OpenCurious Directory · Updated

AI Agent Infrastructure Companies in 2026

AI agent infrastructure companies make the developer tools and runtime services used to build and run AI agents. The category covers agent frameworks and orchestration, persistent memory layers, cloud browsers and computer-use infrastructure, secure code-execution sandboxes, and tool-calling and integration platforms. These companies sell to the developers who build agents, not to the end users of finished agent products.

  1. LangChain
  2. CrewAI
  3. Browserbase
  4. E2B
  5. LlamaIndex
  6. Mem0
  7. Daytona
  8. Composio
  9. Browser Use
  10. Letta

AI Agent Infrastructure Companies to know

We picked these from companies whose main product is a building block for agent developers, such as a framework, memory, browser, sandbox or tool-integration layer. We selected using venture funding reported in independent press, open-source adoption and enterprise usage as of October 2026. We left out companies that ship finished agents (AI Agent Companies), coding agents (AI Coding), tracing and evaluation tools (AI Observability) and general MLOps or vector databases (AI Infrastructure). Every company on the list was active and independent when we checked.

  1. LangChain logo

    LangChain

    Open-source agent frameworks (LangChain, LangGraph) plus the LangSmith platform

    LangChain builds the open-source LangChain and LangGraph frameworks for building LLM applications and long-running agents. It also sells LangSmith, a paid platform for developing, testing, deploying and monitoring agents. The project began in October 2022 and became a company in early 2023.

    Headquarters
    San Francisco, California, USA
  2. CrewAI logo

    CrewAI

    Open-source framework and enterprise platform for multi-agent systems

    CrewAI makes an open-source Python framework, released under the MIT license in December 2023, for building agents and multi-agent systems. It also sells an enterprise platform for building and running agents with access controls, analytics and workflow templates. The company states that it runs more than 450 million agentic workflows a month.

    Headquarters
    San Francisco, California, USA
  3. Browserbase logo

    Browserbase

    Cloud-hosted headless browsers for AI agents and web automation

    Browserbase runs headless browsers in the cloud so AI agents can navigate and act on websites. It also maintains Stagehand, an open-source SDK for browser automation, and offers Search and Fetch APIs for agents. The founder has said the company does not train models and sells infrastructure primitives that agents use.

    Headquarters
    San Francisco, California, USA
  4. E2B logo

    E2B

    Open-source cloud sandboxes for running AI-generated code and agents

    E2B offers secure, open-source cloud sandboxes, built on Linux microVMs, where AI agents can run code and carry out multi-step tasks in isolated machines. It provides Python and JavaScript SDKs and a desktop sandbox for agents that need a graphical interface. The company was founded in the Czech Republic.

    Headquarters
    San Francisco, California, USA
  5. LlamaIndex logo

    LlamaIndex

    Open-source framework and document-parsing platform for agents working over data

    LlamaIndex maintains an open-source Python and TypeScript framework for building LLM agents and applications that work over enterprise data. It also sells LlamaParse and LlamaCloud, managed services for parsing, extracting and indexing documents. The project started as a side project in late 2022 and was incorporated in April 2023.

    Headquarters
    San Francisco, California, USA
  6. Mem0 logo

    Mem0

    Memory layer that gives AI agents persistent memory

    Mem0 offers memory infrastructure for AI agents. It extracts information from interactions, resolves conflicting facts and retrieves relevant memories. Developers can use it as an open-source library or a hosted commercial platform.

    Headquarters
    San Francisco, California, USA
  7. Daytona logo

    Daytona

    Elastic sandboxes and virtual computers for running AI-generated code

    Daytona runs secure sandboxes where AI agents can execute untrusted code, with configurable compute and support for snapshots and forking. It also offers Linux, Windows and macOS virtual desktops for computer-use agents and reinforcement learning. The founders previously built Codeanywhere.

    Headquarters
    New York, New York, USA
  8. Composio logo

    Composio

    Tool-calling and integration platform that connects agents to external apps

    Composio gives developers a platform for connecting AI agents to external tools and applications through managed integrations, authentication flows and tool formatting. The company describes its focus as infrastructure that lets agents learn from shared experience.

