Guide Labs

ActiveAI-native

Frontier models and agents you can learn from, audit, and steer.

What it does

Guide Labs builds frontier AI models and agents that you can learn from, audit, and steer. Our first model, Steerling-8B, is the first and largest inherently interpretable model: it can trace any output to training data, concepts inside the model, and the input, all without sacrificing performance.

Its site says now · captured 2026-08-23

Guide Labs

We are building a new class of interpretable AI systems and foundation models that humans can reliably debug, trust, and understand.

Next to it in AI infra and compute · Interpretable frontier models

  • LinumY Combinator W23

    type text, dream video

  • Paper InstrumentsY Combinator X25

    Open-source frontier models for knowledge work

  • Cactus ComputeY Combinator S25

    Tiny Edge AI For Tiny Devices

  • Stellon LabsY Combinator S25

    Building tiny frontier AI models that run on edge devices

  • Uplift AIY Combinator S25

    Foundational Voice Models for regional languages

  • Kalpa LabsY Combinator F25

    Scaling Generalist Speech models

  • Miso LabsY Combinator X26

    Emotive foundation voice models

  • FastinoPlug and Play PnP 2026

    Fastino provides task-specific language models for efficient and accurate AI development.