page one of the contract. We work in isolated environments per client. And one breach would end our referral pipeline, which is worth more to us than any single engagement. But the residue is trust, and trust has to be earned rather than contracted. You are right to make us prove it before you sign.

“Two days a month is not leadership. My receptionist is here more than that.”

Fair. A CEO who is used to full-time executives equates presence with commitment, for good reason. Leadership means being there when things get hard, not when the calendar says so.

Here is what we have found: most of what a full-time Chief AI Officer does all week does not require a Chief AI Officer. What moves AI forward in a mid-market company is strategic judgment applied at the right moments, not seat time. That said, if your company needs daily AI leadership, fractional is the wrong model and we will tell you so. Not every company is a fit for this, and pretending otherwise is how engagements fail.

“Show me one company where this moved revenue. Not a pilot. Not a dashboard. Revenue.”

You should demand this. The AI consulting space is full of impressive demos that never earned a pound, and “successful pilot” is where initiatives go to die.

We can arrange reference calls with clients who will give you the unfiltered version. But we will also tell you something most firms won’t: not every engagement produces a direct revenue line in the first quarter. Some produce cost savings. Some produce speed. If revenue impact is your bar, the right conversation is about what is realistic for your business specifically, before any money changes hands. An adviser who promises revenue in ninety days without knowing your operation is telling you what you want to hear.

“80% of AI projects fail. Why would yours be different?”

That figure is real. Pertama Partners’ 2026 analysis puts the failure rate of AI projects at roughly 80% against intended business value, about twice the rate of ordinary IT projects, and enterprise pilot data shows that 95% of generative AI pilots never scale to production. A CEO who cites this is not being negative. They are being statistically literate.

Most of that 80% fails because nobody owns the outcome. Our engagements are built around 90-day milestones with kill thresholds defined upfront, and we cut failing projects early rather than invoicing through them. That improves the odds. It does not guarantee them. Some projects will still fail, and anyone who tells you otherwise is selling.

The one nobody says out loud

Most CEOs we meet are in the same position: under pressure to appear fluent in AI while privately overwhelmed by the pace of it, and now expected to evaluate an expert in a field they cannot fully judge. There is no certification body for this role. Anyone can put “AI strategist” on LinkedIn and start charging.

So we don’t ask you to evaluate our AI expertise. We translate every recommendation into a business decision you can already judge: cost, return, risk, timeline. If you can evaluate a P&L, you can evaluate our recommendations. And for the rest, three verification paths exist. Talk to our clients directly, not through testimonials. Review our public work. Or run a two-week paid diagnostic and judge the output yourself. The burden of proof sits with us, where it belongs.

If you are weighing this decision, bring us the hardest version of your question. We would rather lose an engagement by answering it honestly than win one by talking around it.