bg.
Projects / Release verification
Built at ScreenMeet

AI-driven release verification

Testing remote support against a real Windows endpoint.

A Claude-driven test rig I built to run a 14-feature verification matrix and reproduce customer-reported issues on demand.

ClaudeAWS Systems ManagerWindowsRelease verification

Why I built it

Supporting an enterprise remote-support platform means investigating behavior in the environment where customers actually use it. I built this rig to make a live Windows endpoint available for repeatable release verification and targeted issue reproduction.

My contribution

I built the Claude-driven workflow that operates a live Windows endpoint through AWS Systems Manager. It runs a 14-feature verification matrix on each release and can reproduce specific customer-reported issues on demand.

Workflow overview
  1. DefineRelease checks or reported issue
  2. DriveClaude-directed workflow
  3. ExecuteAWS Systems Manager
  4. ObserveLive Windows endpoint

What this demonstrates

  • Support-driven engineering: turning the need to reproduce real customer behavior into a working tool.
  • Applied AI: connecting an AI workflow to an execution environment and a defined verification matrix.
  • Systems integration: working across Claude, AWS Systems Manager, and Windows.
  • Verification design: supporting both repeatable release checks and focused investigation of reported issues.
See my ScreenMeet experience ↗