Automotive release pipeline
Auto App Platform takes your build from APK to storefront on rails: static validation, automated tests on real automotive OS devices, an AI review digest, and fan-out to every store channel you connect. One upload, one timeline, no portal-hopping.
The pipeline
The same road runs under every version you ship. Scroll it the way your build travels it.
01 / upload
Push a build from the portal or straight from CI with the release CLI: APK, localized listings, and store assets in a single shot. Measured on a real release: 1.4 seconds from keystroke to submitted.
$ autostore release app-release.apk
apk ............ 3.5 MB, signature ok
listings ....... en_US, de_DE
assets ......... icon + feature graphic + 3 screenshots
submitted ...... 1.4s02 / static validation
Signature, manifest, and listing completeness are checked the moment the APK lands - plus per-store preflight rules for every channel you target. Problems surface in seconds, before any store ever sees the build.
03 / device test
The build installs, boots, and runs on automotive OS 14 and 15 hardware profiles - two device targets per run. An AI reviewer watches every screen and writes findings. Platform hiccups never count against your app: only a real app failure can hold a release.
infrastructure failure? never your problem: the run is marked platform-error and your release is not penalized
04 / ai review digest
Validation findings, the device run, and store feedback collapse into a single digest per version with fix suggestions and a severity: ok, warning, or blocker. ok and warning publish automatically; a blocker holds the release with a one-click "publish anyway" override. No human in the loop unless it is you.
v1.1.1 digest: device runs report back on both targets. Two advisory findings on contrast in night mode, fix suggestions attached. Clear to publish.
05 / auto-publish
A passing version submits itself to every channel you have connected - idempotently, so retries never duplicate a release. Each store runs its own review; the status of all of them flows back into one timeline: pending, submitted, in review, approved, live.
Store channels
Channels are connected once per developer. From then on, every version that clears the pipeline fans out on its own - and reports back to one dashboard.
A complete store channel operated by the platform: in-car client API, targeted release by country, brand, and firmware ring, and downloads served straight to vehicles.
The first external OEM store is wired end to end: connect once with a token, then every release drafts, submits, and syncs review status and QA issues back automatically.
More lanes are in preparation. The channel model is store-agnostic: preflight rules, submit, and status sync are per-lane plugins - a new lane opens without changing anything about how you release.
AI review
Every store runs its own QA and files its own issues - the same crash arrives three times in three dialects. The platform pulls them all in, and AI merges the duplicates, groups them by root cause, and writes one digest per version with a concrete fix suggestion. Fix once, verify with a fresh device run, resubmit.
CI and API
Everything the portal does is available over the developer API and the release CLI - upload, test, submit, and stream pipeline status straight from your release job, with exit codes your CI can act on.
$ autostore release app-release.apk
upload ......... ok 1.4s
validate ....... ok
device test .... PASS 2 targets, AI-reviewed
digest ......... ok severity: ok
publish ........ submitted to every connected store
$ autostore status v1.1.1
your store: live | partner store: in review | issues: 0 openGet early access to Auto App Platform: one upload, tested in the car, live in every store you connect.
Get in touch