Cadence

The question it answers

"We're spending on AI tools and contractors. Is delivery actually getting better?" Most teams can't answer that, because the evidence is split across systems that don't talk to each other.

How it works

Cadence connects to Jira, GitHub and Anthropic and builds a single model of the delivery pipeline — from triage through code, review, merge and close. Tickets are joined to branches, commits and pull requests, and a data-quality panel shows exactly where that join is incomplete, so the numbers come with a confidence level.

Views

• Overview. Median cycle time, throughput, review turnaround, PRs merged and rework rate, each compared to the previous period; time spent per pipeline stage; and the work that has been stuck the longest.
• Teams and engineers. The same metrics broken down by team and person, including PR size and AI cost.
• AI usage. Token consumption and spend by model and engineer, and cost per shipped ticket.
• AI ROI. An aggregate-only stakeholder view with no individuals, measured from systems of record. It is explicit about what the data can and can't prove, and flags patterns like spend rising faster than throughput.
• Contractors. Blended labor-plus-AI cost against delivery output, sortable and exportable for finance.

Cadence is in development.