This is my fifth Deep Dive on @tryramp
. In the second, I wrote about a hypothetical company growing so fast it hit $1 billion in revenue in its fifth year. Then Ramp did the meme.
BUSINESS
-
Ramp Achieves $1B Revenue in Fifth Year Deep Dive
By
–
-
ASML’s €1.7 billion investment in Mistral AI: European AI funding analysis
By
–
A few comments on ASML's investment in MistralAI There are currently many expressions of joy coming from Europe that the only well-known European AI company, Mistral AI, is now receiving relatively large funding of €1.7 billion from ASML based on a €10 billion valuation. This
-
Focus on strategy, not model choice, for future-proof AI
By
–
11/ Don’t ask, “Which model should we use?” Ask, “What can we build that GPT-6 won’t crush tomorrow?” Because the next wave is coming. Your strategy is your only edge.
-
Phase 5: AI as a System for Founders’ Leadership
By
–
9/ Phase 5: Leadership AI isn’t a feature. It’s a system. Founders must:
• Think in compounding systems
• Train teams in evals, safety, costs
• Sell AI internally with ROI, not hype -
Five principles for building durable AI that outlasts wrappers
By
–
10/ This is how you build AI that lasts:
• Moats that compound
• Differentiation users feel
• Design that doesn’t bankrupt you
• Deployment that scales smart
• Leadership that aligns the org Ignore this and get replaced by wrappers. -
Duolingo’s Data Moat from Years of Learning Data
By
–
3/ Data Moat Duolingo’s edge? Years of learning data → AI no one else can build. Ask:
• Are we collecting exclusive, structured data?
• Does every user make us smarter? No? You’re building a demo, not a product. -
MIT: Zero ROI from GenAI, AI assistant buttons are not strategy
By
–
1/ MIT says most orgs get zero ROI from GenAI. Meanwhile, teams are slapping “AI assistant” buttons on products and calling it strategy. It’s not. If you want to lead with AI – not chase – it starts here:
-
OpenAI’s 5-phase playbook to build AI products that compound
By
–
Most founders are building AI wrong. They treat AI as a feature. But the real winners build moats. OpenAI’s Product Lead just dropped the 5-phase playbook to build AI products that compound. Here’s the strategy every founder must see :
-
XPeng Iron Humanoid Robot Mass Production Begins 2026
By
–
XPeng Iron is an intelligent humanoid robot developed by the Chinese electric vehicle company XPeng. They will start mass production in 2026.
— Chubby♨️ (@kimmonismus) 9 septembre 2025
In 2026, the fierce battle for the mass market for humanoid robots will begin. pic.twitter.com/WomxJahbWsXPeng Iron is an intelligent humanoid robot developed by the Chinese electric vehicle company XPeng. They will start mass production in 2026. In 2026, the fierce battle for the mass market for humanoid robots will begin.
-
Prompting Power: Skip Consultants, Build Closer, Move Faster with LLMs, Data, Questions
By
–
If you know how to prompt, you can: – Skip the consultants
– Build closer to your customer
– Move 10x faster The new market research stack = → LLMs
→ Data
→ Clear questions Prompt smarter. Build better.