What it takes to set intelligence free — Fabian Hedin, CTO, Lovable
Lovable nearly failed after building a sophisticated multi agent framework, Fabian Hedin says. At CODE Summit 2026, its CTO explains what changed: durable agent runtimes, Git inspired histories and forkable computers that support long running coding agents. Sandbox checkpointing, copy on write memory, a Rust rewrite of Vite and verification feedback loops connect reliable software generation to product and business outcomes. Speaker info: - https://x.com/FabianHedin - https://www.linkedin.com/in/fabian-hedin-2377b0144 - https://github.com/lovablelabs/oj Timestamps: 0:00 - Why Lovable's first agent framework failed 3:52 - Building products and business outcomes 7:48 - Removing the limits on agent runtime 8:51 - Inboxes and agent communication 10:06 - Git inspired histories and agent forks 11:10 - Separating prompts from task history 13:03 - Stateless workers and uninterrupted deployments 14:30 - Checkpointing computers and fast resume 16:16 - Forking sandboxes and copy on write memory 18:02 - Rewriting Vite in Rust 19:13 - Inner verification and feedback loops 20:07 - Product feedback and business metrics




Join the discussion
Have something to add?
Sign in to reply, follow this thread, and earn XP when your answer helps someone.