By Andreas Pettersson, Founder, Leaders ADAPT
This is the second of the two client snapshots on the Fractional AI Executive page, written out in full. Details are generalised to protect the client. The sector is not named.
Quick answer: A professional services firm delivering client work through account teams already had AI in the building, but only a fraction of the team used it on real client work and nobody agreed on what good looked like. We built a custom research and reporting engine, published it into the firm's shared skill library so every account team runs the same process, and put an AI quality assurance framework around it. One shared standard replaced an ad hoc process.
Company shape
A professional services firm where client work is delivered by account teams. A company wide AI subscription was already in place. Usage was uneven: a few people used it daily on client work, most used it occasionally, and no two account teams produced a report the same way.
The firm's starting question
We are already using AI every day. Why does it not add up to anything.
What was assessed
Discovery with the leadership and the account leads: how a client report actually gets produced, who researches, who writes, who reviews, and where the quality varied. The answer was that the process lived in people's heads, differently in each head.
What was built
- A custom research and reporting engine. It pulls live data and writes the full client report in the firm's format. It was published into the firm's shared skill library, so every account team runs the same process rather than their own version.
- An AI quality assurance framework. Who reviews what, against which checklist, before anything leaves the firm. Human review became a step in the workflow rather than a hope.
The result
One shared standard replaced an ad hoc research process. The engine's first run surfaced faults on the firm's own website that had gone unnoticed, which is the kind of thing a shared standard finds and an ad hoc process never does.
Where the review moved
The framework put the review where it belonged, with a named reviewer and a checklist for each report, instead of wherever it happened to land that week.
Case story FAQ
What was the problem at this firm?
AI was already in the building. Only a fraction of the team used it for real client work, and nobody agreed on what a good output looked like. Every account team ran its own research process, so the quality of client reports depended on who happened to be on the account.
What was built?
A custom research and reporting engine that pulls live data and writes the full client report, published into the firm's shared skill library so every account team runs the same process. And an AI quality assurance framework, so human review is a step in the workflow rather than a hope.
What changed as a result?
One shared standard replaced an ad hoc research process. The engine's first run also surfaced faults on the firm's own website that had gone unnoticed, which is the kind of thing a shared standard finds and an ad hoc process does not.
Why a quality assurance framework rather than just a better prompt?
Because the failure mode in professional services is not a bad first draft, it is a bad draft that reaches a client. A framework that names who reviews what, against which checklist, before anything leaves the firm, is what makes AI usable on client work. A better prompt makes the draft better. The framework makes the output safe.
Which Leaders ADAPT engagement was this?
The Fractional AI Executive engagement. Details are generalised to protect the client and the sector is not named.
Which engagement this was
The Fractional AI Executive engagement. The pattern repeats across professional services: the tool is already there, the standard is not. What this shape of help costs in the market is on the AI consulting cost page, and the one page AI policy is the first artifact in every engagement like this one.
If AI is in your building and you cannot point to what it changed
The intro call is thirty minutes. Bring one client deliverable and tell me how it gets produced today. We tell you whether a shared standard is a six month engagement or a week of work you can do yourself.
Details are generalised to protect the client. The result is the same one described on the Fractional AI Executive page.
Read the first case story: a multi location services company.
