Engineering Office Intelligence

Anygraph is an applied research lab working on automating digital paperwork. Our approach is to build the knowledge worker of the future: a human-AI employee that is an order of magnitude more productive than any one human or AI.

The current human-AI paradigm is broken. From talking to hundreds of businesses, we have observed the same core problem. This is workslop: AI-generated content that masquerades as good work but lacks the substance to meaningfully advance a task. Someone still has to interpret, correct, and redo.

How can we teach AI to produce meaningful, high-quality work? How can we verify what has been generated? These questions demand a rethinking of how humans and AI work together in professional settings.

By bringing the coding agent revolution to the conventional office, we envision a world without workslop. Our product, Steve, is an AI agent with the skills to get details right, the judgment to know when something is wrong, and the taste to know the difference between draft and done.

We are researchers and engineers building at the edge of what is useful and what is possible. We have studied at Harvard and Cambridge, and our work spans NeurIPS publications, original results in Hodge theory and high-dimensional statistics, and AI systems deployed in healthcare settings where output quality means the difference between life and death.

If any of this excites you, we would love for you to join us at the frontier.

— Zad, Jarell, Rushil, Bryan, and the Anygraph team.