Engineering with Claude across Hometap’s development lifecycle

Product Engineering
Engineering with Claude across Hometap’s development lifecycle

Challenge

  • AI adoption on Hometap's engineering team started with structured evaluation and A/B testing, but consistently comparing results was still difficult without one common, sanctioned assistant.
  • Documentation, code authorship, and code review depended on individual practices rather than a single standardized workflow, making it hard to scale good habits across the whole team.
  • Extending consistent AI-assisted practices across both software engineers and QA, not just individual contributors, remained out of reach without an officially adopted tool and set of practices.

Solution

  • Proxet engineers worked within Hometap's engineering team on their AI maturity journey, from early exploration through structured assessment and A/B testing, to a fully standardized, team-wide approach.
  • Hometap named Claude Code the official AI assistant, with Proxet supporting the determination of a more formal set of AI-use policies and best practices.
  • Developers began pairing their discovery pass with Claude — working from the terminal or the app — with human in the loop steering and model selection.
  • Built a subagent model delegated from a single main Claude thread so multiple tasks run in parallel instead of sequentially.
  • Built custom skills to give Claude-generated code a first automated review before it reaches a teammate.
  • Implemented Claude to review colleagues' code directly, with a human always making the final call. A dedicated agent writes end-to-end tests automatically once a ticket is closed.

Outcome

  • 70% faster bug resolution, with mean time to fix dropping from over 8 days to 2 days.
  • Nearly 2x more technical debt resolved: refactoring and code-health work rose from 58 to 115 issues addressed.
  • Claude Code established as Hometap's official AI assistant for its software engineering and QA organization.
  • Specialized subagents let the team parallelize work across frontend, backend, architecture, and QA tasks from a single main thread, rather than working them one at a time.

Join the growing list of brands that successfully work with Proxet!