Blackstone and Norm AI CTOs: agent capability made our workflow code obsolete
AI Engineer · youtube · 2026-11-05
AI Engineer podcast conversation with Blackstone CTO John Stecher and Norm AI CTO Scott Worland on building tech that survives rapid model improvement.
Key points:
- Data strategy first: Stecher argues solid data strategy and straight-through processing should precede agents, and internal apps can now be discarded and rebuilt rather than maintained.
- ROI vs waiting: how to weigh shipping now against waiting for the next model.
- Evals exposed bad architecture: Norm AI's large hand-labeled eval set revealed that much of its carefully built workflow code was actually in the way once agent capabilities arrived — prompting a full platform rebuild.
- Evals in financial services: designing evals where accuracy matters, and which parts still need human review.
- Both invoke the bitter lesson: invest in infrastructure and data rather than betting on current model limitations.
Valuable for anyone designing enterprise AI platform architecture, eval systems, and build-vs-wait decisions.
More from coding & agent
- Dev builds 3D endless runner with Grok Bot, which even made its own trailer — chongdashu · 2026-11-05
- KAIST's TAC-CoT cuts agent compute 2-3x while boosting accuracy up to +5.9 points — DISLab · 2026-11-05
- asc CLI fully supports App Store Connect API 4.6 as all six PRs merge — rudrank · 2026-11-05
- ByteDance open-sources OpenViking: agent memory that cuts input tokens up to 91% — Promptmethus · 2026-11-05
- Using Codex to audit AGENTS.md and skills for duplicate checks slowing shipping — rudrank · 2026-11-05
- AI coding agent adds unrequested features and approval blockers, then grovels when called out — RealSharpNinja · 2026-11-05