Advisor or operator? Know what you pay for before the second strategy deck.

An AI strategy advisor tells you what to do. An AI operator does it and owns the result. Both can be useful. But many companies pay for an advisor when they need an operator. They get a strategy deck, then another one, and no system in production.

One seller described this in a short video in 2026: “Strategy without decision authority is just a billing cycle.” The seller said that the fractional Chief AI Officer market is splitting. Operators who ship outcomes are moving away from advisors who only report. I agree with the description. This page explains how to tell the two apart and how to buy the one you need.

I work as an operator. I also do advisory work when a client already has a strong team. Read this page with that fact in mind.

What is the difference?

The difference is not the skill level. Some advisors are very skilled. The difference is the authority and the accountability. An advisor recommends and leaves the decision to you. An operator gets the decision and answers for the result.

What does an advisor deliver?

A typical advisory engagement delivers these items:

• An assessment of the current AI use and maturity.
• A strategy with goals and principles.
• A roadmap of use cases, in order of value.
• A vendor short list.
• A governance framework.
• Presentations for the board or the management team.

These items are useful when your team can act on them. They are not useful when nobody has the time, the skills or the authority to act.

What does an operator deliver?

A typical operator engagement delivers these items:

• A selected business problem, measured before the work.
• A vendor decision, made and signed.
• A system in production, with evals, monitoring and cost limits.
• A number after the launch, compared with the number before.
• A team member who owns the system.
• Rules written from the real system.

An operator usually delivers fewer documents. An operator delivers more things that run.

Why do advisory engagements stall?

Advisory engagements stall for predictable reasons. Look for these signs in your company.

Nobody owns the next step. The roadmap says “deploy an AI assistant for support.” Nobody in the company has the time or the skills to do it. The roadmap waits.

The vendor decision does not happen. The advisor gives a short list of three vendors. The company asks each vendor for a demo. The demos look the same. Nobody tests the vendors with company data. After three months, there is still no decision.

The plan does not meet the data. The strategy assumes clean data. The real data is in spreadsheets, old systems and email. The first build meets this problem and stops.

The second deck. After the first strategy, the advisor proposes a deeper assessment or a second phase. The company pays again for more documents.

None of these problems is the advisor’s fault alone. The company bought recommendations. It did not buy decisions or execution.

When is an advisor the correct choice?

An advisor is the correct choice in these cases:

• You have a strong team that can execute. Your engineers and managers have time and AI experience. They need an outside view, not more hands.
• You need a board-level view. The board asks for an independent opinion on AI risk or opportunity.
• You make a large, rare decision. For example, the selection of a platform for the whole company, or a merger where AI capability matters.
• You want a second opinion. Your team has a plan. You want an expert to check it.

In these cases, the advisor’s recommendation goes to people who can act on it.

When is an operator the correct choice?

An operator is the correct choice in these cases:

• You have pilots that never reached production.
• You have no internal person with AI production experience.
• Your team is busy with other work.
• You need a result in this quarter, not a plan for next year.
• You already have a strategy deck and nothing happened.

Most small and mid-size companies that I talk to are in this group.

How do I tell which one a seller is?

Ask these four questions. The answers show the type.

1. “Who will make the vendor decision?” An operator says: “I will, with your approval of the budget.” An advisor says: “I will give you a recommendation.”
2. “What will exist after 60 days?” An operator names a system. An advisor names documents.
3. “Who will build the first system?” An operator says: “I will build it or review it, with your owner.” An advisor says: “Your team or a partner.”
4. “What number will you be accountable for?” An operator names a number and a date. An advisor talks about maturity or readiness.

Neither answer is wrong. But it must match what you need.

What are the four moves of an operator?

Real operators do four things that advisors usually do not do.

What are the risks of an operator?

Operators have their own risks. Know them before you hire.

• The operator becomes a bottleneck. All decisions go through one person. Fix this with an internal owner who learns the work.
• The operator builds what they know. An operator with one favorite framework uses it for every problem. Ask why each choice was made.
• The operator moves too fast. Speed without governance can create risk, for example data in the wrong tool. Ask for rules as part of the work.
• The operator does not leave. The company depends on the operator. Plan the exit from day one. The exit plan guide lists the handover items.

How do I write a contract for an operator?

Write the contract around results, not hours.

Can one person be both?

Yes. Many good fractional leaders do both, in different proportions over time.

At the start, the work is mostly operator work: select one problem, ship one system, measure it. Later, when the internal team runs the systems, the work becomes more advisory: review new ideas, check vendors, report to the board.

This change is healthy. It means that the knowledge moved into your team. Be careful with the opposite direction. A leader who starts as an advisor and never ships will not become an operator later without a change in the contract.

Why do sellers prefer advisory work?

It helps to understand the seller’s side.

• Advisory work has lower risk for the seller. A recommendation cannot fail in production.
• Advisory work scales. One advisor can serve more clients, because there is less hands-on work per client.
• Advisory work is easier to sell. A strategy sounds important. A narrow first system sounds small.

Operator work is harder for the seller. The seller must know your data, your systems and your people. The seller can fail in public. For this reason, fewer sellers offer it, and it often costs more per month. But it usually costs less per result.

What if I already have an advisor?

You do not need to stop the advisor. Change the scope.

1. Ask the advisor to select one item from the roadmap.
2. Ask the advisor to own the vendor decision for that item, with a deadline.
3. Give the advisor an internal owner and a small budget for a build.
4. Set a 60-day date for a system in production.

If the advisor agrees and delivers, you now have an operator. If the advisor refuses, you know that you need a different person for execution. Keep the advisor for strategy, and add an operator for the build.

A hypothetical example

A company with 150 employees paid an advisory firm for an AI strategy. The strategy listed 12 use cases. After four months, none was in production. The firm proposed a second phase: a detailed design for the top three use cases.

Instead, the CEO hired an operator for 90 days with one goal: put the top use case into production. The use case was the summary of sales calls into the CRM. The operator tested two tools with 50 real calls in week two. One tool was clearly better. The operator signed it, connected it to the CRM, wrote an eval set from 100 calls and trained the sales team. The system went into production in week seven. The sales team saved about three hours each week per person. The other 11 use cases waited until this one worked.

The strategy was not wrong. It was not sufficient. The company needed someone to make one decision and carry it to the end.

Next step

Ask the four questions to your current advisor or to your next candidate. If you need an operator, read how my fractional AI CTO engagement works. If your problem is pilots that never reached production, read why AI pilots never reach production. For a 90-day plan that an operator can follow, read the first 90 days guide.