The Daily AI Show Spotify Engineers Stopped Writing Code
33 snips
Feb 13, 2026 Spotify engineers reportedly stopped handwriting code, showing how AI is changing software workflows. Breakthroughs in reasoning models like Gemini 3.0 and proof‑verifying agents get attention. Ultra‑fast models such as Codex Spark and neuron‑powered biological computing are explored. Rapid growth of Chinese open models and big funding rounds highlight how capability and scale are compounding.
AI Snips
Chapters
Transcript
Episode notes
Spotify Engineers Stopped Handwriting Code
- Brian describes Spotify's top engineers not writing handwritten code since December as a real-world signal AI coding has mainstreamed.
- He recounts Spotify shipping 50+ features quickly because AI removed coding bottlenecks and enabled faster experimentation.
Frontier Models Threaten Expert Exclusivity
- Andy and Brian note frontier model improvements imply AI will soon perform large fractions of expert tasks, including advanced math.
- David Kipping's example shows top scientists worry models may reach ~90% of expert capability in their fields.
DeepThink Advances Mathematical Reasoning
- Andy highlights Gemini 3.0 Deep Think's leap to 85% on ARC-AGI-2 and 90% on IMO proof benchmarks as major reasoning progress.
- He points out Google built Aletheia to generate, verify, and revise proofs, signaling models now handle rigorous mathematical reasoning.