FRIDAY FUN: If your AI assistant had a personality and showed up as a pet, what kind of animal would it be... and why? Would it be a wise old owl that gives perfect prompts… or a hyperactive border collie that gets everything done fast but never stops asking questions? 🐍 Is it sly and strategic? 🦜 Chatty and colorful? 🦉 Calm, mysterious, and always on point? Tell us what animal your AI assistant would be—and explain the logic behind it. Maybe even give it a name! 🎯 Bonus points for attaching an image (real pet, AI-generated image, meme, or sketch). Let’s see your AI companion in the wild!

From 2 August 2025, new rules under the EU AI Act will directly affect how accountants use tools like ChatGPT, Claude, and Gemini—especially if you operate in the EU or serve EU-based clients. Here’s what small firm owners need to know (wherever you’re based): ✅ What You Need to Do 1. Be Transparent with Clients If you use AI to generate insights, draft reports, or summarise accounts, you must let clients know—either in the final deliverable or engagement letter. 2. Avoid Prohibited Uses The following are banned, even for internal use: - Emotional profiling - Behavioural manipulation - Biometric surveillance (e.g. emotion detection or facial recognition) 3. Complete a Risk Assessment Each use of AI needs to be classified: - Minimal risk (e.g. summarising public info): little action needed - Limited risk (e.g. AI-assisted report drafting): requires documentation and transparency - High risk (e.g. credit scoring, fraud detection): needs human oversight and strict controls Most AI use in small firms will fall into the limited risk category. 4. Apply the Same Logic as GDPR Treat AI tools like you treat personal data: - Record what you use, when, and why - Note who reviews AI outputs - Add disclaimers or checks where needed 5. Choose Compliant Tools Use AI platforms that: - Offer enterprise-grade security - Allow opt-outs from training - Ideally use EU-based data processing Good options: ChatGPT Team or Enterprise, Claude via Amazon Bedrock, or a local model via Ollama. 🌍 Regional Differences EU Firms: Full compliance required. Treat it like GDPR—document, assess, disclose. UK Firms : If you serve EU clients, the Act applies. If not, you’re exempt—but similar UK guidelines may still apply. Canada / US: You’re only affected if you serve EU clients or use EU-based tools—but similar laws are coming. 🧭 Practical Example A firm in Ireland must fully comply. A firm in the UK must comply only if it serves EU clients. A firm in Canada is exempt—unless it processes EU data or uses EU tools.

OpenAI has just released a brand-new AI model called GPT-OSS — and unlike most tools, this one is completely free, fully offline, and surprisingly powerful. 💡 So why does this matter to us as accountants? Here’s what makes this release a big deal: ✅ Total privacy – You can run this AI on your own machine, offline. That means no client data is sent to OpenAI, Microsoft, or any cloud provider. ✅ Zero ongoing costs – Once you’ve set it up, you can use it as much as you want with no monthly fees or API limits. ✅ Handles complex tasks – This model performs nearly as well as GPT-4 in areas like reasoning, logic, and even math — which is very promising for use in forecasting, advisory, internal planning, and more. ✅ No internet required – Great for those with patchy Wi-Fi or anyone who travels and wants a secure AI assistant on the go. Imagine having an AI assistant you fully control, with no data leakage risk and no subscription cost. This is now possible. ⚙️ You don’t need to be technical to take advantage of this — and you don’t need to start from scratch figuring it out... 👉 If you want to learn how to run LLMs locally, protect client data, and solve the AI privacy and security issue once and for all, I’ve just created brand-new training inside the AI Academy that walks you through everything step-by-step. 🔗 Join the AI Academy here: https://www.skool.com/ai-academy-3665/about Let me know in the comments what you are doing about privacy, security and client confidentiality when using AI. Is this something you are concerned about?

Ever rewatched a Zoom or Loom recording and thought, “Why do I look like I’m talking to my keyboard?” Yeah, same here. I recently saw an ad in my Facebook feed for the iContact Camera—a gadget that helps you look directly into the lens so you connect better with your audience. The idea was great... but the reviews? Not so much. So I did what any AI-obsessed accountant would do… I ran a Deep Research session in ChatGPT. What I found was a clever and affordable way to improve eye contact on camera—without buying another gadget that collects dust. It works for Zoom calls, Loom videos, or anything where face-to-face presence matters. 👇 I’ll drop the exact prompt I used in the comments so you can try it too. And if you’ve found your own tricks for looking more natural and engaging on video, I’d love to hear them!

I want to hear from you—What’s the biggest success you’ve had using AI in your practice so far? Maybe it saved you hours on a task you used to dread…Maybe it helped you deliver more value to a client…Or maybe it just made something click that used to feel complicated. Whatever it is—big or small—I’d love to know. 👇 Share your AI win in the comments so we can celebrate (and learn from) each other. Let’s build a library of real results.