Applied Compute

Applied Compute Inc. develops custom artificial intelligence models and in-house agent workforces for enterprises, a concept the company calls "Specific Intelligence." Founded in 2025 by former OpenAI researchers Yash Patil, Rhythm Garg, and Linden Li, the startup aims to provide companies with a competitive edge by creating specialized AI agents trained on proprietary company data and fine-tuned for specific workflows. The founding team consists of recent Stanford graduates who worked on key projects at OpenAI; Patil was a key member of the Codex software engineer effort, Garg was a core contributor to the first reinforcement learning-trained reasoning model (o1), and Li focused on ML systems and infrastructure for reinforcement learning training.

The company's business model centers on moving beyond general-purpose AI, which operates on public data, to create proprietary AI capabilities that competitors cannot easily replicate. Applied Compute's platform is designed to unlock latent knowledge within an organization to build, validate, and deploy these specialized models and agents in days rather than months. This is achieved through a vertically integrated approach, with an in-house technology stack that includes its own training infrastructure, agent platform, and development tools, running on a cluster of several thousand GPUs. A defining part of its strategy is embedding its own engineers directly within customer teams to work alongside them.

Applied Compute emerged from stealth in October 2025, announcing it had raised $80 million in funding from investors including Benchmark, Sequoia, and Lux Capital, which followed an initial $20 million round in June 2025. The firm serves sophisticated technology companies seeking to embed AI deeply into their operations. Early customers include DoorDash, AI software engineer developer Cognition AI Inc., and AI training data provider Mercor Inc.

Keywords: Specific Intelligence, custom AI models, enterprise AI, in-house AI agents, reinforcement learning, proprietary AI, agentic systems, AI workforce, OpenAI alumni, custom model training, AI for business automation, AI implementation, GPU clusters, AI infrastructure, data-driven decision-making, business process automation, enterprise software, AI development, machine learning systems, workflow optimization