← The notebookField notes · 5 MIN READ

What an AI consultant actually does

Owners ask me two questions that are really one question. What does an AI consultant actually do? And if we hire one, what does the engagement look like from the first call to the last?

A demo, or a pile of licenses, is not an engagement. A real one has a start, a few decisions in the middle, and a number at the end that tells you if it helped.

I'm Co-founder of AI Answered. We do this work with established and larger companies. If you want to see how we take it on, start with how we work as a growth partner. It starts with your numbers.

First, the business and the numbers.

I want to understand how the company makes money, where the time goes, and which numbers the people in charge already trust.

Revenue. Margin. How long a lead sits. What a repeated task costs in hours. Where mistakes show up and what they cost. If those are fuzzy, the engagement starts by making them clear. Skip that, and every later result is a story nobody can check. I wrote the longer version here: AI won't fix a business that doesn't know its numbers.

A plan built on a guess about the business is a guess with extra steps.

Then, where the payoff actually is.

Most companies can list a dozen places the work might help. Sales, support, the site, a slow proposal. A list is not a plan.

We look for a place where a change would show up in a number you already watch. Less time on a task the team does every week. A shorter wait. Fewer dropped handoffs. A step that is holding revenue back.

Choosing one task inside a single team is its own skill. I wrote about that in where to start with AI in your business. An engagement is the wider look. Across the company, where is the payoff real, and where would a project mostly produce a demo?

A few moves, not a pile of them.

Once those spots are visible, we pick a few. A few we can finish. Each one has an owner and a number we will check later.

A move is specific. "Use AI in operations" is a theme. "Turn a sales call into the follow-up the team will actually send, and spend less time doing it" is a move.

We price on the value we can deliver, so this choice is the job. If we can't see the value in the numbers, we wait to build.

Build it with the team, then put it in the week.

People picture the build as the whole engagement. In a good one, it is the middle.

We write down how the work runs today. We try the new version on real inputs. Someone on your team checks the output before a customer sees it. When it holds up, it goes into the week the team already has. Training is a few concrete tasks with the people who do the job. They should be able to run it, spot a bad result, and say so.

If the only person who can explain it is the consultant, it never rolled out. It just visited.

Measure what changed.

Go back to the number from the start. Did the hours drop? Did the wait get shorter? Did errors move? If the point was revenue or margin, did that move?

A number that stayed put is still an answer. We stop, change the approach, or pick a different move.

What we need from your side.

This is where engagements stall.

We need the numbers. We need the people who know how the work really runs, not the version from a slide. We need one owner on your side who can say when something should ship, change, or stop. We need time each week to look at the output and answer questions. We need you to tell us when a result looks wrong.

You don't have to become the builder. You do have to stay in the work. A company that sends a login and leaves the room gets something nobody uses.

If you want a plain check on whether the company is ready for that, it's in how to tell if your company is ready for AI. This note is what the engagement looks like. That one is the look before you start.

What you should have when it's done.

You should know what got built, where it lives, who on your team runs it, and which number it was supposed to move. Your team should be able to keep it, and still own the work if the relationship ends.

If you are choosing who to hire, the questions I'd bring to that call are in what to look for in an AI consultant. Use them on anyone. Use them on us.

Want to see if we're a fit? Start on the growth partners page. We'll look at the business and the numbers first. If we can't help, you should hear that early.

For people on the team who want the tools in plain words, Noah Kruthaupt and I wrote AI for Boomers. Everyday tasks, step by step. The aim is the same as the work. Make the complex simple.

More from the notebook ↗