P
Overview
Polymath AI Labs builds simulated worlds to train AI agents on real-world, long-horizon tasks. Its simulations combine running applications, real tools, and multi-step workflows to teach end-to-end software development and other complex tasks. A flagship product, Horizon-SWE, evaluates an agent’s ability to plan, code, test, and deploy inside a simulated software company. The goal is to create reliable, autonomous AI systems that can operate with minimal human supervision in real-world settings.
Work here?Claim your company
Simplify's Take
What believers are saying
- YC listed Polymath active on August 26, 2026, with six employees.
- Polymath updated its site on September 4, 2026, naming Base10, Cervin, and YC.
- Horizon-SWE launched February 6, 2026 and benchmarked frontier models credibly.
What critics are saying
- No public revenue, customer logos, or contracts prove demand beyond frontier labs.
- OpenAI, Anthropic, and Google can build internal simulators faster than a six-person startup.
- If benchmark quality stalls, Polymath becomes a replaceable services vendor by 2027.
What makes Polymath unique
- Horizon-SWE uses a containerized software company with 73 tools and live traffic.
- It scores agents across feature correctness, DevOps, and engineering quality, not toy tasks.
- YC Winter 2026 founders bring Hume, Plaid, AWS, Amazon, and Berkeley pedigrees.
Help us improve and share your feedback! Did you find this helpful?
Funding
Notable Investors:
Seed funding is usually the first official round after pre-seed, when a startup has a prototype or concept. It’s typically used to develop the product, test the market, and start building the team. Investors here are often angel investors or early-stage venture capitalists.
Seed Funding Comparison
Industry standards
Ind Avg. $3.3M
$130k
$500k
$1.5M
$2M
$2.3M
$3M