Agent Device
CLI to control iOS and Android devices for AI agents
AI-Native Test Automation
Expensify uses agent-device for bug evidence. Its usage story covers collecting evidence from mobile apps and profiling.
Testing utilities in our open-source toolkit for checking React Native behavior.
What you get
After handover, one QA engineer with the agent fleet covers work that previously took ten. Aim for 70% fewer manual testing hours as coverage grows.
Get faster PR feedback with automated checks, working toward a 1-hour turnaround instead of 8, alongside nightly walkthroughs and weekly exploratory tests.
Run the same flows across your agreed web, mobile, TV, and desktop platforms, aiming to find at least 95% of the defects human testers find.
Set goals in plain English. Checked actions replay on every commit without model calls, with agent repair available when your app changes.
Turn bug reports into evidence-ready tickets faster, aiming for 2 minutes instead of 30, with screenshots, recordings, logs, and reproduction steps in your tracker.
Keep the agents, product maps, test data, and regression tests in your environment. Documentation and hands-on training prepare your team for handover.
We build the setup in your environment and train your team to run and extend it, so you can cover more of your app with less manual testing.
DESIGN → BUILD → IMPROVE
We map your QA gaps, agree priority flows, and get the first automated checks running on your app.
We extend testing across agreed platforms, connect your delivery tools, and train your team to run the setup.
We improve reliability, reduce running costs, and add tests or agents through optional support after handover.
Case studies




AI-Native Test Automation
Tell us about your app and the testing work slowing your team down. We’ll connect you with a QA expert to discuss your setup and the next steps.
We’ll connect you with the right expert within 24 hours.
STILL HAVE QUESTIONS?
Yes. We connect to your current tools and test suites, keeping unit and integration tests as the fast foundation. Agents add exploration and replayable end-to-end checks, and we train your team to maintain and extend the setup.
Scripts check predefined cases. Agents can also explore your app, assess visual behavior and translations, and investigate journeys outside your scripted coverage. They return evidence for your engineers to review before making release decisions.
We build around the web, mobile, TV, and desktop platforms agreed in your scope. During migrations, checks can cover both existing and new app surfaces.
After a check confirms a flow’s outcome, the agent’s actions are recorded for replay on every commit without model calls. When a step breaks, an agent can repair and re-record it, or strict CI mode can stop the run. Live AI checks and repairs can still use a model.
The agents, product maps, test data, and regression tests live in your repositories, CI, and cloud. Handover includes documentation, runbooks, and training so your team can operate the setup. Ongoing support from our team is optional.
We agree on the price after discovery, based on the scope. Work can be billed for time spent, delivered for a fixed price, or tied partly to an agreed outcome. Infrastructure, model, and third-party tooling costs are separate; ongoing maintenance is optional.
We compare the agreed workflows with your starting point: manual QA time, feedback time, defects found, and coverage. We also track run time, running cost, maintenance, and the human effort needed to review findings.
We need access to the relevant repositories and CI, runnable test builds, test accounts and data, and a QA or engineering lead who can agree the first flows. We also agree which environments and model providers the setup can use.