This week / Lists / Agent infrastructure
Updated November 13, 2026
Agent infrastructure startups to watch, week of November 9, 2026
The picks and shovels: sandboxes and cloud computers for agents, browsers, memory, tool access, search, orchestration frameworks and evaluation.
This week's top 3: Firecrawl (score 86, headcount +22% in 3 months); Arcade (score 75, headcount +10% in 3 months); Modal (score 74, headcount +54.4% in 3 months).
| # | Company | Momentum | Headcount | Headcount 3m | Web visits 3m | Last round |
|---|---|---|---|---|---|---|
| 1 | Firecrawl API that turns websites into clean, LLM-ready data for agents. | 86 | 50 | +22% | +80.4% | Series B, $75M (Sep 2026) |
| 2 | Arcade Tool-calling platform that handles authorization so agents can act in user accounts. | 75 | 22 | +10% | +18.3% | Series A, $60M (Jun 2026) |
| 3 | Modal Serverless compute for AI workloads, including sandboxes used to run agent code. | 74 | 193 | +54.4% | +22.9% | Series D, $750M (Oct 2026) |
| 4 | Steel Open-source browser API for AI agents and web automation. | 72 | n/a | n/a | +47.8% | n/a |
| 5 | Exa Search engine and API built for AI agents and LLM applications. | 67 | 184 | +49.6% | -44.6% | Series C, $250M (May 2026) |
| 6 | Cua Open-source infrastructure for computer-use agents, with sandboxed macOS, Linux and Windows environments. | 66 | 8 | 0% | n/a | Seed (Jan 2026) |
| 7 | Browserbase Headless browser infrastructure for AI agents, plus the Stagehand browser automation framework. | 65 | 83 | +45.6% | +58.8% | Series B, $40M (Apr 2025) |
| 8 | Composio Tool and integration layer that gives agents authenticated access to hundreds of apps. | 65 | 4 | +100% | +193% | Series A, $29M (Apr 2025) |
| 9 | LangChain Open-source agent frameworks (LangChain, LangGraph) and the LangSmith observability platform. | 64 | 385 | +57.1% | +9.8% | Series B, $125M (Oct 2025) |
| 10 | Daytona Fast, stateful sandboxes and composable computers for running AI-generated code and agents. | 60 | 69 | +7.8% | +21.1% | Series B, $60M (Oct 2026) |
| 11 | Runloop Devboxes and benchmarking infrastructure for AI coding agents. | 57 | 17 | +6.3% | +86.5% | Accelerator/Inc (Oct 2025) |
| 12 | E2B Open-source cloud sandboxes where AI agents run code and use tools securely. | 55 | 46 | +24.3% | +25.5% | Series A, $21M (Jul 2025) |
| 13 | Braintrust Evaluation and observability platform for teams building AI products and agents. | 52 | 177 | +36.2% | +13.5% | Series B, $80M (Feb 2026) |
| 14 | Anchor Browser Cloud browsers for agents that need to log in and act on websites reliably. | 49 | 15 | 0% | n/a | Seed, $6M (Oct 2025) |
| 15 | Hyperbrowser Cloud browser infrastructure for running web agents and scraping at scale. | 45 | n/a | n/a | -20.7% | n/a |
| 16 | Zep Context and memory infrastructure for agents built on a temporal knowledge graph. | 40 | 5 | -28.6% | +24.6% | Pre Seed (Mar 2024) |
| 17 | Mem0 Memory layer for AI agents and assistants, available open source and as a managed service. | 35 | 21 | +5% | -15.5% | Series A, $24M (Oct 2025) |
| 18 | Letta Platform for stateful agents with long-term memory, from the creators of MemGPT. | 28 | 17 | 0% | -44% | Seed, $10M (Sep 2024) |
| – | AgentOps Observability and monitoring for AI agents in development and production. | n/a | n/a | n/a | n/a | n/a |
| – | Orgo Virtual desktops for computer-use agents, each with its own isolated machine. | n/a | n/a | n/a | n/a | n/a |
Scores compare each company with all 89 tracked companies, not only this list. Read the methodology, see past weeks, or open a company for every number and its source.