---
title: "How Meko fits in your stack"
url: "https://docs.mekodata.ai/architecture/how-meko-fits/"
---

# How Meko fits in your stack

Where Meko sits in your architecture and how it deploys

Meko is the agent-native data infrastructure that enables multi-agent systems to learn together, building collective memory and shared knowledge that compounds across the entire system.

It supports:

- **Collective memory**. Learning compounds across the entire system, not just per-agent memory.
- **Shared knowledge**. Meko builds knowledge over time from conversations, real-time data sources, and slower-changing knowledge bases.
- **Decision Traces**. Connect raw conversations to execution traces of how LLMs think, agents process memory and knowledge, learn from it, and share that learning.

Meko is serverless, and multi-tenant (multi-agentic). It provides memory, conversation history, shareable knowledge, and full chain-of-thought traceability.

## Architecture

Meko integrates with any agentic framework through a standard MCP interface. It is built on top of a unified distributed PostgreSQL data layer that supports vector, SQL, NoSQL, graph, and search.

Meko sits between your agent SDK and the underlying storage, providing a unified data layer that replaces multiple standalone systems.

![Meko Architecture Diagram](/images/architecture.svg)

### What you get

A [datapack](/architecture/core-concepts/datapacks/) is the isolation unit for agent data in Meko. Datapacks can be shared by many users and connected to many agents. A datapack incorporates the following:

| Capability                    | Powered by (internal)                                                   | API                     |
|-------------------------------|-------------------------------------------------------------------------|-------------------------|
| Conversation history          | PostgreSQL (YugabyteDB YSQL)                                            | `conversation_*` tools  |
| Memory                        | Durable facts with hybrid semantic, keyword, and entity-aware retrieval | `memory_*` tools        |
| Knowledge base                | RAG pipeline (pg\_dist\_rag) + vector search                            | `knowledgebase_*` tools |
| Decision traces               | LLM reasoning traces and execution logs                                 | Observability dashboard |
| Observability, security, RBAC | Spans the full stack, not bolted on per-component                       | —                       |

![Meko Datapack Internals Diagram](/images/meko-datapack-internals.svg)

The underlying storage technologies are managed internally and not directly accessible.

### What Meko replaces

Without Meko, building an agentic application typically requires wiring together:

- A relational database for structured data
- A vector database for embeddings and similarity search
- A search system for memory retrieval
- Per-database observability, security, and RBAC
- Custom integration code for stitching together search results from each database

![What Meko Replaces Diagram](/images/meko-what-replaces.svg)

Meko replaces all of this with a single system. The memory, knowledge, and conversation layers share state and operate as one.

### Framework compatibility

Meko works with any agentic framework, including:

- CrewAI
- LangChain / LangGraph
- OpenAI Agents SDK
- Any MCP-compatible client

## Deployment

- **Cloud**: Currently, Meko is a cloud service, managed by YugabyteDB.