ai / software delivery system
Software Factory
A connected AI-assisted delivery system moving work from issue and incident to code, testing and release.
System designer & hands-on engineer · Developed at Powered By Rental
Working delivery system · Developed in a live SaaS business
AI enablement · Engineering & systems · Business & leadership
The situation
A small team needed to move product work through development, review, testing and release while maintaining a live service. Adding AI to isolated coding tasks would leave the handovers and QA bottlenecks untouched.
My contribution
I designed and personally engineered much of the Software Factory while leading Powered By Rental. It brings specialist agents, connected tools and human review into one delivery workflow. The rental product and commercial work are covered in a separate case study.
The work
Jira provides the work to investigate and Sentry provides operational signals. MCP-connected tools give specialist workflows the context to triage issues, investigate incidents, implement changes and review code. GitHub Actions and layered testing support validation and release. Automation is selective: people review changes, resolve ambiguity and test where confidence is insufficient. Failures have shaped the boundaries, rather than being treated as exceptions to hide.
The result
In one ten-day sprint, two engineers and one QA specialist tested and deployed around 250 tickets, against an estimated previous baseline of around 50. This is a reported example rather than an audited productivity measure. The system continues to evolve.