Applied Compute\'s Agent Workforce Targets Niche AI with $80M

A stealthy startup from ex-OpenAI researchers, Applied Compute, has emerged with $80 million in funding to argue that general-purpose AI is just the beginning. The company, backed by heavyweights like Benchmark and Sequoia, is betting that the next major leap in productivity will come from what it calls "Specific Intelligence."

Inside sources close to the deal shared with StartupHub.ai that the round's valuation is close to $700 million.

The core idea is that while models like GPT-4 are powerful generalists, true competitive advantage lies in specialist agents. Applied Compute plans to build these specialists by training custom models on a company’s unique, internal data, the "latent knowledge" locked away in documents, databases, and internal tools. The result is a bespoke Applied Compute agent workforce, a team of digital employees designed for specific tasks within a specific corporate environment.

From Generalists to Specialists

According to a company announcement, this approach moves beyond simply fine-tuning. Applied Compute deploys an "in-house agent workforce that reports to your team," effectively creating a new tier of AI labor that is deeply integrated into a company's operations. This allows businesses to build proprietary AI capabilities that don't depend on the public release schedule of major AI labs.

The startup is already working with sophisticated, AI-forward companies like DoorDash, Mercor, and, notably, Cognition, the creators of the AI software engineer Devin. This suggests that even the builders of advanced general agents see a need for more specialized, in-house AI.

The founding team is composed of Yash Patil who was a key member of OpenAI's agentic Codex effort, Rhythm Garg who was a core contributor to its first reasoning model, and Linden Li who worked on the ML infrastructure for reinforcement learning. With a team composed largely of former founders and top researchers, Applied Compute is making a high-stakes play for the next phase of enterprise AI: turning a company's own data into a private, hyper-competent digital workforce.