Contents · 6 min read
01 / 06 · The proof
What must the first phase prove?
In short
Agree the proof before work starts. Leaders should make better decisions first. Working agents and ranked choices should follow. Wider gains come after those choices change how the firm works.
Take a wealth manager whose value rests on senior advisers’ client ties. Its first test is whether leaders can use AI to decide which relationships need a succession plan first.
The CEO is deciding whether to back a wider AI program. A clear financial case would help, and asking for one is fair. But the firm has not yet changed its work. Any large gain is still a forecast.
For the wealth manager, the first phase must prove that its leaders can use AI to decide which client relationships need a succession plan first.
The test is a better succession decision made in the leaders' own work.
02 / 06 · Week one
Better judgment should start in week one
In short
Leaders should use AI in their own work from week one. The first result is a better decision on a real business choice.
For the wealth manager, the CEO and team start with the client ties. They use AI to prepare real work. They test their own view of what may change.
A class can build skill, and that helps. But the test is what leaders can now see or decide. A sound first phase can start in week one without access to live systems.
The work should stay close to live choices. For this firm, that means the future of its client ties. Leaders learn what AI can do by using it on that issue.
The partner must turn what leaders learn into working agents the firm can keep.
03 / 06 · Working proof
When should working agents appear?
In short
Working agents should appear within three to four weeks. Leaders should be able to use them and judge their work. The firm should own them.
The first phase of GRAIL's program puts working agents in leaders' hands within three to four weeks. That is a short test with real output.
For the wealth manager, each agent should reflect the firm's own business logic. Its leaders should be able to use the agent. They should also judge its work against their own needs.
A polished demo can teach the team, and that has value. But a demo alone is weak proof. The stronger result is a working agent that the firm owns.
The agent does not prove a large financial gain. It proves that leaders can use an agent built around the firm's own logic.
04 / 06 · Ranked bets
Who turns the evidence into ranked priorities?
In short
Management should rank three to five opportunities by business return and strategic relevance. The partner supplies the working tools and the proof behind them. Management owns the choices across the firm.
A long list of use cases can show broad interest. That is useful. But it gives the CEO no clear choice about what comes next.
For the wealth manager, the team should first name the one or two core changes. The fate of its client ties is one such change. Each opportunity should then face two tests.
Business return
Management asks how the opportunity could improve the firm. The answer must come from its own work and data. There is no standard figure that fits every company.
Strategic relevance
Management asks whether the opportunity acts on a core change. For this firm, that means the strength and life of its client ties.
The partner must deliver working agents and evidence from leader use. The partner should not make the business choice for the team.
The team orders three to five opportunities by business return and strategic relevance. The highest-priority opportunities must now change how the firm works.
05 / 06 · Changed work
What creates real impact?
In short
A priority creates wider impact only when the firm changes its work. People must use the new flow and act in new ways. Those changes create the base for later gains.
For the wealth manager, a chosen agent must sit inside changed work. Managers must decide where people use it. They must also decide who judges its output.
A pilot added to the old flow may work at the task it was built to perform. The AI program reports it as a success, and it is. Yet the result stops at that task because the firm still works in the old way.
That contrast matters. AI pasted onto old work can automate around the edges. Real change needs new work and steady tests by the people who run it.
Early use gives the firm proof. Redesigned work can then make client relationships less dependent on one adviser. Financial impact can be judged after the firm sees that change in its own data.
The order protects the CEO from false claims. It also stops a sound first phase from being judged by a result it cannot yet produce. Only then can the CEO judge the case for more work.
06 / 06 · The gate
When should the CEO commit to the next phase?
In short
Commit only when the last phase has produced enough proof. Ask for a clear result at each gate. Do not approve scale based only on a forecast. Wait until the firm has changed how it works and reviewed the results.
The CEO is right to ask when money will show up. But the answer must rest on the firm's own proof. For the wealth manager, the CEO can use five gates:
Start with direct use
The management team uses AI from week one. The partner helps make that use real.
Ask for owned proof
By week three or four, the partner hands over working agents. The firm owns what was built.
Make management choose
The team ranks opportunities. Business return and strategic relevance set the order.
Change the work
Managers rebuild the work around the chosen priorities. They own the change in how people act.
Earn the next phase
The CEO reviews the proof before approving scale. If the proof is weak, the firm can stop and keep the work.
This sequence does not claim a financial gain before changed work has produced one. The next commitment follows the proof already in hand.
Set the first proof
The management team’s next task is to define what the first phase must prove, then judge whether that evidence earns a wider commitment. Across three workshops over three to four weeks, See The Opportunity puts AI into leaders’ own work and produces custom agents for each leader. The evidence from that use becomes a ranked roadmap of opportunities, giving management a clear basis for deciding what to fund next.
Three to four weeks, three workshops with your management team: hands-on use of AI, custom AI agents for each leader, and a ranked roadmap of opportunities. Fixed fee, no access to your live systems.
Questions CEOs ask about this
What concrete results should our leadership team expect from an AI engagement, and when?
An executive AI engagement should improve leaders’ judgment from week one. The CEO must decide whether to fund wider change; getting it wrong approves scale without proof. Within three to four weeks, leaders should have working agents. They should also have ranked priorities based on real use. Those results show whether the next phase has earned support.
When should working AI agents appear?
Within three to four weeks. Leaders should be using them in their own work by then. The firm should also own what was built.
When should we expect ranked opportunities rather than a use-case list?
By the end of the first phase. Management should rank opportunities by business return and strategic relevance.
Which results depend on management and which belong to the partner?
The partner builds the working tools and supplies proof. Management decides which opportunities matter across the company and owns the change in work.
How should early results connect to later financial outcomes?
Leaders use AI on a live choice. They receive agents the firm owns. Management ranks three to five opportunities by business return and strategic relevance. Managers redesign the chosen work. The CEO reviews company results before funding scale.
Founder, GRAIL
Johan has spent fifteen years in management consulting and executive leadership, including as Chief Product Officer of a SaaS company. He founded GRAIL to help European leadership teams work out what AI changes at the core of their business. He is the author of The Execution Revolution and the host of the ThinkRoom podcast.
Related reading
Sources
- What we do: the three-to-four-week first phase and the working results it should produce. (accessed October 8, 2026)
- Why GRAIL: results from week one, client ownership and self-contained phases. (accessed October 8, 2026)
- Wealth management CEO briefing: why client ties and succession sit at the core of the firm's value. (accessed October 8, 2026)
- You're Reading the Wrong Curve: why AI added to old work does not amount to wider change. (accessed October 8, 2026)
Next scheduled review: January 21, 2027