Agent Infra

An open-source, serverless, and scalable search infrastructure for AI products supporting vector, full-text, regex, and metadata search.

Cogni is an MCP memory server designed for LLMs and AI agents, utilizing entity-graph spreading activation to enable deterministic, multi-hop, and cross-vocabulary retrieval without using an LLM or GPU in the retrieval path.

Open-source framework that orchestrates multiple autonomous AI agents through role-playing collaboration, using intuitive crew/agent/task abstractions. CrewAI Cloud and Enterprise add hosting and operational tooling.

Cronloop is an agent orchestration platform that schedules and runs Claude Code or Codex tasks at intervals ranging from five minutes to once a week. The platform executes agents in sandboxed environments, maintains state memory between runs, encrypts environment credentials, and provides a remote MCP server to manage agents directly via ChatGPT or Claude.

An open-source, plugin-based agent harness from DeepSeek AI designed for building AI agents.

Agent-native analytics engineering toolkit: point it at your warehouse and dbt project and it learns the structure, writes transformations, and pinpoints exactly what to fix when schemas drift. Built for analytics and data engineers who want coding agents to do more of the data work.

An open-source LLM application development platform for building, testing, and deploying agentic workflows, RAG pipelines, and AI assistants.

An open-source, visual developer platform for building and orchestrating AI agents, chatbots, and retrieval-augmented generation workflows using a drag-and-drop interface.

GreatArrow.ai provides a shared memory and tools layer for AI assistants like Claude, ChatGPT, Gemini, and Cursor. It maintains unified context across different AI interfaces using a four-layer memory model and allows assistants to execute actions across connected systems via MCP integrations.

Declarative language and toolchain for AI agents: define agents, tools, and prompts in HCL, then compile to a framework or manage them with plan/apply semantics on a hosted platform. Billed as "Terraform for AI agents".

Low-level orchestration framework for stateful, long-running agents, built by LangChain. Provides state management, persistence, and controllable graph execution; used in production by Klarna, Replit, and Elastic. LangGraph Platform adds managed deployment.

LlamaIndex provides document parsing, indexing, and orchestration workflows for AI agents and RAG pipelines. Its suite includes LlamaParse, a VLM-powered cloud parser for complex layouts, and LiteParse, a local open-source alternative for parsing PDFs, Office docs, and images.

Lumify is a sports intelligence API and hosted Model Context Protocol (MCP) server designed for AI agents and coding assistants. It provides structured, real-time sports data, odds, splits, and predictive intelligence (such as event probabilities, fair prices, and expected value) for sports like MLB, NFL, NBA, NHL, soccer, and tennis. It includes native SDKs, OpenAPI schemas, and LLM-optimized documentation to facilitate zero-hallucination agent integrations.

A managed hosting and orchestration platform designed to run AI agents in isolated microVMs. It supports automated sleep/wake states, encrypted secrets, and deployment via CLI, SDKs, or API, with templates for popular agent frameworks.

A task-specific small language model (SLM) optimized for converting user intent into valid, schema-bound tool calls. It reduces context-accumulation costs and execution latency by offloading schema binding from frontier models, and can be deployed via TrustedRouter or on-premises/VPC.

A shared knowledge base and context management layer designed for AI agents to read and write from. It exposes a Model Context Protocol (MCP) interface, automatically routes incoming updates, resolves conflicting data, and refactors large documents to keep them token-friendly.

An open-source client-side vector search engine that runs in the browser via WebAssembly. It queries IVF-LVQ8 indexes stored as immutable Parquet files on remote object storage using HTTP range requests, executing query embedding and ranking locally.

A fully managed vector database designed for building AI applications, agent memory, semantic search, and recommendation systems.

Qdrant is a high-performance, open-source vector search engine and database written in Rust. It supports native hybrid search (dense and sparse vectors), built-in multivector support, advanced metadata filtering, and real-time indexing for AI applications like RAG and recommendation systems.

Salestrics is a unified revenue workspace and CRM that acts as a hosted Model Context Protocol (MCP) execution proxy. It allows AI agents running in IDEs or chat clients to read and perform CRUD operations on live CRM, support, and billing data with built-in human-in-the-loop security gates.

An ingestion and pre-processing platform designed for GenAI that extracts, parses, chunks, and embeds unstructured data from over 64 file types into structured formats ready for AI models and analysis.

Weaviate is an open-source vector database and AI development platform that enables developers to store, index, and search high-dimensional vectors. It includes built-in embeddings generation, a natural language Query Agent, and features for building Retrieval-Augmented Generation (RAG) and personalized AI experiences.