Consulting an SME on AI, in five conversations

Most AI consulting work is sold as a sprint. A two-week audit, a slide deck, a roadmap. The slide deck is sometimes good. The roadmap is sometimes followed. Often, six months later, the SME has spent more on tools than it makes in margin and is no further forward than when we started.

I run my SME engagements differently. They are not sprints. They are five conversations, in a particular order, over six to twelve weeks. None of them require a deck. All of them are designed to leave the SME more capable, not more dependent. Here is what each one is for.

## Conversation one; the actual business

I do not start with AI. I start with the business. I want to know what the founder is paid to think about, not what they would like AI to do. I ask three questions. Where do you lose the most money to manual work this month? What is the most repetitive customer interaction you handle? What is the boring task your best person spends two hours a week on that they hate?

The answers are usually surprising. They are almost never the things the founder mentioned in the intro call. The intro call is theatre. The first real conversation is diagnostic.

## Conversation two; the candidate use case

Once I have the diagnosis, I bring back exactly one candidate use case. Not three. Not five. One. Founders have lots of ideas. The job in conversation two is to let go of all the others and commit to the one that matters.

The candidate use case has three properties. It saves real hours, or makes real revenue, in the next sixty days. It can be tested with a tool that already exists; not built. It does not require a data project to start.

If a use case fails any of those three tests, it is the wrong first use case. Not bad; just not first.

## Conversation three; the boring tooling

The third conversation is the one most consultants skip. We sit at a table with the actual tool open and configure it together. The founder watches. The operator they are going to hand it off to participates. The goal is not to ship a polished product. The goal is to make sure the founder has seen, with their own eyes, that the tool works on their data and produces a result they trust.

The founder leaves this conversation with the tool already integrated and a small playbook for the operator who will run it day-to-day. I leave the conversation with one simple metric we are going to look at in conversation four.

## Conversation four; the real numbers

Three to four weeks later, we sit down again and look at the metric. Hours saved per week. Tickets resolved per agent. Conversion on the part of the funnel we touched. We are not looking at vanity numbers. We are looking at the one number we agreed on in conversation three.

Two things happen in conversation four. Either the number moved, in which case we extend, or it didn''t, in which case we kill it cleanly and try a different candidate. The ability to kill a use case quickly is the SME''s superpower. Most large enterprises cannot do this. SMEs can. Use the advantage.

## Conversation five; the handoff

The last conversation is the one I am most proud of, because it is the one that ends the relationship. We sit down and write the SME''s own one-page playbook for evaluating future AI use cases. It is in their voice; it uses their examples; it lives on their drive, not mine. By the end of the conversation, the SME has internalised the diagnostic and can run the next round without me.

That is the goal. Not to be retained. To be replaced. The SMEs that have grown the most after working with me are the ones that ran their next three use cases on their own and only called me back in twelve months for a strategy review.

## Why five conversations beats a sprint

A sprint generates artefacts. The five-conversation model generates capability. Artefacts age fast. Capability compounds. Most of my repeat business comes from founders whose first engagement looked like a slow conversation and whose second engagement, two years later, was an entirely different problem.

If you are an SME founder reading this and considering AI work, the question to ask any consultant in the first call is simple. "At the end of this engagement, will my team be able to run the next one without you." If the answer is no, find a different consultant.