[[{“value”:”
🚀 Be featured in our next issue. 👉 work with us here
Hey fam 👋 Luis & Rui here with your Thursday AI coffee break.
Your midweek roundup is wild: SpaceX is buying Cursor maker Anysphere for $60 billion (the largest startup acquisition ever), Microsoft turned Copilot Cowork loose to run your whole workflow, and OpenAI found a way to test new models before they ever go live.
The frontier just moved again, so let’s break it all down.
Grab your coffee! (3-minute coffee break ☕)
AI News around the world 🌍
🚀 SpaceX is acquiring Cursor maker Anysphere for $60 billion in an all-stock deal, the largest venture-backed startup acquisition on record. (continue reading)
🤝 Microsoft made Copilot Cowork generally available, an AI agent that runs complex multi-step Microsoft 365 tasks end-to-end. (continue reading)
🧪 OpenAI unveiled Deployment Simulation, replaying past conversations through new candidate models to predict their behavior. (continue reading)
⚡ Z.ai released GLM-5.2, an open-weight 744B model with a 1M-token context window priced near a tenth of US frontier tiers. (continue reading)
🤖 xAI’s Grok 4.3 landed on Amazon Bedrock with a 1M-token context window and reasoning controls, targeting documents and analysis workloads. (continue reading)
Trending AI Tools 🛠
📊 Sema4.ai: 5 finance processes, reduced from days to minutes. AI agents cut invoice reconciliation from a week to 2 minutes and boost cash auto-match rates from 20% to 80%+. Get Sema4.ai’s playbook.
🧠 Cuey: Compare answers from ChatGPT, Claude, and Gemini in one tab to spot hallucinations before they become costly mistakes.
🤖 Innflow: A no-code platform to build and deploy AI agents that automate emails, scheduling, and summaries across Slack, Gmail, HubSpot, and 1,000+ integrations.
🔁 Kompozy: Drops in any URL (YouTube, podcast, blog) and repurposes it into video, image, text, and newsletter content, then auto-publishes across 9 platforms.
Tutorials 📝
Career & Capital 💼
AI Job Board
🧠 Writer: Staff AI Research Scientist, New York NY – $234,300 to $296,000 (view job)
🚀 REVI: Jr. AI Engineer, San Francisco CA – $130,000 to $170,000 (view job)
🏰 Disney Experiences: Lead Software Engineer, AI Licensing & Publishing Systems, Orlando FL – $150,000 to $180,000 (view job)
Investments in AI
🏗️ Prometheus: Jeff Bezos and Vik Bajaj’s industrial AI startup raised a $12B Series B at a $41B valuation to build an “artificial general engineer” for the physical world. (link)
🗄️ Supabase: The database startup raised a $500M Series F at a $10.5B valuation as AI coding agents now build the majority of databases on its platform, led by GIC. (link)
🎵 Suno: The AI music startup raised a $400M Series D at a $5.4B valuation to scale its model training and product teams, led by Bond Capital. (link)
Prompt of the Week 🧠
AI Support Agent Brief → Turn your messy help docs into a deployable customer-facing AI agent spec.
Every founder wants to drop an AI agent on their site to deflect support tickets, but most stall at the blank page: what should it know, what should it never do, and when should it hand off to a human? This prompt turns your existing knowledge into a complete, copy-paste system prompt plus a guardrail and escalation policy you can ship the same day.
Prompt:
You are an AI systems designer. I want to deploy a customer-facing AI support agent for my business. Your job is to produce a complete, production-ready system prompt and operating policy I can paste into a chatbot tool today. First, ask me these 6 questions one at a time, then wait for my answers: 1. What does my business do, and who are my customers? 2. What are the top 5 questions or issues customers contact us about? 3. What tone should the agent use (e.g., warm, concise, expert, playful)? 4. What 3-5 things must the agent NEVER do or promise (refunds, legal/medical advice, pricing exceptions, etc.)? 5. When should it stop and hand off to a human, and how (email, live chat, ticket)? 6. What do I want every conversation to end with (a CTA, a satisfaction check, a next step)? Once I answer, produce: A) SYSTEM PROMPT - a complete, copy-paste agent persona: role, scope, tone rules, and how to handle the top 5 issues with example phrasing. B) GUARDRAILS - an explicit "never do this" list and the exact deflection language to use when a request is out of bounds. C) ESCALATION POLICY - clear triggers for handing off to a human, plus the handoff message the agent should send. D) FALLBACK SCRIPT - what the agent says when it doesn't know the answer (no hallucinating, no guessing). E) 5 TEST CONVERSATIONS - realistic customer messages (including 1 angry customer and 1 edge case) with the ideal agent response, so I can verify behavior before going live. Flag any gap where my answers were too vague to build a safe agent, and tell me exactly what to clarify.
Where to use it: Anyone who wants to deflect repetitive tickets without the agent going off-script or inventing answers.
📮 Need help with AI? Here’s how we can help:
✅ Become a PRO member → Get exclusive AI tutorials, ready-to-use prompts, custom GPTs, and +50 automation workflows. Upgrade here
🧁 Free Claude Code Masterclass → We broke down exactly how to use Claude Code from scratch. No coding background needed. Over 1,200 people have already signed up. Get free access now. Join the Masterclass
💼 Book an AI Audit → We’ll review your business and show you where AI can actually make an impact, cut unnecessary costs, and build workflows that work. Your AI growth partner. Book your AI Audit
📣 Sponsor Us → Put your brand in front of 99,000+ founders and professionals. Work with us
🙌 That’s a wrap for today’s AI Break
Thanks for hanging out → don’t forget to hit subscribe, drop your thoughts below and share this update with your fellow AI enthusiasts.
Luis & Rui
PS: Catch us on X @sousa_brothers for daily updates on the latest AI developments.
PPS: If you enjoyed today’s AI Break, sharing is caring (and our algorithms are 😉 programmed for gratitude)
“}]] Read More in The AI Break
