Uploaded February 2026 | Updated September 2026, 2 weeks ago

---

Formal verification already consumes years of human effort.

In this episode, Lukas Biewald talks with Carina Hong, Founder & CEO of Axiom, about why verification is becoming the real bottleneck in high stakes AI systems.

They discuss how Axiom uses AI to take on the tedious checking that stretches verification cycles across years, starting with formal mathematics and extending to hardware and software. 

Carina also explains why Axiom’s approach to auto-formalization mirrors spec driven models like Kiro from AWS.

Since recording this episode, Axiom has announced that AxiomProver solved all 12 Putnam problems from the Putnam Exam held on 6th December, 2025, and released the full proofs on 7th January, 2026. Read more here:
axiommath.ai/territory/from-seeing-why-to-checking-everything

Connect with us here:
Carina Hong: linkedin.com/in/carina-hong
Axiom: linkedin.com/company/axiommath
Lukas Biewald: linkedin.com/in/lbiewald
Weights & Biases: linkedin.com/company/wandb

Songs from Carina’s Playlist-
music.youtube.com/watch?v=wGaGtUE99hM&si=odw9xjpmN8EE4d-q
music.youtube.com/watch?v=Oo_1LvfDGa0&si=BQ7LQOKLhrCdg4Rh
music.youtube.com/watch?v=ifJzxwyzVvo&si=2OqbFEt8F3Bvl_Rf
music.youtube.com/watch?v=3FR4d6yq5YE&si=DMD_GKEqm5Brtj-j
music.youtube.com/watch?v=ah9a9KtGxiY&si=HkeGpcqcuQx4Pkih

00:00 Trailer
01:12 Introduction
02:08 Understanding Axiom's Reasoning Engine
06:32 Human vs AI in Mathematical Problem Solving
22:47 Applications and Future Milestones
25:29 Code Migration and Database Consistency
29:30 Auto Formalization and Its Challenges
36:48 Commercialization and Intellectual Alignment
38:02 Future of Mathematics with AI
44:30 Rock and Roll Influence
49:23 Final Thoughts