Applied Compute, a startup founded by former OpenAI researchers, has announced it secured $80 million in funding to advance its vision of "Specific Intelligence" for enterprise clients.

The funding round, backed by prominent investors including Benchmark, Sequoia, Lux, Elad Gil, Victor Lazarte, and Omri Casspi, positions the company with a valuation close to $700 million, according to industry sources. This investment aims to scale deployments and expand the team developing in-house agent workforces. Applied Compute's core offering, termed "Specific Intelligence," focuses on developing highly specialized AI agents trained on a company's proprietary data. This approach aims to unlock "latent knowledge" within an organization, creating custom models that power an in-house agent workforce. The company asserts that while general AI models are useful, true competitive advantage in AI comes from specialized agents deeply integrated into specific company operations. The startup is already collaborating with sophisticated, AI-forward companies such as Cognition, DoorDash, and Mercor AI. These partnerships highlight a growing demand for tailored AI solutions that move beyond generic models, allowing clients to achieve state-of-the-art performance on customer evaluations and build and validate models in days rather than months. Applied Compute emphasizes that its proprietary agents offer an edge by not relying on the public release schedules of major AI labs. The founding team comprises Yash Patil, a key contributor to OpenAI’s agentic software engineer effort (Codex); Rhythm Garg, a core contributor to OpenAI's first RL-trained reasoning model (o1); and Linden Li, who focused on ML systems and infrastructure for RL training at OpenAI. Two-thirds of the team are former founders, bringing extensive technical backgrounds, from top AI researchers to Math Olympiad winners. This high-density, low-latency team is leveraging its deep expertise to build the next generation of AI agent workforces.