The Powerful Alternative To Fine-Tuning
Categories: VC, Startup, Design
Summary
Poetic has developed a recursively self-improving AI system that outperforms frontier models at a fraction of the cost, eliminating the need for expensive fine-tuning and making AI capabilities accessible to startups without massive computational budgets.
Key Takeaways
- Poetic achieved 54% accuracy on ARC AGI v2 at $32 per problem, beating Gemini 3 Deep Think's 45% accuracy which cost $70+ per problem—demonstrating 2x cost efficiency with better results.
- Optimizing for Humanity's Last Exam cost less than $100k compared to hundreds of millions spent on frontier model training, making state-of-the-art results accessible to seven-person teams.
- AI harnesses automatically adapt to new frontier models without requiring retraining or fine-tuning investments, protecting startups from obsolescence when better models are released.
- Fine-tuning becomes economically obsolete for most startups since new models continuously outpace custom fine-tuned versions, but recursive self-improvement systems maintain competitive advantage.
- Builders should experiment daily with AI tools rather than over-engineer solutions, as capabilities improve so rapidly that iterative testing beats lengthy planning.
Related topics
Transcript Excerpt
The world is changing so quickly. This is probably a little bit obvious, but you should just try things and and like every day do something with AI. Last summer, I took a weekend and used um GPT5 to help me build an iPhone app. I hadn't done that in a decade. And yeah, it's so fast and so easy. And that was, you know, an age ago. That was like 8 months ago. Uh now it's even faster and easier. Don't limit yourself. like anything that you imagine, you should just try to use AI and see how far you can get with it and you'll be, you know, making the world better. Welcome to another episode of the Light Cone. Ian Fischer is the co-founder and co-CEO of Poetic, which is building recursively self-improving AI reasoning harnesses for LLMs. Previously, he spent a decade as a researcher at Google De…