QA.tech

Visit Website

Introduction

QA.tech is a hosted AI testing platform that calls itself “the verification layer for the AI software factory”. According to the official site, it maps your application, analyses the blast radius of a change, and verifies the user journeys that change affects.

Its agents run dynamic tests against a live web or native mobile app on every pull request, exercising it visually like a real user and reporting what works before merge. QA people teach the agent journeys and rules; engineers get a per-PR review; DevOps can trigger it from CI or query it from agents over MCP.

Features

1. PR verification: Agents test the deployed preview of each pull request and report results before merge.
2. App mapping and blast radius: The site says QA.tech maps your app and works out which journeys a change touches, rather than running everything every time.
3. Teachable agent: QA teams add user journeys, test users, knowledge and rules. The in-product chat can set up things like an admin test user to unblock tests.
4. Web and mobile: Listed surfaces include web, mobile web, iOS and Android apps. Mobile testing is on the Growth plan and above.
5. Exploratory testing: Exploratory and dynamic PR testing are included from the Starter plan.
6. API, voice and i18n testing: API testing, voice testing and geo/i18n testing are listed on Growth and Enterprise.
7. MCP, API and CLI: The pricing table lists MCP, API and CLI access on every paid plan, so coding agents can query results.
8. Coverage reports: Coverage and performance reports are listed from Growth upward.

Pros

1. Free start: A $0 trial plan with 10 credits and one application lets you test the loop on your own app.
2. Change-aware: Focusing on affected journeys keeps runs relevant as PR volume grows.
3. Agent friendly: MCP access means a coding agent can read verification results without a human relaying them.
4. Broad surface list: Web, mobile, API and voice on higher plans reduce the number of separate tools.
5. Budget controls: Spending limits and budget alerts are listed on every plan.

Cons

1. Credit-based billing: Plans are measured in credits; it takes a few weeks of real use to predict monthly cost.
2. Mobile and API gated: Starter covers web only. Mobile, API and scheduled runs need Growth, which is custom priced.
3. Hosted agents: Tests run on QA.tech’s service against an environment it can reach.
4. Agent verdicts need review: An AI agent’s “works” is evidence, not a guarantee; important flows still deserve deterministic assertions.
5. Short retention on Starter: Test data is kept 30 days on Starter, 90 on Growth.

Usage and Links (Example)

A typical setup from the official site:

1. Sign up on the free plan and add one application URL.
2. Connect your repository so each pull request’s preview deployment is tested.
3. Teach the agent key journeys, test users and rules.
4. Review per-PR reports, or let a coding agent read them over MCP.

See the docs for CI triggers, MCP setup and supported platforms.

Resources:

• QA.tech website
• Pricing
• Docs

Pricing

From QA.tech’s pricing page:

1. Free trial: $0. 10 credits, 1 application, web testing and PR testing.
2. Starter: $89 a month. 50 credits a month, up to 5 users, standard model access, up to 2 applications.
3. Growth: Custom. Custom credits, up to 50 users, mobile, API, voice and i18n testing, scheduled plans.
4. Enterprise: Custom. Custom users, fine-tuned model option, unlimited environments, faster runs.

Confirm current credit rules on the site before you buy.

Recommended for?

1. Teams with preview deployments per pull request who want an agent to check affected journeys before merge.
2. QA leads who want to define journeys and rules rather than write scripts.
3. Engineering orgs using coding agents that can consume results over MCP.
4. Not a fit for teams that need every check to run inside their own network without a vendor cloud.

Conclusion

QA.tech is an agentic PR verifier: it maps the app, picks the journeys a change affects and tests them on a live preview. A free trial and a published Starter price make it easy to evaluate.

Mobile, API and scheduled testing sit on custom plans, and agent verdicts still need human judgment on critical paths. Try it on one app with real pull requests before rolling it out.