Every business I've worked with on AI implementation has had the same gap. Not a technology gap — a visibility gap. The owner knows, roughly, how things work. The team knows their own part. But nobody has ever sat down and traced a single workflow from start to finish, through every system it touches, noting every manual step along the way.
That exercise — which takes about two hours — is more valuable than any AI tool. And it's the step almost every implementation skips.
Why This Step Gets Skipped
It doesn't feel like technology work. When a business decides to "implement AI," the instinct is to open the tool and start using it. Mapping workflows feels like admin, like delay, like the opposite of getting on with it.
But here's what happens when you skip it: the AI ends up connected to nothing. It has no live data, no business context, no understanding of the process it's supposed to support. So it stays as a text generator rather than becoming a workflow engine. The implementation stalls. The subscription keeps running.
The system map isn't delay. It's the foundation that makes everything else work.
What the Session Actually Looks Like
When I do this with a new client, we sit down together — at their office, not on a call — and I ask them to walk me through one specific activity. Not a description of how it's supposed to work. A live walkthrough of what actually happens.
Pick something that happens at least weekly: quoting a new customer, processing a supplier invoice, chasing an overdue payment, preparing the numbers for a management review. Anything that takes time and involves more than one system.
As they walk through it, I'm noting everything: every tool they open, every file they create or copy from, every decision they make, every step where information moves from one place to another. When it gets handed to someone else, I ask that person to walk me through what they do with it.
Two hours in, we have a map on paper. It's rarely pretty — and that's exactly the point.
What the Map Typically Reveals
Every business I've done this with has been surprised by something. Usually several things.
- Duplication. The same piece of information being typed into two different systems because they don't connect. Customer details entered in the CRM, then re-entered in the accounting software. Job information logged in the scheduling tool, then copied into an invoice manually.
- Invisible bottlenecks. A step that only one person knows how to do. A weekly export that depends on someone remembering to run it. A manual check that exists because a system once gave wrong data and nobody ever fixed the root cause.
- Wasted outputs. Reports that take an hour to produce and sit unread. Summaries emailed to the owner that get filed without being looked at.
- Connection points. Places where two systems could easily talk to each other — but don't. These are often where an AI workflow can be inserted with the least disruption and the most immediate payback.
How to Do This Yourself
You don't need a consultant in the room to run a version of this. Pick one workflow — something you do at least weekly — and spend an hour tracing it honestly.
Write down every tool you open. Every file you create. Every decision you make. Every time information moves from one place to another, manually. Every step that only works because you or one specific person knows how to do it.
At the end, ask yourself three questions: Where is the same information being entered more than once? Where does the process depend on someone's memory rather than a system? Where is time being spent producing something nobody uses?
Those answers are your starting points. The highest-value AI workflow is almost always sitting at the intersection of the most time lost and the most money at risk.
What Comes Next
Once you have the map, the implementation question becomes much simpler. Instead of "how do we use AI?" you're asking "which specific step in this specific workflow should we automate first, and what does it need to connect to in order to work?"
That's a question with a concrete answer — and one that makes the difference between building something that runs every week and compounds over time, versus buying a subscription that sits unused.
If you'd like to run this session together — in person, at your office, with your team — book a free discovery call and we'll go from there.