Scaled Cognition

Scaled Cognition is an artificial intelligence model lab building what it calls Super Reliable Intelligence for the enterprise. Founded by chief executive Dan Roth, a former Microsoft corporate vice president of conversational AI, and chief technology officer Dan Klein, a University of California Berkeley professor, who together built and sold one of the first agentic AI companies to Microsoft, the company raised a hundred million dollar Series A led by Khosla Ventures with Genesys in 2026 at a valuation around seven hundred fifty million dollars. Its pitch is that reliability, not raw capability, is the real barrier to putting AI to work.

The flagship product is a model, APT, short for Agentic Pretrained Transformer. The company says it delivers the conversational quality of leading models while eliminating hallucinations and guaranteeing policy adherent performance, and that it is smaller, faster, cheaper, and more accurate than frontier models. Reliability is engineered into the architecture rather than bolted on afterward, which is what lets APT power agents that take real action where the stakes are high, on bank balances, medical records, flight changes, and insurance claims. Crucially, APT runs in a virtual private cloud or fully self hosted, so an enterprise owns its AI outright rather than renting it from a third party model provider.

Beyond the model, Scaled Cognition sells a full platform for putting agents into production. It includes agentic tooling to build agents, simulation and evaluation frameworks to test them before they go live, and live monitoring to watch them once they do, giving enterprises what they need to build, test, and operate reliable AI at scale. Customer experience is the first application. Genesys, a cloud contact center leader serving more than eight thousand organizations across a hundred countries, already uses APT for agentic virtual agents inside its platform, and companies using the model are projected to automate more than a billion customer service interactions within a year.

The larger ambition is the six hundred billion dollar business process outsourcing market, the sprawl of outsourced customer service, information technology support, human resources, and finance. The thesis is a reversal of a long trend, that enterprises which once shipped this work to third party providers now want to insource it as artificial intelligence workforces they own and control, and Scaled Cognition wants to supply the engine. The bet is sharp and unproven at scale, since the promise of a model that will not give a wrong answer runs against the probabilistic nature of the technology, and the company is wagering that its architecture backs the claim.