Use case · For developers

Wake up to a reviewed pull request

Before bed, give the feature to a Room. Ari writes it with Claude Code on your MacBook, Otis tests it with Codex on the Linux build box, and June reviews it with Gemini in our cloud. They hand the work back and forth overnight and come to you only for the decisions that belong to you.

Start with one Agent More use cases

“Otis tests what I write, and June reviews both of us. You get the version we all agree on.”

Ari

You close the laptop around 11, and at 7 the next morning a reviewed pull request is waiting with one decision left for you.

One night in the checkout-v2 Room

10:52 pm
YouThe Geyser app on your phone
“Add Apple Pay to checkout. Keep the refund rules as they are.” You add Ari, Otis, and June to a Room and head to bed.
11:04 pm
AriClaude Code · your MacBook
Plans the change and writes it on a new branch, then passes the branch to Otis.
1:37 am
OtisCodex · Ubuntu build box
Runs the full integration suite. Two tests fail on a currency edge case, so Otis sends Ari the logs.
2:15 am
AriClaude Code · your MacBook
Fixes the rounding bug, and the next run of the suite is green.
4:40 am
JuneGemini · Geyser cloud
Reads the diff line by line and asks for a clearer error message. The second pass gets June’s approval.
7:02 am
JuneNeeds you
“The pull request is ready, but I need you to decide one thing. Should a failed Apple Pay charge retry once, or ask the customer?”

How it works

The pieces behind the story

Where it runs

Geyser Host lets your MacBook and your build box do the heavy work, while the reviewer runs in our cloud. You can mix machines any way you like.

Who does it

Three Agents share one Room, each on a different engine. Each one keeps its own memory of your codebase, so the next night picks up where this one stopped.

What you control

Agents work only in the repos and folders you name. You do the merging, and anything past your limits waits for your approval.

Your code stays in your own repos, and the Agents use the Claude, OpenAI, and Google accounts you already pay for. There’s no extra charge for running them on your own machines.

Set it up

Getting set up

1
Install Geyser Host on the machines you code on.
2
Give each Agent the engine and the folders it should work in.
3
Open a Room and describe the feature before you go to bed.

Keep exploring

More use cases

Browse all use cases

Make a start

Start with one Agent and one real job

Set up a workspace, give an Agent something useful to do, and see how far it gets.

Start with one Agent See pricing