---
session_id: "1352467"
title: "What Will You Build When AI Is 10x Faster?"
phase: live
last_checked: 2026-11-12T13:24:21-08:00
render: 20261112-132421-c4d6ee
---

# Links: What Will You Build When AI Is 10x Faster?

## Speaker

**Natalie Serrino**, Co-founder, Gimlet Labs

- X: [@nserrino](https://x.com/nserrino) ("you can find me on X at nserino")

## Company

**[Gimlet Labs](https://gimletlabs.ai/)**: "Gimlet Labs is an applied research lab
focused on building high performance AI systems." (from the site)

## Resources

- **Product** · [Token Mint — Gimlet Labs](https://gimletlabs.ai/token-mint-game).
  The token-factory game from the talk: fast tokens versus many tokens, and the
  upgrades that push the Pareto frontier up and to the right. Laptop only.
- **Blog** · [The emerging role of SRAM-centric chips in AI inference](https://gimletlabs.ai/blog/sram-centric-chips)
  (Gimlet, 2026-03-05, by Natalie Serrino). The accelerator class the talk pairs
  with GPUs (the transcript's "asteroid-centric accelerator").
- **Blog** · [Splitting LLM inference across different hardware platforms](https://gimletlabs.ai/blog/multivendor-prefill-decode-disaggregation)
  (Gimlet, 2025-10-13, by Natalie Serrino). How a model is split into pieces that
  run on different chip types, the approach behind the Snake demo.
- **Blog** · [Recursive Self-Optimization at Gimlet](https://gimletlabs.ai/blog/recursive-self-optimization)
  (Gimlet, 2026-11-03). The telemetry-to-model self-improvement loop the talk
  names ("we're building recursive self-optimization for this").

## Gaps

- LinkedIn (`linkedin.com/in/natalieserrino`, from the roster) answered 999 with
  no page title naming her, so it is not corroborated and is left out.
- The roster's `gimletlabs.ai/blog` is the company blog, not a personal page; its
  posts are listed above instead.
- No official recording or slides found yet. The dossier rechecks every 6 hours
  during the event.