Cascade

Distilling Proprietary Intelligence

Y CombinatorWinter 2026ActiveB2BInfrastructure

What they do

Problem

Companies struggle to adapt AI models to their proprietary data and workflows, leading to slower, more expensive, and less accurate intelligence.

Solution

Cascade aligns models to a company’s proprietary data and workflows to produce higher-throughput intelligence that performs better, faster, and cheaper.

Who it's for

Organizations with proprietary data and workflows (typically enterprises) looking to tailor models to their specific use cases.

Founders

Adam AlSayyad

Founder

Adam is the Co-Founder and CEO of Cascade. Previously, he was a researcher at the Berkeley Artificial Intelligence Research (BAIR) Lab, where his work focused on graph reasoning models, and agentic safety under some of the world's leading ML and AI safety researchers. Adam studied Computer Science at UC Berkeley.

LinkedIn · X

Haluk Cem Demirhan

Founder

Haluk is the Co-Founder and CTO of Cascade. Previously, he built production monitoring infrastructure and scaled agent systems at companies like Netflix and Amazon. His research at BAIR Lab covered long-horizon memory optimization and failure mode taxonomies for AI agents. Haluk studied Computer Science and Mathematics at UC Berkeley.