Polymath builds simulation environments to train and evaluate long-horizon AI agents. The company creates simulated worlds where AI agents learn to operate autonomously, combining running applications, real tools, and multi-step tasks that reflect real-world complexity. Its flagship product, Horizon-SWE, is a benchmark that tests frontier models on end-to-end software engineering tasks. Polymath is a Y Combinator Winter 2026 company based in San Francisco.
Polymath
ActiveSimulation environments to train & evaluate long-horizon AI agents
Total raised
$500K
1 round
Stage
Seed
Jan 2026
Team
1-10
since 2026
Pricing
—
Founded
2026
San Francisco, United States
Agent-ready
—
Simulation environments to train and evaluate long-horizon AI agents
Simulated worlds where agents learn to operate autonomously
Combines running applications, real tools, and multi-step tasks
Environments reflect real-world complexity
Horizon-SWE benchmark for end-to-end software engineering tasks
Tests frontier models on multi-step, realistic workloads
Y Combinator Winter 2026 company based in San Francisco
12/100
Early
MCP server
Public API
Webhooks
OAuth 2.0
SDKs
No public agent surfaces detected yet.
Jan 2026 Seed $500K ● Y Combinator
Capital network
$500K raised ·1 backer·10 network links
- Backers1
- Shared portfoliocompanies these backers also fund
- Extended networkfunds that co-invest alongside them
Resolve AI
AI SRE for complex production environments
AI AgentsAI Developer Tools
deepset
Haystack framework and deepset Cloud for enterprise LLM apps
AI AgentsAI Developer Tools
Wordware
Build AI agents by writing in plain English
AI AgentsAI Productivity
Wonderful
Multilingual enterprise AI agents for customer service
AI AgentsAI Productivity
Cursor
The AI code editor built for productive engineers.
AI CodingAI Agents
Oasis Security
Agentic access management to secure AI agents and non-human identities
AI AgentsAI for Cyber Defense
- What does Polymath build?
- Polymath builds simulation environments to train and evaluate long-horizon AI agents, combining running applications, real tools, and multi-step tasks that reflect real-world complexity.
- What is Horizon-SWE?
- Horizon-SWE is Polymath's flagship benchmark that tests frontier models on end-to-end software engineering tasks.
- Why focus on long-horizon agents?
- Long-horizon, multi-step tasks are difficult for AI agents, so Polymath builds realistic simulated worlds where agents can learn to operate autonomously and be evaluated rigorously.
- Who is Polymath for?
- It is aimed at AI researchers and teams developing autonomous agents who need realistic environments and benchmarks, and it is a Y Combinator Winter 2026 company based in San Francisco.
Discussion
Sign in to join the discussion.
Sign inExplore more around Polymath
Contextual paths to related AI startups, deals and rankings.