macro-inc/macro

Macro is a unified workspace for teams: email, chat, docs, tasks, agents, calls, and CRM — @-linked together with shared AI memory.

Macro – an integrated AI‑augmented workspace

What it is – Macro is an open‑source, AGPL‑v3‑licensed platform that tries to replace the collection of tools most startups use (email, chat, docs, issue tracking, CRM, etc.) with a single, tightly‑linked web app. It is built with a SolidJS front‑end and a large Rust code‑base for the back‑end, and it ships a full set of services (≈42) that run together via Docker/Compose or Pulumi.

Core idea – Every piece of information lives in a bidirectional graph and can be @‑linked across the system. An email can be turned into a task, a task can reference a PR, a CRM record can be mentioned in a chat message, and agents (AI assistants) can read or edit any of those objects because they share the same underlying database.

Key components

Block What you get
Email Unified multi‑account inbox, keyboard‑first navigation, AI helpers for drafting, searching attachments, and creating tasks directly from mail.
Messages Channel‑style chat optimized for technical discussions, with inline replies, thread collapsing, and automatic linking to tasks, PRs and CRM records.
Tasks Linear‑style issue list that is created from any context (email, doc, chat) and stays linked to its source.
Docs Real‑time collaborative markdown editor built on CRDTs (conflict‑free replicated data types), with version history, offline support and full @‑linking.
Canvas 2‑D board where you can drop docs, tasks, emails, etc., and keep the same linking semantics.
Agents AI assistants that have team‑level memory – they can query the whole workspace, draft replies, create tickets, or edit documents. The memory is stored as markdown and can be exported.
Calls Recorded calls that are transcribed and added to the graph so agents can reference them.
File storage Auto‑imported from mail and chat, searchable by content.
Pull requests GitHub PRs appear as first‑class objects, linked to tasks and chat.
CRM Contact and company objects that sync with email, can be @‑mentioned, and automatically create channels for deals.

AI integration – Macro exposes a Model‑Control‑Protocol (MCP) that lets any LLM (OpenAI, Anthropic, Claude, etc.) act on the workspace. Agents can read the unified memory, run actions (create tasks, edit docs, send mail) and are not limited by rate‑limits. The platform also provides a search surface that lets agents pull information from PDFs attached to emails.

Running the code – The repo contains:

  • apps/web – the SolidJS client (also packaged for Tauri desktop and mobile).
  • services – the Rust micro‑services (workers, Lambda‑style handlers).
  • crates – reusable Rust libraries.
  • packages – shared TypeScript utilities for collaboration.
  • infra – Pulumi IaC for cloud deployment.
  • Docker compose files for local development.

The README points to a RUNNING_LOCALLY.md guide for spinning up the full stack or just the front‑end against the hosted backend.

Why it matters for AI/ML folks – Macro is a real‑world example of a knowledge graph backed workspace where LLMs can be given full context (emails, chats, docs, CRM records) in a single query. Its open‑source nature lets researchers experiment with custom agents, memory‑synchronisation strategies, or CRDT‑based collaborative editing without building the plumbing from scratch.

License & community – AGPL‑v3, fully open source, with a contribution guide, security bounty program, and a small team that is hiring. The project encourages self‑hosting and offers commercial support for enterprises.

Related

  • Project
  • Project
  • Project
  • Project
  • Project