Cognition

Cognition's Tavrn customer case reports Devin embedded across engineering and operations after six weeks and handling four out of five questions previously routed to engineers through scoped, logged access. This named operational outcome is independent of annualized company revenue.

Cognition reported crossing a $1B annualized revenue run rate by September 2026 after rapid enterprise adoption of Devin and Windsurf across major global companies.

Cognition reported crossing a $1B annualized revenue run rate by September 2026 after rapid enterprise adoption of Devin and Windsurf across major global companies.

Cognition's October 2025 SWE-1.5 technical report describes coding RL environments combining correctness tests, code-quality rubrics and browser-agent end-to-end grading, with human experts attempting to circumvent graders through reward hardening.

Cognition develops Devin, cloud coding agents, multi-agent orchestration, specialized software-engineering models such as SWE-1.6 and SWE-check, and an integrated agentic IDE platform through Windsurf.

Cognition develops Devin, cloud coding agents, multi-agent orchestration, specialized software-engineering models such as SWE-1.6 and SWE-check, and an integrated agentic IDE platform through Windsurf.

Cognition reported crossing $1 billion annualized revenue run rate on September 25, 2026. This commercial-scale metric is independent of existing September financing and strategic-relationship evidence.

Cognition raised over $2B at a $48B valuation in September 2026 and has strategic enterprise relationships with customers and partners including Cognizant, Citi, Mercedes-Benz, Goldman Sachs, AWS and others.

Cognition raised over $2B at a $48B valuation in September 2026 and has strategic enterprise relationships with customers and partners including Cognizant, Citi, Mercedes-Benz, Goldman Sachs, AWS and others.

Cognition documents end-to-end reinforcement learning in software task environments, joint iteration on the Cascade agent harness and model, and grading using tests, code-quality rubrics and browser-based end-to-end checks for SWE-1.5.

Cognition combines autonomous agents, coding models, IDEs, cloud execution, testing, enterprise integrations and large-scale software engineering deployment, making its technical skills exceptionally portable across AI developer-tooling employers.

Cognition combines autonomous agents, coding models, IDEs, cloud execution, testing, enterprise integrations and large-scale software engineering deployment, making its technical skills exceptionally portable across AI developer-tooling employers.