EngProd Metrics

Tuneloop provides engineering productivity metrics purpose-built for AI-assisted development. Four headline metrics, each filterable by any dimension and expandable with drill-down charts.

Session Outcome Rate

Are AI agents actually producing results?

The share of sessions that ended in a shipped outcome. You control what counts as a successful outcome — pr_merged, ticket_resolved, or LLM-judged success.

Code Churn Rate

Is AI-written code sticking, or getting rewritten?

Measures the rate at which AI-authored lines are modified or deleted within a trailing 30-day window, compared against human-authored churn. A high AI churn rate relative to human churn signals that agents are producing code that doesn't survive review or real-world use.

Time to Ship

How long does it take for work to produce results?

Measures the time from first session to shipped outcome — merged PR, resolved ticket, or shipped feature. Tracks how your AI stack affects overall delivery speed.

AI Defect Rate

Are AI-assisted changes causing problems?

The share of AI-assisted artifacts (PRs, commits) that led to a defect within a configurable lookback window. Defects include reverts, bug tickets linked to the change, ticket reopens, and hotfixes. Compare against your baseline to understand whether AI-assisted changes are more or less reliable.

Filtering Dimensions

Every metric can be sliced by:

Model
Harness
Complexity (trivial, routine, substantial, open-ended)
Repository
Team
Custom tags