AGI, Islam, and 20 Years of AI | Waleed Kadous, Cluesmith

Описание: 40+ patents. 20+ years building AI at Uber, Anyscale, and Canva. Then Waleed Kadous left corporate, learned Claude Code, and decided to build solo.

His thesis: what used to take a team of 20 can now be done by a team of two. So he's doing it — bootstrapping Cluesmith, a micro venture studio, building an open source Islamic AI that answers 30,000 questions a month, and teaching a course called AI for Imams.

In this conversation, we get into why he left, what he thinks about AGI, and why he believes entrepreneurial generalists will win the AI era.

We cover:
→ Why he walked away from corporate AI after 20 years
→ Building a bootstrapped micro venture studio at Cluesmith
→ His AGI framework: selling something, part of a cult, or a dilettante
→ What's actually missing — metacognition, embodiment, training data
→ Ansari: open source Islamic AI and fatwa shopping
→ Why he'd never build a non-RAG app again
→ $700B in AI infra, 95% of pilots fail — bottom-up vs top-down
→ The 20-60-20 rule for AI adoption
→ Vibe coding: "It's a chainsaw — you can do a lot, but you can hurt yourself pretty badly"
→ Why entrepreneurial generalists will win

---TIMESTAMPS:
0:00 — From corporate AI to solo founder
1:25 — What he's building at Cluesmith
2:39 — "AGI is imminent" — selling, cult, or dilettante
4:30 — XAI and what's happening there
4:55 — What's actually missing for AGI
7:17 — Demis Hassabis and who he respects
8:14 — Ansari: Islamic AI and fatwa shopping
9:48 — Why he'd never build a non-RAG app
12:09 — 8 years running AMCRAN and sadqa jariya
15:01 — $700B spent, 95% of AI pilots fail
16:42 — The 20-60-20 rule for AI adoption
17:50 — Entrepreneurial generalists will win
20:45 — Quick fire: AGI, vibe coding, Claude Code, UBI

---Product in Motion is a show about how AI-native startups make decisions while the rules are still being written.