He did but Sam made it look like he was poaching OpenAI exclusively. This helps attract all those Zuck didn’t hire (because he doesn’t need all the good folks out there) to go find roles at OpenAI.

This is the definition of GTM.

The web is going to become a collection of walled gardens. Just like physical borders exist, I think content will go the same way. Substack, Quora and Reddit could very easily wall off search engines and only work with ChatGPT. Or they could build their own ChatGPT like.

AI is needed to build the brain to fly these autonomously.

May we scale smoothly, exponentially and uneventfully through superintelligence. - Sam Altman.

This is not the unbundling of Bloomberg. This is the unbundling of fintech. The tech in fintech is being replaced by ai. Finai will be larger than fintech. AI will give us tools that we could never have imagined. Give us workflows that we would never have even dreamed of.

7️⃣/7. 🗣️ Your turn: Would an API-first Scope-3 engine change how your org reports carbon? Where would it plug in, and what blockers do you foresee?. Share objections, wishlist features, or “shut-up-and-take-my-money” GIFs . 👇let’s debate before we build. #AI #ClimateTech #LLM.

6️⃣/7.💸 Why CFOs pay:.✓ Slash audit prep from 6 weeks to 6 hours.✓ 30–50 % cost cut vs Big-4 consultants.✓ Hard numbers unlock green financing discounts (literally cheaper loans). ROI speaks.

5️⃣/7.📊 UX that wins budget:.• Green-amber-red dashboard for procurement VPs.• API endpoint /v1/carbon for ESG tools.• Auto-generated disclosure text that drops straight into EU CSRD reports (zero copy/paste).

4️⃣/7.🚚 Real-time pipeline:.Invoice lands ➡️ parsed in <2 s ➡️ emissions estimate stored ➡️ webhook fires to Slack:. “New PO #8732 adds 4.6 t CO₂e (electronics, air freight).”.Ops team sees it before the ink dries.

3️⃣/7.🛠️ Model stack (tiny but lethal):.1.OCR → layout-aware transformer (detects line-items).2.NER fine-tuned on logistics terms (“kg”, “teu”, “lane”).3.Carbon factors lookup graph (GHG Protocol).4.LoRA-tuned LLM that outputs CO₂e / SKU / supplier with confidence scores. Runs.

2️⃣/7.👀 Data you already have:.• PDFs of vendor invoices & purchase orders.• Freight tracking feeds (EDI, CSV, APIs).• ERP tables (SAP, Oracle). Just messy enough that no off-the-shelf GPT prompt nails it. Opportunity spotted.

1️⃣/7.🌍 Enterprises sweat over “Scope 3 emissions”—the carbon hidden in every invoice, shipment & vendor contract. Manual spreadsheets ≠ scalable. Time for a micro-model to own this pain. #ThousandMistrals.

5️⃣/5.Let a thousand Mistrals bloom. Drop the one niche you can’t stop thinking about ⤵️. Builders are watching. Investors are too. #AI #LLM #startup #buildinpublic #ThousandMistrals.

4️⃣/5.If you’re a maker wondering, “what’s left to build?”.👉 List 5 domains you know better than OpenAI does. 👉 Cross with modality (text/voice/vision) + task (classify/summarize/generate). Somewhere in that grid is your unfair advantage.

3️⃣/5.Why they win:.moats come for free (domain-specific corpora).2.Infra bill ≪ VC round—run on a GPU you actually own.3.DevRel > Ads. APIs, docs, demos, OSS kits… builders love tools that disappear.

2️⃣/5.Think:.• Voice-native LLMs for India’s 19,500 dialects. • PII-sniper models that catch secrets before they hit Git. • 50 MB edge nets moderating content on cheap phones. Niches so sharp they feel obvious in hindsight.

1️⃣/5.🚀 The next trillion-$ opportunity in AI isn’t “another GPT.”. It’s a thousand micro-models laser-focused on problems nobody else sees—yet. #ThousandMistrals.

🧒📚 Parents don’t need more options—they need trusted curation. Imagine a model trained on:.• child-development research,.• diversity & empathy scoring,.• thousands of parent reviews. Output? Three perfect books or toys for a specific age, interest, and budget—no.