An agent that knows your whole workspace.
AI that works from everything you have: your email, messages, tasks, docs, and calls, all with your exact permissions. It answers, takes action, and runs on a schedule.
Drafted it. Left the To field empty so you can drop in Alex, BoxGroup, and the rest of the cap table.
Hey,
Quick heads up: we're launching Macro on Product Hunt on July 1.
This is the V1 launch: Macro as the open source company OS, one platform replacing Slack + Notion + Linear + Superhuman.
How you can help:
- Upvote + comment the morning of July 1
- Repost / quote our launch tweet
- Forward to founders & eng leaders in your network
Aiming for #1 Product of the Day: the first hour matters most.
Jacob
On track for Friday. The launch deploy is the only blocker, security cleared Monday, and the truth came from the go / no-go call, not the doc that said otherwise.
Macro turn the Atlas go / no-go into tasks and assign them.
Macro summarize this thread for someone joining late, and list the open questions.
Launch is set for Friday. Eng is finishing the staging deploy; security cleared Monday.
Open: who owns the launch email?
Add any MCP server under Settings → Connectors and your agent can reach it too.
macro https://mcp-server.macro.com/mcp
Claude Code, Codex, or any MCP client gets the same tools:
Open source under the AGPLv3. Read it, self-host it, make it yours.
Star on GitHubWhat people ask their agents
Paste these into an agent chat, an automation, a channel, or your own MCP client.
“What changed on Project Atlas this week, and what’s blocking it? Check tasks, channels, and call transcripts.”
“Go through yesterday’s planning call and create a task for each action item. Assign each to whoever took it.”
“Draft a reply to this thread. Warm but brief. We can do June 20, but ask them to send the revised scope first.”
“Every Friday at 4pm: recap the week in #general, covering what shipped, key call decisions, and anything that slipped.”
“Find the task for the bug I’m fixing, read the linked discussion, then set it to In Review with a summary comment.”
“Import my Notion docs from the “Engineering” workspace as Macro docs. Keep the folders, skip anything archived.”
Context beats a clever model.
A general chatbot has no idea what you're working on, so you spend half your time pasting context in. A Macro agent already has all of it, and it can actually act on it.
Questions, answered
What can an agent actually do?
An agent has access to everything in your workspace, so it can answer questions across your email, messages, tasks, docs, calls, and CRM. It can also act: draft and send email, create documents, create and update tasks, search the web, even run code. Ask in an agent chat (c + a), mention @Macro in a channel, or call it from any MCP client.
How is this different from ChatGPT or Claude?
Unified context. Ask a general chatbot “what’s the latest on Project Atlas” and it has to search Slack, Notion, Drive, and email one at a time, then guess which conflicting doc is current. It still can’t open the call recording that holds the real answer. Because everything already lives in Macro, your agent reads from one source of truth, including that call.
What context does an agent have access to?
Everything you can see, and nothing you can’t. Agents inherit your exact permissions, so they read the same email, channels, docs, tasks, and calls you do. On a team, they also get team-level memory: not just your chat history, but the work, emails, docs, and calls happening across your team.
Is it safe to let an agent take actions?
Agents only ever act as you. Anything they create or send is attributed to your account, not a bot, and they can’t touch anything you don’t already have access to. A good habit is to start prompts with “draft, don’t send” until you’re comfortable, then review and send from the composer.
What are automations?
Automations are agents on a schedule: recurring jobs like a daily inbox brief or a Friday team recap. Create one in the agents module under the Automations tab, write the instructions (you can @mention channels, docs, and people), and pick a cadence. Each run’s result appears in your inbox and in the agents tab.
What is @Macro in channels?
Mention @Macro in any channel to pull a fast agent into the conversation, with the channel as context. It’s perfect for “summarize this thread for someone joining late” or “list the open questions nobody has answered” without leaving the discussion.
Can I connect my own tools?
Yes. Macro is built as an extensible platform. Connect any MCP server under Settings → Connectors (Notion, Slack, Linear, and more) and your agent can reach those tools too, alongside everything native to Macro.
Can I use my Macro workspace from Claude Code or my IDE?
Yes. Macro exposes a remote MCP server at https://mcp-server.macro.com/mcp. Add it to Claude Code, Codex CLI, or any editor that speaks MCP and sign in through Macro’s OAuth flow. Your coding agent then gets the same reach: search the workspace, read threads and docs, create documents, send email, and update task properties.
Is it open source?
Yes. Like the rest of Macro, agents are fully open source under the AGPLv3, not “open core.” Read it, self-host it, build on it. To build on Macro under a different license, contact licensing@macro.com.
Give your agent the full picture.
Connect your account in 30 seconds and put an agent to work across your email, calls, docs, and tasks.