AI-Driven QA Automation Agents

I want to replace my manual QA flow with a chain of autonomous agents. The idea is simple: feed each agent everything we already have in Jira (acceptance criteria, user stories, existing defects), add the latest sitemap and a persistent knowledge store of the application, and let the agents turn that information into working Playwright tests coded in Python. A human reviewer will still approve the generated workflow, but from that point on the agents should take over script creation, execution, reporting, and defect logging.

Scope of work
1. Workflow & test-case generator
 • Connect to Jira’s REST API, read the acceptance criteria, combine it with the sitemap and knowledge base, and output a structured list of test cases.
 • Flag unclear items for human review before moving on.

2. Script builder
 • For every approved test case, generate Playwright code in Python.
 • Store scripts in our existing repository, following our naming conventions and page-object pattern.

3. Executor & reporter
 • Run the suite on Windows build agents, capturing logs, screenshots, and videos.
 • Summarise results in a markdown/HTML report and attach artifacts.

4. Defect creator
 • Automatically open or update Jira defects for any failed step, linking back to the evidence.

Acceptance criteria (done = agent works without manual help)
• End-to-end flow completes with one button or API call.
• All generated scripts pass linting and Playwright’s own test runner.
• Reports appear in Jira as a comment on the related story.
• Defects include reproducible steps and attachments.

Please keep the implementation modular so we can extend to Linux or macOS runners later. If this sounds like something you can deliver, let’s talk timelines and milestones.