Notes on AI employees for client work
How firms can hand repeated client work to AI, and keep review, judgement and client data where they belong.
What is an AI employee? How it differs from an AI assistant or an AI agent
An AI employee holds a role: its own clients, services, calendar, reviewer and record. Here is how that differs from an AI assistant or a general AI agent, and why it matters for client work.
29 October 2026 · 2 min readHow to review work an AI prepared, without redoing it
Reviewing AI-prepared work line by line wastes the time it saved. A practical way to review it: focus on the few things that need judgement, check sources, tick assumptions and let risk set the depth.
29 October 2026 · 2 min readTest an AI on your past work before it touches a client
A demo proves an AI works on someone else's example. A probation run on your own past jobs, compared with what your team actually delivered, shows whether it works on yours.
29 October 2026 · 2 min readHow to stop chasing clients for documents
Chasing clients for statements, invoices and sign-offs eats hours every month. A practical system: ask once with a clear list, make uploading easy, start on what has arrived and follow up on a calendar.
29 October 2026 · 2 min readAI for accounting firms: which work to hand over first
Where should an accounting firm start with AI? Start with work that repeats every month, has clear inputs and a checkable result, and keep judgement and sign-off with your people.
29 October 2026 · 2 min readClient data and AI: questions to ask before you say yes
Before any AI tool touches client data, ask how it is kept apart, whether it trains models, how long it is kept, who can see it and what is logged. A checklist for service firms.