Evidence at a glance
Based on the linked sources. The author's reported result and independent confirmation are separate checks.
- AI involvement
- Reported models: Claude. Tools: Kokoro. The writeup below describes the work attributed to AI.
- Reported result
- Works. This is the directory's source-based outcome classification.
- Independent confirmation
- Not reported.
- Human work and limitations
- The entry summarizes a documented experiment in AI-assisted software development. It highlights the iterative nature of the build process and the identification of specification gaps as a key outcome of the testing phase.
Original sources
What was done
An experiment using Claude to build a sales commission calculator from a Markdown specification. The process successfully passed 24 acceptance checks and 48 automated tests, while identifying and resolving 10 specification gaps.
The author utilized Claude to reverse engineer a QuotaPath alternative by providing a single Markdown specification for a small-team sales commission calculator. The development was structured into four distinct phases: managing reps, plans, and deals; implementing payout math and statements; handling approval, locking, clawbacks, and payroll export; and creating rep-facing statements with a what-if calculator. According to the author, the resulting tool passed 24 acceptance checks in a real browser and 48 automated tests. The build process identified 10 gaps in the original specification, all of which were subsequently addressed. The author notes that while the tool functions, users should consult guidance on AI-built tool production readiness before relying on such software in production environments. The project documentation highlights that specific edge cases, such as deal cancellations after approval, initially caused calculation discrepancies that required spec refinement. The author explicitly distinguishes between the AI-built alternative and the commercial QuotaPath platform, noting specific scenarios where the paid service remains preferable.
Sources: Source 1