I spent the best part of a month wondering why my AI kept losing track of who I was. Turns out it wasn't me getting something wrong. If you use Cowork and you've set up Global Instructions (Settings → Cowork → Global Instructions) to tell Claude who you are, how you like to work, your safety rules... you might have noticed they occasionally revert to an old version. You update them. They look right. Days later, they're back to something you wrote weeks ago. I checked properly. It's a known bug, not user error: - Settings → Cowork → Global Instructions can silently revert to a stale saved version - It happened to me repeatedly, on two different devices - It's logged on Anthropic's own bug tracker, with several separate reports of the same thing - As of now, it's still unresolved I've linked one of the open reports below if you want to see it for yourself 👉 https://github.com/anthropics/claude-code/issues/40175 The workaround that's worked for me... don't rely on that Settings box alone. If you have a folder connected to your Cowork session, put your real instructions in a file there instead (a CLAUDE.md file, if that means anything to you). Claude reads that straight off your files every time, and it isn't affected by this particular bug. Annoying to find. Easy enough to work around once you know. Has this happened to you, or has your Claude been behaving itself? 👇

Ask an accountant how they keep client data safe with AI and you'll almost always hear the same line: "I've turned off training." It's a reasonable thing to believe. It's also, on its own, close to worthless as a compliance position. Turning off training stops one thing... the provider reusing your conversations to train future models. Everything else still happens. Your prompt still leaves your computer. It's still processed on a server, usually abroad. It's still retained for a while. It's still, in the language of the law, being handled by a third party on your behalf. If you process client data through AI, and virtually every firm does, you're the data controller and the AI provider is your processor. That's a defined legal relationship with eight separate obligations attached. The training toggle covers roughly one of them. This isn't hypothetical any more either. The PCRT bodies, including the ICAEW, published AI guidance in January. The IRS issued its first Circular 230 guidance on AI in June. The UK's Data (Use and Access) Act landed in February. None of it created new duties. All of it confirmed the existing ones apply, and that regulators are actually looking now. I put together a free guide walking through all eight things compliance actually requires, in plain English, with a simple framework for deciding what's safe to do with client data. Fifteen minutes to read. Attached below. The full course covers the other four chapters this guide doesn't (agentic AI, connecting AI to QuickBooks and Xero, prompt injection, and the 2026 rules by jurisdiction) ... that's inside Premium if you want to go deeper. Where does your firm genuinely stand on this right now... sorted, working on it, or hadn't thought about it until this post? 👇

A few months ago, someone who isn't an accountant built the world's first AI accountancy firm. Eleven AI "colleagues". Each with a name, a job title, and a workstream. And a managing partner called Grace Ledger. I sat down with Alexis Kingsbury ... the man behind it, and author of the number one Amazon bestseller Accrual Intentions ... to find out what actually happened. It's not a story about AI replacing accountants. It's far more useful than that. In the conversation we get into: - What "agentic AI" really means, and why it changed everything in 2026 - How his AI team built 108 files in 8 minutes ... and where that went badly wrong - The moment the AI confidently pushed him to take paying clients with no AML registration - Why you should never let AI do your calculations, and what to do instead - Why confident, beautifully presented AI errors are harder to spot than a junior's - What this all means for junior roles and the future of the profession My honest take ... this was one of the most eye-opening chats I've had all year. Alexis is refreshingly straight about where AI is brilliant and where it's genuinely dangerous. 👉 The recording is here (Please help support the YouTube channel by Liking the video and subscribing). Have a watch, then tell me in the comments ... would you ever let AI loose on your own year-end accounts? 👇

I wonder if all this hype about AI replacing us is overlooking an important opportunity for us. • AI isn’t replacing experts. We’re joining the conversation at a different place. Expertise isn’t necessarily being replaced by AI. We’re simply joining a more mature, focused conversation. We’re not competing with AI to maintain our relationships. We’re using the skills that AI brings to complement how we already work with our clients. • The value of expertise is as important as ever. AI didn’t invent conversations. It has simply reminded us that our relationships started with human conversations. Over time the listening, asking questions, making time for thinking together was replaced by having the answer. • AI prepares us for better human conversations. Clients come to us when they’ve organized their thinking. They’ve asked the what-if questions that they weren’t able to say out loud yet. They allowed themselves to be vulnerable because AI gives them permission to share their thinking without judgment. 💡Idea: Encourage your clients to tell you about their experience with AI. The professionals who are curious enough to have a different conversation with their clients are the ones who will be valued, not replaced.

I caught ChatGPT trying to improve a sentence that I'd written. It suggested replacing "That is what I'm still thinking about" with "That's what I'm still thinking about." There was just one problem: I'd written "That's." So I called ChatGPT on it: "Hold on. That's not what I wrote. Where did you get that?" We talked it through and realized what had happened. It wasn't looking at my sentence. It was mentally carrying over an earlier version of the conversation. It wasn't a big deal. I didn't think that AI was broken. I was really just curious how ChatGPT came up with the word that didn't exist. It struck me how differently a lot of people treat AI compared to humans. If a coworker made the same mistake, we wouldn't declare them unreliable and not fit to work with forever. We'd just ask them to hold on a minute so we could figure out where the mistake happened. People work through a misunderstanding and move on. What's funny is that the same conversation that contained the mistake also had some of the best insights. Correcting one mistake doesn't mean losing confidence in either AI or a person. I'm not right all the time, and I'm comfortable with that. Maybe AI deserves the same space to be wrong sometimes. It gives us room to question, think, and collaborate.