Every week: the AI news that actually matters to a finance function, a prompt you can use on Monday, and the hype you can safely ignore. Written by Umar Din FCCA, who trains finance teams to use AI on real month-end work.
Model launches, pricing changes and agent releases, translated into what they mean for a finance team.
One worked prompt each week, built for real finance tasks like variance commentary or a board paper.
What is worth acting on this week, and which headlines you can safely skip.
1 Oct 2026
OpenAI dots, Meta Muse, Instinct and Microsoft Autopilot explained for finance teams. Why agents go off-script, plus two prompts to use them safely.
24 Sept 2026
Two new models, half the price, and a test to see if you're overpaying.
20 Sept 2026
Anthropic's own CFO says Claude writes his statutory financials. What that proves about AI in finance, and what it doesn't.
10 Sept 2026
OpenAI declared the AGI era. The same model scored 62.7% or 99.9% depending on setup. What that means for finance teams, plus a prompt to test any model.
3 Sept 2026
Anthropic reversed its data retention policy for enterprise customers. What Enterprise Frontier Safeguards changes if your risk team blocked Claude.
29 Aug 2026
Anthropic merged Claude's memory across chat and Cowork. Why persistent context is a governance question, plus a research brief you can reuse five ways.
20 Aug 2026
Stripe's $7bn bet, and why UK firms are hiring because of AI, not despite it.
13 Aug 2026
A practical guide for finance managers: how to prompt properly, what AI agents actually are, and how to build your first trustworthy finance agent this week.
6 Aug 2026
New OpenAI and Gallup data on how finance teams use AI, plus a RACEF variance commentary prompt you can test on a month you've already closed.
Prefer longer reads? The best editions become in-depth guides on our blog.