AI Summary
→ WHAT IT COVERS Molly Graham interviews Follow Max Mullen, Instacart co-founder and early-stage investor, about separating AI hype from practical reality. They examine specific use cases, the speed of AI improvement, accessibility of AI expertise, and how everyday users — not engineers — can extract real value from current tools. → KEY INSIGHTS - **The Six-Week Rule:** If an AI tool failed at a specific task more than six weeks ago, that result is no longer valid. Models improve so rapidly that previously impossible tasks become routine within weeks. Treat every prior failure as expired data and retest regularly — the capability gap closes faster than most users realize. - **Prompt simplicity has replaced prompt engineering:** Until roughly six to eight months ago, getting quality AI output required learning specific prompting techniques and workarounds. That barrier has largely disappeared. Users can now state requests in plain, casual language — including incomplete sentences — and receive near-accurate results, making the tools accessible without any technical training. - **Pick one platform and build memory:** Repeatedly using a single AI assistant — ChatGPT, Gemini, or Claude — allows it to accumulate context about your role, preferences, and needs. Over time, this eliminates the need to re-explain background details with each session, producing faster and more personalized outputs without additional effort from the user. - **AI handles roughly 80–90% of complex tasks:** Follow Max Mullen used three AI platforms simultaneously to generate a full eight-day Japan family itinerary — including hotel clustering, activity pacing, and age-appropriate planning — in minutes. A human travel agent was still needed for the final booking step, illustrating where AI currently stops and human involvement remains necessary. - **Vibe coding enables non-engineers to build functional apps:** Using only spoken dictation on an iPad, Follow Max Mullen's eight-year-old built a working website with multiple games in under twenty minutes. This approach — directing AI to write and deploy code while the user only approves outputs — means functional software tools can now be created without any programming knowledge. → NOTABLE MOMENT Follow Max Mullen described uploading personal medical lab results to ChatGPT and asking it to interpret the data as a medical professional would. Rather than a diagnosis, it generated a precise list of questions to raise with his doctor, making the subsequent physician call measurably more productive. 💼 SPONSORS [{"name": "LinkedIn", "url": "https://linkedin.com/worklife"}, {"name": "Range Rover Sport", "url": "https://rangerover.com"}, {"name": "Upwork", "url": "https://upwork.com"}, {"name": "AT&T Business", "url": "https://business.att.com"}, {"name": "Wayfair", "url": "https://wayfair.com"}, {"name": "Skylight", "url": "https://myskylight.com/worklife"}, {"name": "Bill", "url": "https://bill.com/proven"}] 🏷️ Artificial Intelligence, AI Productivity Tools, Vibe Coding, Future of Work, Technology Adoption