I have been spending a lot of time these days building AI systems for companies and as these capabilities are beginning to mature I am spending more time building evaluations. Here is a resource that has helped me, and I hope it helps you too. I was so inspired in fact that I will be teaching my own version in October!

𝗧𝗵𝗲𝘀𝗲 6 𝘄𝗵𝗶𝘁𝗲 𝗽𝗮𝗽𝗲𝗿𝘀 𝗳𝗿𝗼𝗺 Google 𝗮𝗿𝗲 𝗮 𝗴𝗼𝗹𝗱 𝗺𝗶𝗻𝗲 I often refer to them every now and then If you're trying to Adopt AI in your business, I'd highly recommend you to go through these at least once - 1️⃣ 𝗙𝗼𝘂𝗻𝗱𝗮𝘁𝗶𝗼𝗻𝗮𝗹 𝗟𝗟𝗠 𝗠𝗼𝗱𝗲𝗹𝘀 𝗮𝗻𝗱 𝗧𝗲𝘅𝘁 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝗼𝗻 🔗 https://buff.ly/DDfoJyq 2️⃣ 𝗣𝗿𝗼𝗺𝗽𝘁 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 🔗 https://buff.ly/NVI5Fv6 3️⃣ 𝗘𝗺𝗯𝗲𝗱𝗱𝗶𝗻𝗴𝘀 & 𝗩𝗲𝗰𝘁𝗼𝗿 𝗦𝘁𝗼𝗿𝗲𝘀 🔗 https://buff.ly/PFwO1Na 4️⃣ 𝗔𝗴𝗲𝗻𝘁𝘀 (𝗠𝘆 𝗳𝗮𝘃𝗼𝘂𝗿𝗶𝘁𝗲 𝗼𝗻𝗲) 🔗 https://buff.ly/l1O4lDh 5️⃣ 𝗦𝗼𝗹𝘃𝗶𝗻𝗴 𝗗𝗼𝗺𝗮𝗶𝗻-𝗦𝗽𝗲𝗰𝗶𝗳𝗶𝗰 𝗣𝗿𝗼𝗯𝗹𝗲𝗺𝘀 𝗨𝘀𝗶𝗻𝗴 𝗟𝗟𝗠𝘀 (𝗺𝘂𝘀𝘁 𝗳𝗼𝗿 𝗔𝗜 𝗔𝗱𝗼𝗽𝘁𝗲𝗿𝘀) 🔗 https://buff.ly/HEd79Sq 6️⃣ 𝗠𝗟𝗢𝗽𝘀 𝗳𝗼𝗿 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗔𝗜 🔗 https://buff.ly/yWSAxGC

I kept seeing MVPs take weeks, so I tried a tighter loop. After 20+ Lovable builds and $29k in a few months, here’s the flow: Day 1: problem → 1-page PRD → static skeleton Day 2: 1–2 core features (tiny scope) Day 3: auth in 2 prompts Day 4: Stripe + SEO basics Day 5: clean deploy + demo PRD kickstart prompt (use Perplexity first): You will create a PRD I can paste into lovable.dev. Ask 5–8 clarifying questions in one list. Wait for answers. Then output: 1) ~50-word summary 2) Pages with exact routes + 1-line purpose 3) 6–8 user stories + 1 acceptance check each 4) Data objects (names + behaviors; no fields) 5) UX flow (happy path + empty + failure) 6) Two-week plan 7) Copy (3 hero, 5 microcopy) 8) Skeleton build prompt Constraints: non-technical, consistent route names, no DB fields. Stripe prompt (clean template): Add Stripe subscriptions and a simple paywall. Plans: - Basic $29/mo - Pro $290/yr Requirements: - Update subscriber status in real time - Lock premium pages until subscription is active Two quick tactics: Verify skeleton with a Route Snapshot that matches your PRD character for character. If anything drifts, restore and rerun smaller. Debug faster with “investigate first”: list 3 likely causes and how to confirm each, wait for approval, then implement. If this is useful, I can share the full checklist and the exact prompts I run (free Notion). If mods are cool with links I’ll drop it in the comments, or DM me “loop” and I’ll send it. I also host a small weekly Q&A where we swap prompts and unblock each other.

I found one insightful post on LinkedIn from Mustafa Suleyman, CEO Microsoft AI : Here is my view : AI should optimize for the needs of the user not ask the user to believe it has needs itself. It should help humans by automating repetitive or manual tasks, always under human supervision. What it should not do is flip the relationship, where humans feel like they exist to serve or protect the AI.

Hey everyone! I've been thinking a lot about the bigger picture with AI. It's a game-changer, and it feels like we're just scratching the surface. So, I'm curious: How do you think AI will change your industry in the next 5 years? What's one prediction you have for how it will impact the way we work, create, or do business? Share your thoughts below! I can't wait to see what you all come up with.