MacroscopeCode Review
AI code review on every pull request. Approvability approves the safe ones; Check Run Agents enforce your own rules.
Command your own software factory
AI code review that approves safe PRs, and cloud coding agents that verify their work.
Book a demoPull requestsReviewed in the cloud
- Fix typo in onboarding email#2481 · agentMacroscope approved
- Update README setup steps#2479 · danaMacroscope approved
- Rework checkout flow#2477 · agentNeeds a human
- Regenerate lockfile#2476 · agentMacroscope approved
- Update test snapshots#2474 · agentMacroscope approved
~91%of Pydantic's PRs are approved by Approvability, without a human4XPydantic went from 100–150 PRs a week to about 50015h → 2hWith Murmur, work that used to take Pydantic 15 hours now takes two3XCars24 tripled the share of one team's code reaching production in three months
“We feel like Macroscope has solved code review.”
Code review is the new bottleneck
Your coding agents open more PRs than people can review. Macroscope reviews every change in the cloud before it lands and approves the ones that don't need a human.
Each team sets its own guardrails
The policy lives in .macroscope/approvability.md. Anything uncertain goes to a human.
One platform for AI software development
Review, measure and build, all in the cloud.
Status
Status shows what changed and why, with pushed vs. landed coding time.
Agent
Answers questions about your codebase, writes code and opens pull requests, from Slack.
MurmurBeta
Cloud coding agents, each in its own VM, that verify their own work before they open a PR.
Cloud coding agents, with Murmur Beta
The part of the software factory that writes the code. Agents own changes end to end, from implementation through review.
Its own cloud VM
Each agent runs in a dedicated cloud VM that can run your full stack.
Bring your own agent
Use your Claude or Codex subscription and manage your own API keys.
A fleet in parallel
Dozens of agents in parallel, each in its own cloud sandbox.
Verifies its own work
Agents run tests, launch services and interact with the UI before opening a pull request.
Self-driving PRs
Murmur listens to GitHub and deals with review comments, failing checks and merge conflicts.
Works with Macroscope review
Agents answer review comments from macroscope[bot] until checks pass.
Works where you work
Murmur integrates with Slack, Linear, GitHub or your own systems.
Direct it from your terminal
Use your local machine to direct a fleet of cloud agents with the CLI or MCP.
Ship faster, with every change reviewed
Book a demoQuestions
What is an AI software factory?
A way of building software that goes beyond a single AI software engineer: coding agents write much of the code, and the system around them reviews, verifies and lands it. Macroscope runs that system: AI code review in the cloud for every pull request, and cloud coding agents in Murmur, now in beta.
How does Macroscope auto-approve safe PRs?
Approvability runs cloud code review on each pull request against the policy your team writes in .macroscope/approvability.md. PRs that meet it are approved automatically; anything uncertain goes to a human reviewer.
Do humans still review code?
Yes. Each team sets its own guardrails, and Macroscope sends every pull request it is not sure about to a person. Your engineers spend pull request review on the changes that need them, not on every change.
How is Macroscope different from other AI code review tools?
Automated code review usually stops at comments. Macroscope also approves the PRs that meet your policy, runs your own rules as Check Run Agents, and Status shows what changed and why, with pushed vs. landed coding time.
What is Murmur?
Murmur, in beta, runs background coding agents in the cloud. They are self-verifying agents: each works in its own VM with your full stack, checks its own work and opens a pull request that Macroscope then reviews.