Cognee alternatives
Other ai memory & retrieval platforms include Letta, Mem0, Zep, Graphiti, Qdrant, Weaviate. They share a category, but their workflows, coverage, and terms can differ. These are options to evaluate, not ranked recommendations.
Letta
Build agents that retain and manage state and memory across interactions.
Mem0
Add persistent personalized memory to AI applications and agents.
Zep
Provide AI agents with memory and contextual knowledge from ongoing interactions.
Graphiti
Build temporally aware knowledge graphs for agent memory.
Qdrant
Build vector search and retrieval systems with an open-source vector database.
Weaviate
Store data and embeddings for semantic search and AI retrieval.
Pinecone
Manage vector retrieval infrastructure for AI applications.
Chroma
Store and retrieve embeddings for AI applications with an open-source database.
LanceDB
Store and search multimodal data and embeddings in a vector database.
Vespa
Build scalable search and recommendation applications with vector and keyword retrieval.
pgvector
Add vector similarity search to PostgreSQL using an open-source extension.
Marqo
Build multimodal search using text and image embeddings.
Agentset.ai
The open-source platform to build AI apps that deliver reliable answers. Production-grade RAG in minutes, no expertise needed.
AI/ML API
Access 1000+ AI models through a single API for text, image, video, audio, code, embeddings, and more. Compare pricing, latency, and capabilities—place with AIMLAPI.
Heabsy
AI implementation partner for regulated companies: document assistants, RAG search, agents. Inference on own hardware in the EEA.
Agentcloud
Agent Cloud is like having your own GPT builder with a bunch extra goodies. The GUI features 1) RAG pipeline which can natively embed 260+ datasources 2) Create Conversational apps (like GPTs) 3) Create Multi Agent process automation apps (crewai) 4).
Recall
Recall is a personal AI knowledge base. Save, summarize, and chat with your articles, videos, podcasts, PDFs, and notes — with ChatGPT, Claude, or Gemini.
Erag
an AI interaction tool with RAG hybrid search, conversation context, web content processing and structured data analysis with LLM / GPT.
SinglebaseCloud
Singlebase is an AI-native backend platform and Firebase alternative, combining document and vector databases, authentication, storage, and built-in AI services to power modern applications at scale. Build and scale backend + AI services.
Llm-Embed-Jina
Embedding models from Jina AI.
Local-Rag
Ingest files for retrieval augmented generation (RAG) with open-source Large Language Models (LLMs), all without 3rd parties or sensitive data leaving your network.
NeuroLink
The pipe layer of an AI nervous system — one interface connecting provider neurons to your application, across three inference types: generate, stream, and a calibrated decide (via TypeSafe Jev). MCP-native, voice (TTS/STT/realtime), RAG, memory, file.
RAGFlow
RAGFlow is a open-source Retrieval-Augmented Generation (RAG) engine that fuses RAG with Agent capabilities to create a superior context layer for LLMs.
SwarmVault
The local-first LLM Wiki: open-source knowledge graph builder, RAG knowledge base, and agent memory store. Built on Andrej Karpathy's pattern. An Obsidian alternative for personal knowledge management, AI second brain, and durable Claude Code / Codex /.
WFGY
WFGY is heading toward WFGY 5.0 Polaris Protocol, a major open-source release for AI reasoning, RAG, agents, and real-world workflows. Includes Problem Map, Global Debug Card, WFGY 4.0, and the CFV Easter Egg.
ruvnet/ruflo
🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code /.
How should you evaluate a switch?
Test retrieval quality on your own information, including access restrictions and outdated documents. Confirm deletion, retention, indexing updates, embedding costs, and how the application handles missing evidence. Include migration work, access, and cancellation terms in the evaluation. A shared category does not make two products direct substitutes.
