Y Combinator Launch · Mar 2026
YC launch describing Polymath’s simulated worlds for long-horizon agents, including running apps, real tools, and evolving tasks that mirror real work.
Company research
Public company, workplace, funding, and market signals
Updated Jul 30, 2026
Overview
Polymath is a San Francisco-based applied research/data lab building simulation environments and benchmarks for training and evaluating long-horizon autonomous AI agents, with an emphasis on reliability, autonomy, and frontier-model collaboration.
Primary product
Simulation environments and benchmarks for long-horizon AI agents, including Horizon-SWE
Founded
2026
Headquarters
San Francisco, California, United States
Team size
1-10
Industry
Software Development
Sub-industry
AI agent infrastructure / reinforcement learning environments
Offices
2261 Market Street, Suite 72022, San Francisco, CA 94114
Business model
B2B
enterprise software
data/benchmark licensing
research partnerships
Funding
People & investors
Work & culture
Small, research-heavy team focused on safe, highly capable AI; hiring engineers, researchers, and operators to build long-horizon agent infrastructure with frontier model labs.
Compensation
Public job postings list Member of Technical Staff roles at $150k–$300k plus equity and a Strategic Projects Lead at $150k–$250k plus equity; the AI Research Resident role is remote contract and does not list pay.
Benefits
Health insurance
Dental insurance
Vision insurance
Free meals
Bay Club membership
Commuter benefits
401(k)
Unlimited PTO
Top-tier equipment
Product & market
Differentiators
- —Simulation environments that mimic real, evolving work systems
- —Focus on long-horizon, multi-step agent reliability rather than single-task coding
- —Horizon-SWE benchmark for end-to-end software engineering workflows
- —Real tools, running applications, and live-traffic production-grade scenarios
- —Positioned as environment infrastructure for frontier labs
Technology
Reinforcement learning
Simulation environments
Containerized environments
OpenHands agent harness
Production-grade software environments
Real tools and services
Multi-step agent evaluation
Verifiers
Customers
Frontier AI/model labs
Leading model labs
AI labs that need simulation environments for long-horizon agents
Competitors
OpenMind
Ineffable Intelligence
AgileRL
Scale & contact
Latest news
Forbes · Mar 2026
Forbes highlighted Polymath among the YC batch, framing it as infrastructure for AI agents through simulated environments where agents rehearse real-world workflows before production.
Polymath blog · Feb 2026
Introduced Horizon-SWE, a benchmark for long-horizon, multi-tool software engineering tasks in production-grade environments, measuring autonomous-agent performance across planning, coding, testing, deployment, and monitoring.
Polymath blog
Polymath said it raised an $8M seed round to build the environment layer for training and evaluating autonomous agents, with support from Base10, Cervin Ventures, SurgePoint Capital, Founders Future, Berkeley Frontier Fund, Y Combinator, and other angels.
Jobs and careers at Polymath
0 jobs at Polymath
No open positions
Check back later for new openings