Neura Market
**Memori BYODB Documentation** logo

**Memori BYODB Documentation**

Free

Agent-native memory infrastructure for LLM applications

FreeFree tier
Inputs: text
Type
Open Source
Company
Memori Labs
Visit Website GitHub (40 stars)

About **Memori BYODB Documentation**

Memori is an agent-native memory infrastructure for LLM applications, agents, and copilots. It continuously captures interactions (conversations, tool calls, decisions, workflow steps) and extracts structured knowledge, then intelligently ranks, decays, and retrieves the most relevant memories across sessions. Memori's Advanced Augmentation turns raw conversations into searchable memories, and its Agent Trace Execution converts execution history into structured memory primitives that agents can recall and reuse. The tool runs asynchronously to minimize impact on response times. Memori BYODB (Bring Your Own Database) gives users full data ownership and the freedom to choose from many databases (CockroachDB, MariaDB, MongoDB, MySQL, OceanBase, Oracle, PostgreSQL, SQLite, TiDB) and managed providers (AWS RDS, Neon, Supabase). It supports multiple LLM providers (OpenAI, Anthropic, Gemini, Grok, OpenAI-compatible via base_url) and integrates with LangChain, Agno, and Pydantic AI. Intelligent Recall surfaces memories ranked by relevance and importance with decay, and supports semantic search. The tool is open source (self-hosted free) with a cloud free tier and production plans.

Key Features

Bring your own database (BYODB) – supports CockroachDB, MariaDB, MongoDB, MySQL, OceanBase, Oracle, PostgreSQL, SQLite, TiDB, and compatible managed providers
Intelligent Recall – memories ranked by relevance and importance with decay, semantic search for manual recall
Advanced Augmentation – extracts structured knowledge from raw conversations and tool calls
Agent Trace Execution – captures tool calls, decisions, workflow steps, outcomes as structured memory primitives
Asynchronous background processing minimizes latency impact on response path
Multi-LLM support – OpenAI, Anthropic, Gemini, Grok, OpenAI-compatible providers via base_url
Integrations with LangChain, Agno, and Pydantic AI
One-line setup – connect database and LLM, memory capture and recall work automatically
Dashboard for API keys, usage, and (cloud) Graph Explorer and Playground
Full data ownership and compliance – data stays in user's database, no third-party storage

Pros & Cons

Pros
  • Full data ownership and control – data stays in your own database on your infrastructure
  • Database freedom – supports a wide range of SQL databases and managed providers
  • Intelligent recall with relevance ranking and decay reduces memory clutter
  • Traces execution events (tool calls, workflow steps) for richer memory beyond conversation
  • Asynchronous processing minimizes latency impact on real-time responses
  • Open source free self-hosted option with community support
  • Easy setup – one-line connection and automatic memory capture/recall
  • Supports multiple LLM providers and frameworks (LangChain, Agno, Pydantic AI)
Cons
  • Requires users to manage and maintain their own database infrastructure (BYODB model)
  • Cloud free tier has limited memory capacity (5,000 created, 15,000 recalled)
  • Production pricing is high for small teams (starts at $60K/year for Team plan)
  • Relatively new tool; documentation and ecosystem may still be evolving
  • No built-in UI for managing memories beyond basic dashboard (cloud only has Graph Explorer/Playground)

Best For

Building AI agents with persistent memory across sessionsCustomer support copilots that remember user preferences, history, and contextPersonalized AI assistants that recall past interactions and decisionsLLM applications requiring context-aware responses without clutterEnterprise applications needing compliance, data ownership, and custom analyticsMulti-agent systems sharing memory pools with role-based access control (ReBAC)

FAQ

What databases does Memori BYODB support?
Memori supports CockroachDB, MariaDB, MongoDB, MySQL, OceanBase, Oracle, PostgreSQL, SQLite, and TiDB, as well as managed providers like AWS RDS/Aurora, Neon, and Supabase through their compatible engines.
Is Memori open source?
Yes, Memori offers an open source free self-hosted version that you can run on your own infrastructure. The source code is available on GitHub.
How does Intelligent Recall work?
Intelligent Recall ranks memories by relevance and importance, with an intelligent decay that causes older or less relevant facts to recede over time. This ensures your AI stays contextually aware without clutter. Memories can also be retrieved via semantic search.
What LLM providers are supported?
Memori directly supports OpenAI, Anthropic, Gemini, and Grok (xAI). It also supports OpenAI-compatible providers (such as Nebius, Deepseek, NVIDIA NIM, Azure OpenAI) via the base_url parameter. Bedrock is supported through LangChain ChatBedrock.
What is the pricing model?
Memori offers a free open source self-hosted version, a free Cloud tier (with limits: 5,000 memories created, 15,000 recalled), and production plans starting at $60K/year for a single production agent.