I'm Sam, CEO and co-founder of Coherence Labs. For five years I was a software engineer at Google, building full-stack features for Google Web Search, training and deploying LLM-based systems at scale, and closing the data gaps between teams as LLM systems rolled out across the company.
Before Google, I spent three years at Caliber Communications building out their entire machine learning system: an internal data-labeling program, a CNN training pipeline, and an inference system that supported real-time object detection across hundreds of video streams. I did this while studying Computer Science at McMaster University.
In 2026, I left Google to found Coherence Labs with a mission to bring the benefits of intelligence technology to everyone. Our first product, Axiom Reason, used LLMs to put formal logic in anyone's hands. Today I work directly with business owners to adopt intelligence technology with confidence: making their processes, goals, and guardrails explicit, then delivering tools that save them time, make them money, or improve their product.
Writing by Sam
Complexity Made Coherent
You bring your complexity, we provide coherence. Our new strategy is to help businesses easily and confidently adopt intelligence technology, with value they can measure and a pace that works for them. Axiom Reason will be continuing on as a free and open-source toolkit.
Towards Practical Philosophy
Words like "existence" and "real" have stalled philosophical progress. A pragmatic alternative offers a way forward: grounding our concepts in their consequences for action. By adopting the word "actual" to describe anything that affects how we act, we can sidestep intractable metaphysical debates and begin formalizing the domains that matter most.
Reality Can Be Modeled
Despite the modern abundance of data, communicating meaning remains a fundamental challenge. Addressing this paradox requires accepting a foundational proposition: reality can be modeled. Because language itself uses abstract symbols that map to reality, the mere act of communication assumes this premise. Consequently, asymptotically solving human miscommunication depends on improving the models backing our language rather than simply transmitting more data.
Bring us your complexity
We're taking on our first clients now. Tell us about your business and where you're stuck.
Get in touch