Your builds
Runs on your build machines or cloud accounts.
Services Testing & QA
Playtest agents for every build.
We set up agents that play your builds overnight, check them against what should happen and file reports your QA team can act on in the morning. Your testers confirm and prioritise.
How it runs
One night of testing.
- Handled by an agent
- Faster with AI
- Your team signs off
-
Nightly build
CI produces a build and starts the run on its own.
Handled by an agent -
Agents play it
Scripted routes and exploring agents drive the game through a test hook.
Handled by an agent -
Checks and captures
Crashes, soft locks and frame spikes are caught with a clip and the game state.
Handled by an agent -
Triage
Duplicates are matched against your tracker and repro steps written for new issues.
Faster with AI -
Tester confirms
Your QA team verifies, sets priority and assigns.
Your team signs off
Example run
What a day of it looks like.
An illustrative log of the setup at work. Your tools, numbers and review points will differ.
- Agent
Build 1482 received from CI, 12 agents started.
- Agent
Routes: 38 of 40 critical paths completed.
- Agent
Soft lock: the north door in level 3 stays shut if the lever is used twice.
- Agent
Frame spike: 41 ms in the market area, clip and state attached.
- With AI
Triage: 2 new issues filed, 5 matched to open tickets.
- Your team
QA lead confirms both new issues, sets priority and assigns them.
What we set up with you
The pieces behind it.
Agents can only report what they can see, so most of the setup is instrumentation.
- Test hooksA small interface in the build that exposes game state and accepts inputs.
- Routes and goalsCritical paths scripted once, with goal-driven agents exploring around them.
- Failure rulesWhat counts as broken: crashes, soft locks, out-of-bounds, missing assets, frame budget.
- Regression baselinesScreenshots and timings compared with the last good build.
- Tracker reportsIssues filed with clip, log and state, matched against open tickets.
- Device coverageRuns spread across the machines or cloud devices you already have.
Where it fits in your studio
Your tracker
Reports arrive in the tracker your team already uses.
Your testers
People confirm every issue before it is assigned.
Your CI
Runs trigger from your existing pipeline.
Starting out
A first pilot.
A first pilot covers one mode or set of levels with a nightly run.
From you
Access & a lead
Build access, a way to add a test hook, and access to your tracker.
From us
A working setup
Nightly reports for that area, the hook and routes, and a comparison with your manual pass.
Next step
Tell us what you're making.
A few lines is enough. We'll reply within one to two business days with a suggested starting point.