ACE: Universal Memory Infrastructure for AI

> Stop explaining your architecture to ChatGPT every morning.
>
> ACE gives your AI tools (Claude, ChatGPT, Gemini, Cursor) persistent, structured memory that lives on your infrastructure.


---

The Problem

AI coding agents are amazing, but they're super-talented interns with amnesia.
- They forget your architectural decisions from last week.
- They re-introduce bugs you already fixed.
- They don't know why your codebase is structured this way.

The Solution

ACE (Autonomous Context Engine) is a memory layer that sits between you and your AI. It tracks decisions, patterns, issues, and context - so every AI session starts with full knowledge.

$3

| | ACE3 | Mem0 | Zep | ChatGPT Memory |
|---|---|---|---|---|
| Structured entity types (10) | Yes | No | No | No |
| Knowledge graph (12 relationship types) | Yes | No | Yes | No |
| AI Builder (plan generation) | Yes | No | No | No |
| Observer (AI monitoring AI) | Yes | No | No | No |
| Customer-hosted database | Yes | Partial | Enterprise | No |
| MCP-native (29 tools) | Yes | Basic | No | No |

---

Quick Start

$3

``bash
npm install -g @ace3-memory/ace
`

$3

`bash
ace login # Opens browser for authentication
ace start # Server runs on localhost:7777
`

$3

`bash


Remember something

ace remember --key "auth-decision" --value "We use JWT with 30d expiry"

Recall context

ace recall "why did we choose postgres?"

REST API (works with any AI tool)

curl -X POST http://localhost:7777/api/v1/my-project/decisions \
-H "Authorization: Bearer $ACE_JWT_TOKEN" \
-H "Content-Type: application/json" \
-d '{"decision": "Use PostgreSQL", "rationale": "Vector extensions + ACID"}'
`

---

Architecture

`
[ Your Machine ]
|
+-- [ ACE Server ] (localhost:7777)
| |
| +-- [ Your PostgreSQL ] (Neon, local, AWS, etc.)
|
+-- [ AI Tools ] (Claude, ChatGPT, Cursor) --> queries ACE via REST/MCP
|
+-- [ Dashboard ] (ace3-ai.com) --> visualize & manage memory
`

You provide: The database (PostgreSQL) and infrastructure.
We provide: The intelligence, dashboard, and tools.

---

Features

| Feature | Description |
|---------|-------------|
| 10 Entity Types | Memories, decisions, issues, work logs, architecture, patterns, best practices, observations, tags, relationships |
| Knowledge Graph | 12 relationship types with temporal traversal and visualization |
| Semantic Search | OpenAI embeddings + pgvector for intelligent recall |
| AI Builder | Generate project plans with GPT-4/Claude/Gemini |
| Observer | AI that monitors your AI for errors, drift, contradictions |
| Background Agents | 6 AI agents that maintain memory (cleanup, graph maintenance, triage, etc.) |
| MCP Server | 29 tools for Claude Desktop, Claude Code, Cursor, VS Code, ChatGPT, Gemini |
| REST API | 210+ endpoints for any AI tool or custom integration |
| Dashboard | Web UI at ace3-ai.com for visual memory management |

---

Pricing

All plans include a 14-day free trial with full Pro access.

| Tier | Price | For |
|------|-------|-----|
| Starter | £14.99/mo | Solo developers |
| Pro | £39.99/mo | Teams (10 users included, +£3/extra) |
| Founder | £299 one-time | Lifetime Pro+ (25 users, limited to 100 licenses) |
| Enterprise | Custom | Unlimited everything, SSO, SLA |

View Pricing

---

Contributing

We welcome contributions to:
- SDKs: Python SDK (ace3-memory` on PyPI, see ROADMAP.md Phase 1)
- Integrations: VS Code extension, browser extensions
- Parsers: Add support for Rust, Go, Java

---

Built by a developer, for developers.