    Headquarters
    San Francisco, California, USA
  9. Browser Use logo

    Browser Use

    Open-source library and cloud service that let AI agents operate websites

    Browser Use converts websites into a text-like form that AI agents can read and act on, rather than relying only on vision. The MIT-licensed library is open source, and the company also sells hosted browser infrastructure and web-automation agents. The founders met at ETH Zurich and joined Y Combinator's Winter 2025 batch.

    Headquarters
    San Francisco, California, USA
  10. Letta logo

    Letta

    Platform for stateful agents with long-term memory, from the MemGPT team

    Letta builds developer tools and a hosted service for stateful AI agents that keep long-term memory. It grew out of MemGPT, an open-source project from UC Berkeley's Sky Computing Lab, and its advisors include Ion Stoica and Joey Gonzalez.

    Headquarters
    San Francisco, California, USA

AI Agent Infrastructure Companies at a glance

AI Agent Infrastructure Companies: headquarters and focus
CompanyHeadquartersFocus
LangChainSan Francisco, California, USAagent framework, orchestration
CrewAISan Francisco, California, USAagent framework, multi-agent
BrowserbaseSan Francisco, California, USAbrowser infrastructure, computer use
E2BSan Francisco, California, USAsandboxes, code execution
LlamaIndexSan Francisco, California, USAagent framework, document parsing
Mem0San Francisco, California, USAagent memory, open source
DaytonaNew York, New York, USAsandboxes, code execution
ComposioSan Francisco, California, USAtool calling, integrations
Browser UseSan Francisco, California, USAbrowser automation, computer use
LettaSan Francisco, California, USAagent memory, stateful agents

Frequently asked questions

What is an AI agent infrastructure company?

It is a company that makes the developer tools and runtime services used to build and run AI agents. Examples include agent frameworks (LangChain, CrewAI, LlamaIndex), memory layers (Mem0, Letta), cloud browsers (Browserbase, Browser Use), code-execution sandboxes (E2B, Daytona) and tool-integration platforms (Composio).

Which AI agent infrastructure company has raised the most recently reported funding?

Of the companies listed, LangChain has the largest reported recent round. It raised a $125M Series B led by IVP at a $1.25B valuation in October 2025. Browserbase raised a $40M Series B led by Notable Capital at a $300M valuation in June 2025.

Which companies provide sandboxes for running agent code?

E2B and Daytona. E2B offers open-source microVM sandboxes and raised a $21M Series A led by Insight Partners in July 2025. Daytona offers sandboxes plus virtual desktops for computer-use agents and raised a $24M Series A led by FirstMark Capital in February 2026.

How do Browserbase and Browser Use differ?

Browserbase mainly sells cloud-hosted headless browsers, plus the open-source Stagehand SDK and Search and Fetch APIs. Browser Use began as an open-source library that turns websites into a text-like form agents can act on, and now also sells hosted browsers. Browserbase raised a $40M Series B led by Notable Capital; Browser Use raised a $17M seed led by Felicis.

Which companies focus on memory for AI agents?

Mem0 and Letta. Mem0 sells a memory layer that extracts, reconciles and retrieves memories, and raised $24M across its seed and Series A in October 2025. Letta, from the MemGPT team at UC Berkeley, builds stateful agents with long-term memory and raised a $10M seed led by Felicis in September 2024.

How is agent infrastructure different from AI agent companies?

Agent infrastructure companies sell building blocks to developers, such as frameworks, memory, browsers, sandboxes and tool integrations. AI agent companies sell finished agents, like support or sales agents, to the businesses and consumers who use them.

All AI company directories →

About this directory: OpenCurious curates this list for researchers, founders, operators and buyers. Company descriptions and headquarters are based on publicly reported information as of and change often; figures a company reports about itself are its own claims. This is an editorial selection, not a paid ranking: no company paid to be listed, and companies appear in no particular order. Company names and logos are trademarks of their respective owners and are used only to identify each company. See our editorial policy. To suggest a company or a correction, email hello@opencurious.com.

The OpenCurious Newsletter

Stay curious. Get the best of the open web.

New directories, researcher tools, and open-source finds — straight to your inbox. No spam, ever.

  • New directories
  • Researcher tools
  • Open-source finds

Occasional emails from OpenCurious. Unsubscribe anytime. Privacy Policy