A GitHub-style pull request view, Fast-v3 by default, and per-plan review settings

Highlight
Featured this release
A GitHub-style view for pull requests
The pull request page reads like a GitHub conversation, with the latest QA.tech review inline.
The header shows the PR state (Open, Draft, Merged or Closed), the source and target branches, and a diffstat next to the Conversation and Files changed tabs. The description and latest review from AI QA testing sit in one timeline, with older reviews under "Previous reviews". The sidebar shows the review score, classification, a Low/Medium/High risk badge and a link to the repository.

More on this: PR testing and PR and CI/CD testing.
# Direct link
Update 02 of 04
Fast-v3 is the default web agent
Web tests that don't name an agent now run on Fast-v3, the recommended pick for new applications.
Fast-v3 is now listed first in the agent picker and shown as the QA.tech recommendation for applications without a default agent. Tests and applications already set to Fast-v2 stay on Fast-v2.

More on this: web testing and regression testing.
# Direct link
Update 03 of 04
OS Tools for every organization
Agents can read email and receive SMS through OS Tools in every workspace, with no opt-in.
Every browser agent can open and close OS Tools during a test, and the inbox is summarized for the agent up front. Settings → OS Tools shows Email and SMS inbox tiles. Fast-v3 uses OS Tools for email links, one-time codes and SMS.

More on this: web testing, authentication testing, SaaS onboarding testing and notifications testing.
# Direct link
Update 04 of 04
Review instructions and rerun budget per test plan
Tune how the review agent treats failures for each test plan, including how often it reruns a failed test.
Open a test plan and use the new Review tab:
- Review instructions – guidance on priorities, acceptable flakiness, what to ignore and who owns what. Pre-filled with a default and marked Default or Custom.
- Reruns per failed test – Off, 1, 2 (default) or 3. With Off, the agent classifies failures from the evidence alone.
Runs without a review chat, such as single tests and chat-started runs, still use the project's rerun setting.

More on this: regression testing and QA testers.
# Direct link

A GitHub-style view for pull requests

Fast-v3 is the default web agent

OS Tools for every organization

Review instructions and rerun budget per test plan

Your team moves fast. Can your testing keep up?