Stop babysitting terminals.
Blume follows every session for you.
LESS CHECKING IN
We’re hiringDownload
Give your coding agents a brief. They plan, build and test the change. Blume brings you the code and the test results so you can decide what ships.
Align archived ticket handling
Separate catalog sync detection
Refresh billing settings flow

Blume follows every session for you.
LESS CHECKING IN
Blume follows every session for you. Each change, from the first instruction to the merged commit.
FROM BRIEF TO MERGE
Your agents plan, build and test. You approve outcomes and make decisions. Follow each delivery through review and into release.
YOUR FACTORY FLOOR
Follow one delivery
Open one delivery to follow its brief, remaining work and test evidence. Review the revision and decide what ships.
The delivery
One product change with its brief, task tree, agent runs, code revision, tests and screenshots, intact from the first instruction to production.
Your agents plan, build and test. You approve outcomes and make decisions. Follow each delivery through review and into release.
Remaining work
Your agents plan, build and test. See what’s left at a glance.
Your agents plan, build and test. You approve outcomes and make decisions. Follow each delivery through review and into release.
Every step, screenshotted
Review test steps and screenshots. Approve the revision you ship.
Your agents plan, build and test. You approve outcomes and make decisions. Follow each delivery through review and into release.
Approve the revision
Open one delivery to follow its brief, remaining work and test evidence. Review the revision and decide what ships.
Your agents plan, build and test. You approve outcomes and make decisions. Follow each delivery through review and into release.
Interactive product preview · Illustrative data
As agents produce more code, the hard part is keeping their work visible, trustworthy and reviewable.
Know what every agent is doing and what is left before you ship. Today, that means piecing together terminals, chat logs and branches.

An agent saying “done” is only the beginning. You need evidence that the change works and meets your team’s definition of done.
Several agents can build at once. Your attention is still finite. Bring the work and its evidence together so you can make one clear decision.
What you get
The process tells agents what “done” means, and every claim arrives with its evidence.
Agents work longer without supervision, and several can work in parallel without silent collisions.
How The delivery model lets agents coordinate, and the agentic loop lets the agent catch issues before involving you.
Every claim has evidence tied to a specific revision. You see what’s left at a glance.
How Evidence is pinned to the exact task, delivery and commit. Visual verification at zero friction.
Acceptance criteria become tests before review, and suites rerun before every review.
How Bug reports are reproduced with a failing test first, then fixed and rerun.

Get started
Download Blume, connect your agents, and review the evidence.

Install the plugin into Codex or Claude Code. Every conversation connects on its own.
Your agents plan, build and test. See what’s left at a glance.
Review test steps and screenshots. Approve the revision you ship.
Vision · On the roadmap
A shared, versioned domain model gives every agent the same language, rules and edge cases. Each session starts with your team’s context, not a blank slate.
One model, shaped by real work

Entities, relationships and the words your team uses
Business invariants, permissions and boundaries
Critical paths and their expected outcomes
Known edge cases across time zones and currencies
Conventions and patterns agents must preserve
Reviewed decisions, findings and bug reproductions keep it current as your product changes.
Plan
Write acceptance criteria in the language your team uses.
Build
Follow existing business rules and architecture boundaries.
Test
Catch permission gaps, currency mistakes and dates shifting across time zones.
About us
Blume is built by two engineers who met at NTNU in Trondheim and have been building software together and apart ever since.
We believe watching your coding agents work should be effortless — and that the tools you rely on should respect your data.
Co-founder · Oslo, Norway
Peder has spent his career helping organizations — from large enterprises to startups — succeed with software. He holds a master’s degree from the Norwegian University of Science and Technology, where his thesis explored forecasting models and cost-optimization software for digital advertising.
He is an AWS-certified solutions architect and a certified application security engineer, with a deep background in building secure, reliable systems.
LinkedInCo-founder · Toronto, Canada
Olav studied at the Norwegian University of Science and Technology and has been building for the web ever since. Most recently he helped build Wordware (YC S24) in San Francisco, turning AI infrastructure into a platform anyone could use to create AI workflows.
He brings that same conviction to Blume: powerful tools should feel effortless.
LinkedInCoding agents changed how software gets built — but keeping track of what they're doing, what they cost, and what they touch is still harder than it should be.
We're building Blume to make that effortless: one place to watch every coding agent, designed local-first so your conversations stay yours.
Your first delivery
Download Blume, pick one change and follow it from the first brief to the revision you approve.
Know what your agents built.
We’re hiring · Open rolesBlume 2.0.26 unifies screenshot viewing, gives notifications their own read state and app badge, speeds up reviews, and cuts background database load across busy workspaces.
Read the post →