Every serious tool for building, running, and shipping AI agents — frameworks, memory, evals, browser infra, voice, sandboxes, and more. Hand-reviewed, no filler.
Open-source multi-agent framework, community continuation of the original AutoGen research project.
Observability for AI agents with session replays, cost tracking, and integrations for 400+ frameworks and LLMs.
Vercel's unified TypeScript SDK for AI apps and agents with streaming and multi-provider model support.
Hosted, human-like Chromium instances that let agents handle logins, forms, and sites without APIs.
Marketplace of thousands of ready-made scraping and automation Actors, callable by agents via MCP.
MCP runtime handling agent authorization, prebuilt tools, and governance for business systems.
Open-source (ELv2) tracing, evaluation, and experimentation platform for agent development from Arize.
Enterprise phone-call automation with AI voice agents and bundled per-minute pricing for the full stack.
Eval and observability platform for AI products: trace production behavior and run quality experiments.
Open-source Python library plus hosted agents and browser infrastructure for AI-driven web automation.
Managed headless browser infrastructure with APIs for agents to navigate, log in, and extract web data.
Anthropic's Python/TypeScript library exposing the Claude Code agent loop, tools, and context management.
Open-source agent memory that turns documents and context into a queryable knowledge graph.
Tool and auth layer giving agents managed OAuth and executable actions across 1,000+ apps.
Evaluation and observability platform built on DeepEval, the open-source LLM testing framework.
Python framework and enterprise platform for building crews of role-based AI agents that work together.
Open-source infrastructure for executing AI-generated code, with sub-90ms sandbox creation and Git support.
Open-source Firecracker-based sandboxes for safely running LLM-generated code, with sub-200ms startup.
AI voice platform with TTS, STT, and cloning, plus conversational agents across phone, chat, and WhatsApp.
Web search and contents API built for AI agents, backed by an index of 100B+ documents.
Open-source web scraping and crawling API that turns websites into clean, LLM-ready data.
Public cloud with fast-booting VMs; its Sprites product runs persistent Linux machines for agent workloads.
AI evaluation and observability platform that turns offline evals into production guardrails.
Google's open-source kit for building, evaluating, and deploying agents in Python, TypeScript, Go, and Java.
Open-source orchestration framework from deepset for production agents and RAG pipelines.
Open-source AI gateway and observability layer to route, log, and analyze LLM requests.
Cloud browser platform positioning itself as web infrastructure for AI agents, with scraping and session APIs.
Search foundation APIs: URL-to-markdown Reader, web search, embeddings, and rerankers for RAG and agents.
Open-source LLM engineering platform for tracing, prompt management, and evaluations; self-host or cloud.
MIT-licensed orchestration library for building stateful, controllable agents as graphs of steps.
LangChain's platform for tracing, evaluating, and monitoring agents, usable with any framework.
Platform for stateful agents with self-editing long-term memory, built by the creators of MemGPT.
Open-source realtime media stack and Agents framework, with a managed cloud for voice AI in production.
Data framework for agents and RAG over documents, with LlamaCloud for hosted parsing and indexing.
TypeScript agent framework with built-in workflows, memory, and observability; Apache-2.0 with a cloud tier.
Memory layer that lets agents retain and recall user context across sessions; open source plus hosted API.
Serverless cloud for AI workloads: inference, training, and ephemeral sandboxes for untrusted agent code.
Open-source integration platform connecting products and agents to 900+ APIs, with MCP tool exposure.
Lightweight Python library from OpenAI for agents with tools, handoffs, and guardrails; few abstractions.
Comet's open-source platform that logs every agent step and automates eval workflows to catch errors.
Open-source framework from Daily for building voice and multimodal conversational AI pipelines.
Open-source CLI for testing and red-teaming LLM apps, with a commercial security platform on top.
Type-safe Python agent framework from the Pydantic team; swap models by changing a string.
Platform for contact-center voice agents that book appointments, qualify leads, and hand off to humans.
Devboxes: stateful micro-VM environments plus benchmarking and observability for AI coding agents.
Microsoft's open-source SDK for building AI agents and plugins in C#, Python, and Java.
Low-cost Google SERP API returning web, news, images, maps, and scholar results as JSON in 1-2 seconds.
Open-source browser API for running fleets of cloud Chromium sessions behind agent workloads.
Stripe's official SDKs, MCP server, and agent skills for adding payments and billing to AI agents.
Memory API for agents with user profiles, vector search, and document ingestion across formats.
Real-time web access API for agents: search, extract, crawl, and research endpoints in one service.
Developer platform for building, deploying, and monitoring phone and web voice agents at scale.
Weights & Biases toolkit for tracing, evaluating, and monitoring LLM apps and production agents.
MCP endpoint that lets AI assistants execute real actions across Zapier's 9,000+ connected apps.
Agent memory service built on temporal knowledge graphs for fast retrieval of conversation and business data.