Find the workflows where automation actually pays back.
For ops-heavy businesses with 10 to 500 employees, the best AI project is usually not a chatbot. It is a repetitive workflow that stops eating the week.
Can this be automated, or is everyone just used to suffering through it?
When a team says "this can't be done" or "this would take forever," sometimes they mean the manual version would take forever. That is different from the automated version being impossible.
My job is to separate those cases. Some work should stay human. Some work needs a better process first. Some work is a two-afternoon script wearing a forever-problem costume.
Operations work leaves tracks.
Spreadsheets and reports
Recurring reports, crash-prone workbooks, manual exports, formula forests, and anything only one person knows how to run.
Portals and re-keying
Orders, vendor portals, customer forms, status updates, and copy-paste between systems that never quite talk to each other.
Documents and inboxes
Invoices, POs, PDFs, customer emails, routing rules, and the messy language-heavy work where AI may actually help.
We sort work into four buckets.
- Script it. Clear rules, stable inputs, exact output.
- AI-assist it. Messy language or documents, with human review.
- Document it first. The rules exist in someone's head but are not written down.
- Leave it human. The judgment is the work, or the error cost is too high.
That sorting step keeps you from paying a token tax on work that never needed AI, and from trusting a model with work that needs accountable human judgment.
Bring the workflow people complain about.
I will help you decide whether it is automation-ready, AI-assisted, or not worth touching.