Find the work worth automating.
The first hour of our workshop, written down so a small team can run it alone. Ninety minutes, one room, the people who do the work. You leave with a ranked shortlist and one thing to try first.
For teams of a few to a few dozen people. Larger organisations usually need the full day with us, because the interesting work crosses team boundaries.
- 1
List what repeats
20 minEveryone writes down, alone, the pieces of work they did more than three times last month. Not projects, not goals: the actual repeated jobs. "Answering the same policy question." "Chasing a supplier for a delivery date." "Re-typing an order from an email into the system." Aim for ten each; short and concrete.
The best entries sound boring. Boring is what repeats.
- 2
Cross out the judgment calls
10 minRead them out. Strike anything where a person has to decide something that matters and be accountable for it: pricing, hiring, safety, a complaint from an angry customer. Keep the rest. Keep also the jobs where the judgment is small and the checking is large.
A job that is "mostly copying, with a decision at the end" stays on the list; the decision can stay with a person.
- 3
Score what is left
25 minFor each job, three quick scores out of three. Worth it: how much time or error it costs today. Effort: how hard it looks to hand over. Data: does the answer already live in a system you can name, or in someone's head? Then one name: who would own it if it were automated.
If nobody can name where the answer lives, score data as one, not zero. It is a question to ask, not a reason to stop.
- 4
Pick one
15 minSort by worth minus effort. Take the top one that has a named owner and a data score of at least two. That is the job you try first. Write the one-line version: who does it today, what comes in, what goes out, what "done" looks like.
Resist picking the most exciting one. Pick the one you can check in a week.
- 5
Test it before you fund it
20 minAnswer four questions about the job you picked. Where would the answer come from? Which system holds it? Who grants access? What happens on a bad day? Most ideas fail here, and failing here costs nothing. If it survives, you have something worth a first working version.
Sometimes the honest answer is a rule in a system you already own, or a process change and no software. That is a good outcome too.
The scoring sheet
| The job | Worth it (1–3) | Effort (1–3) | Data (1–3) | Owner |
|---|---|---|---|---|
| Answering the same policy question | 3 | 1 | 3 | |
| Writing first-draft supplier replies | 3 | 2 | 2 | |
| Matching invoices to delivery notes | 2 | 2 | 3 | |
The three filled rows are examples, not results. Print it, or copy it into whatever your team already uses.
What to do with the shortlist
Keep it. Revisit it in a quarter. If the first job works, the second is already ranked. If your team grows, or the interesting jobs turn out to cross into other teams' systems, that is the point where a day with us earns its place.
Discuss a day with your teamsWritten in September 2026 by Nexus. Free to use inside your company. It describes a method, not a result; what your team finds depends on your team.