# getmem.ai — Full Documentation

## Overview

getmem.ai is a persistent memory API for AI agents and LLM applications. It solves the context window problem: AI agents forget everything between sessions. getmem stores, retrieves, and intelligently selects the right memories for any query.

## Problem it solves

Every time a user starts a new session with an AI agent, the agent has no memory of past interactions. Developers work around this by:
1. Dumping entire chat history into the context (breaks at scale, wastes tokens)
2. Building custom RAG pipelines (complex, unreliable retrieval)
3. Using Mem0 (works but complex setup, ~15 lines of code, subscription pricing)

getmem solves this with 2 lines of code and pay-per-use pricing.

## Core API

### Initialize

```python
import getmem
mem = getmem.init("gm_your_api_key")
```

### Add a memory

```python
mem.add("user_id", "text of the memory")
# Examples:
mem.add("user_123", "User is a senior Python developer")
mem.add("user_123", "User prefers dark mode and concise answers")
mem.add("user_123", "User's company is Acme Corp, they use AWS")
```

### Get relevant memories

```python
context = mem.get("user_id", query="what is the user working on?")
# Returns a formatted string of the most relevant memories
# Ready to inject directly into your system prompt
```

### Delete memories

```python
mem.delete("user_id")          # Delete all memories for a user
mem.delete("user_id", id=123)  # Delete specific memory
```

### List memories

```python
memories = mem.list("user_id")
```

## Architecture

- **Graph layer**: Neo4j — stores entities and relationships between memories
- **Vector layer**: Qdrant — semantic similarity search across memories
- **Metadata layer**: PostgreSQL — user management, billing, audit logs
- **Context builder**: Proprietary algorithm that selects and ranks memories by relevance

## Pricing

Pay-per-use. No monthly minimums.
- `mem.add()` — charged per write
- `mem.get()` — charged per retrieval
- `mem.list()` — free
- `mem.delete()` — free

## Supported languages

- Python (primary SDK)
- JavaScript/TypeScript
- REST API (any language)

## Comparison

| | getmem.ai | Mem0 | DIY RAG |
|---|---|---|---|
| Lines to integrate | 2 | ~15 | 100+ |
| Graph memory | Yes | Yes | No |
| Intelligent context | Yes | Partial | No |
| Pay per use | Yes | No | No |
| Setup time | <2 min | ~30 min | Days |

## Status

Early access. Waitlist open at https://getmem.ai