Notes

The instinct is to add a tool, a role, a process or a dashboard.

Often, people are already carrying more activity than the work needs.

The hardest and most valuable work is deciding what should stop.

Nobody gets praised for removing a dashboard, which is part of why they survive so long.

Reporting records what happened. It should also help shape what happens next.

Start with the decision: who makes it, what they need to know and when.

Then choose the few numbers that can inform it.

If nobody can name the decision a report serves, it is a habit rather than a report.

When work slows down, the reflex is to redraw the structure.

First, check whether people know who decides, who contributes and when to escalate.

Clear decision rights can resolve a delay without moving anyone.

Reorganising is the expensive way to answer a question you have not asked yet.

A company can appear well organized until the owner steps away.

Then routine decisions wait for permission.

Check whether people have the information, authority and trust to act.

The test is simple. Go away for two weeks and see what is still sitting there waiting for you.

The temptation is to make the existing process faster.

But a confused process can produce mistakes faster, too.

Establish what should happen, who is responsible and where judgment is needed.

Then automate what is ready.

Messy data, vague processes and unclear decisions give AI a poor foundation.

Start with the work: what happens, who decides and which information can be trusted.

Then you can see where AI helps, and where it only adds a confident voice to a bad number.

For Ripley, I keep the durable record in readable, versioned files.

Clear names, standing rules and a running log make decisions and corrections inspectable.

The model can use the record; I can check and change it.

I can open any of it, see when a line was written, and change it. That matters more than which model is reading it.

A booking tool can reveal how clients choose. An operations tool can expose a difficult handoff.

Its limits may reveal a need you could not have specified at the start.

The first version is research you get to keep.

Duplicate spreadsheets, hallway approvals and tasks done twice are clues.

Ask what each workaround makes possible and why the formal process falls short.

The people carrying the complexity can often explain it most clearly.

Start with how they are doing.