Public early alpha · open source

Define your vision.
Agents build the reality.

The self-building application framework where autonomous AI agents scaffold, wire, and extend reviewable Django/React software. Start with a public-alpha stack and grow it addon by addon.

Stay up to dateHow it works
MCP-native
Self-hosted or angee.sh cloud
Docker deployment

Install the CLI

$curl -fsSL https://angee.ai/install.sh | sh
$angee up# your company is running

Requires Docker. View on GitHub

Product snapshots

One workspace, many agent-built surfaces

Notes, files, platform controls, agent sessions, and account flows live in the same composed application shell.

Angee product collage showing the notes workspace, platform console, files workspace, agent console, and branded login

Why teams choose ANGEE

Stop hand-wiring infrastructure. ANGEE ships it solved — then lets AI extend it to anything you need.

Self-building applications

Describe what you need. ANGEE's agents scaffold reviewable Django + React addons — models, GraphQL, permissions, UI, resources, and migrations — against the same contracts humans use.

Zero-config architecture

Addons declare their contracts once. The composer turns those declarations into a runnable project, so routes, schema, resources, and permissions stay deterministic instead of hand-wired.

MCP-native agents

Every agent is a first-class resident with fine-grained permissions, declared capabilities, and per-session grants. Connect Claude, Cursor, or any MCP client directly to your running ANGEE instance.

Enterprise-grade, out of the box

Django 6, GraphQL, REBAC permissions, encrypted fields, typed React, and operator-managed stacks are built in. Self-host today; managed cloud is coming soon.

How ANGEE works

From open-source alpha to a composed platform — then keep extending with AI

01

Bootstrap

Clone, initialize, run

Clone the open-source runtime, install the ANGEE CLI, and bring up the example stack: auth, REBAC permissions, GraphQL, subscriptions, MCP, and the current base addons.

git clone → angee init --dev → angee dev
02

Compose

Pick your building blocks

Compose base addons such as iam, integrate, agents, mcp, storage, knowledge, parties, messaging, social, platform, and operator. The composer emits the runnable project deterministically.

Addons → angee build → Runtime
03

Extend

Describe it, AI builds it

Need something custom? Describe the capability in natural language. AI agents scaffold an addon, wire it into the platform contracts, and validate it against the schema — you review and ship.

Describe → Agent Scaffold → Review → Ship
04

Deploy

Self-hosted or cloud, your choice

The same manifest that runs locally deploys to your own infrastructure — or, soon, to a managed angee.sh instance. The platform can self-update under operator supervision. You stay in control.

angee up → self-hosted or angee.sh

Three-layer architecture

Addons, agents, and ecosystem modules compose into one deterministic Django + React runtime

Addons

Modular building blocks

Django apps with source models, GraphQL schema parts, permissions, resources, and React surfaces. Add one to the project and the composer emits the deterministic runtime.

  • Build-time composition
  • GraphQL + MCP
  • REBAC permissions
  • React routes + views

Agents

Autonomous AI workers

First-class residents with fine-grained permissions, declared capabilities, and per-session grants. Content redaction, context firewalling, and a full audit trail come standard.

  • Permission-scoped execution
  • MCP protocol
  • Context firewalling
  • Full audit trail

Ecosystem

Community-powered growth

Base addons and agent-created modules form a growing library. IAM, resources, integrate, MCP, operator, and platform keep the shared surface consistent.

  • Base addon library
  • Agent-generated modules
  • Shared tool registry
  • Composable templates

Platform Architecture

How the layers connect

MCP surface
·
Agent orchestration
·
Permission-scoped execution
Identity
+
Messaging
+
Files
+
Your addons
GraphQL + subscriptions
Permissions + Audit
Models + resources

Who do you trust?

Agents with identities, permissions with no back doors, an audit trail that can't be talked out of — on infrastructure you own.

The answer is architecture

Growing ecosystem

Pre-built addons for common problems, plus the products built on ANGEE

Platform Core

Included in every installation
iam

Identity, the swappable user model, and the REBAC permission hub

resources

Tiered data import/export for master, install, demo, and seed resources

mcp

Authenticated MCP server at /mcp, exposing addon tools to agents

integrate

OAuth connections, credentials, and the bridge/capability runtime

agents

Inference provider/model catalogue and the runtimes agents run on

platform

Admin console for schema exploration, addons, and the marketplace board

Addons

Enable what you need
messaging

Threads, messages, and channel bridges that ingest mail and social conversations

storage

Drives, folders, and content-addressed files with chunked uploads

knowledge

Markdown-first knowledge base with vaults of versioned pages

parties

People, organizations, addresses, affiliations, and directory sync

social

Public feeds, engagement metrics, following/timeline, and reactions built on messaging

Built on ANGEE

Payloads you can angee up

The endgame

ACM — Agentic Company Management

Business software has always managed things for humans to work on. ACM is what comes after: AI agents that actually run the back office, on a substrate you can read, audit, and own. angee up your business — and the agents take their shift.

CRM

Customer relationship management

Manages your customers. Humans do the work.

ERP

Enterprise resource planning

Manages your resources. Humans still do the work.

ACM

Agentic company management

Agents operate the company — bookkeeping, invoicing chase, procurement, pipeline hygiene — under structural permissions, with human approval gates.

Every company gets an AI back office. The one they can trust is the one they can read.

Meet ARPEE, the agentic ERPWho do you trust?

Battle-tested tech stack

Built on proven open-source foundations — best tool for each layer, no proprietary lock-in

Backend

Python 3.14+Django 6Strawberry GraphQLFastMCPdjango-zed-rebacchannels + uvicornuv

Frontend

React 19TypeScript 6RefineTanStack RouterVite 8Tailwind 4

Infrastructure

Angee CLIDocker Composeprocess-composeGraphQL codegenpytestVitest

Pricing that scales with you

Start free with self-hosting. Move to angee.sh managed instances when you want the convenience.

Open Source

Full framework for teams with in-house DevOps

Free/ forever
  • Complete source code access
  • Docker Compose deployment
  • All core apps & agent framework
  • Community ecosystem apps
  • Community support
  • Bring your own LLM
Get started

angee.sh

Managed ANGEE instances — your payload, our operations

Coming soon
  • One command from repo to running instance
  • ARPEE or fyltr payloads — or your own addons
  • Compute-priced workspaces, never per-seat
  • Backups, SSO & team management
  • Managed updates & monitoring
  • Sovereign / BYO-cloud tier planned
Join the waitlist

Enterprise

White-glove deployment with compliance controls

Contact us
  • On-premises or private cloud
  • Role-based access policies
  • Audit logs & compliance tools
  • Dedicated account manager
  • Custom integrations & SLA
  • Training & knowledge transfer
Talk to sales

Frequently asked questions

Everything you need to know about ANGEE

What exactly is ANGEE?

How is this different from other AI coding tools?

What is MCP and why does it matter?

What tech stack does ANGEE use?

Can I use my existing code with ANGEE?

How does "extend with AI" actually work?

Is ANGEE really fully open source?

What is ACM (Agentic Company Management)?

What is angee.sh?

Can I use ANGEE in production?

Ready to let agents build for you?

Angee is already open source in early alpha. Join the update list for release notes, addon progress, and managed-cloud news.