Hanno Labs: Causal AI Lab Behind Bosun (2026)
Hanno Labs (est. 2025) builds frontier AI for causal modeling and knowledge systems. Makers of Bosun, the open-source Apache 2.0 judge model for agent memory graphs.
Founded: 2025 · HQ: United States · Team: 1-10 · CEO: Julie Saltman
About Hanno Labs
Hanno Labs is an AI research lab founded in 2025 by Julie Saltman and Stephen Solka, building at the intersection of causal modeling and complex knowledge systems. The lab's stated mission is to apply frontier AI to the problems that still require human judgment: domains where existing LLM tools fall short because they lack the ability to reason about causality, evidence strength, or conflicting facts over time. Julie Saltman, CEO and co-founder, brings 11 years of regulatory law experience at the US Department of Justice, is a published author, and teaches at Georgetown Law as an adjunct professor. She holds degrees from Yale and Michigan Law. Stephen Solka, CTO and co-founder, has 15 or more years building AI and machine learning systems at fintech startups and served as VP Engineering at AngelList. The team is advised by former Department of Defense scientists who specialized in predictive modeling and technological surprise analysis. The lab ships products across two tracks: enterprise tools for regulated industries and open-source models for the developer community. On the enterprise side, Hanno Labs operates StanddAI for due diligence in private market investing, Hanno for predictive regulatory intelligence, Hey Lefty for autonomous research monitoring agents, and Exoclaw as an enterprise autonomous agent framework. On the open-source side, the lab released Bosun in June 2026, a programmable judge model for curating agent knowledge graphs. Bosun represents the lab's clearest published technical statement: agents managing long-lived memory need a fast, calibrated, instruction-following model to decide edge by edge which connections in a knowledge graph are supported, non-redundant, and still factually current. The Bosun release includes WarrantBench, an evaluation benchmark measuring steerability for arbitrary judgment rules, a benchmark class largely absent from standard LLM leaderboards. Funding details are not publicly disclosed. The lab operates as a private company at an early stage, shipping quickly across both enterprise and open-source channels.
Mission
Apply frontier AI to the problems that still require human judgment
Products
- Bosun: https://huggingface.co/Hanno-Labs/bosun-4b
- StanddAI: https://standd.io
- Hanno: https://hannolabs.ai
- Hey Lefty: https://heylefty.com
- Exoclaw: https://hannolabs.ai