Jellyfish adds tools to measure AI coding costs and productivity

Jellyfish announced in a press release new tools for measuring how AI coding products affect engineering work, productivity, and costs. The platform combines signals from engineering systems to compare human work, work assisted by AI, and autonomous agent activity.

Lifecycle Explorer shows how engineering time is distributed, while AI Cohorts tracks interactions with tools including GitHub Copilot, Cursor, and Claude Code. Metrics Explorer analyzes contributions from engineers and autonomous agents, and lets users create custom metrics with natural language. Jellyfish Assistant provides a chat interface for asking questions about engineering data.

The release also adds tracking for AI skill adoption, team behavior, token usage, and spending by tool and model. Spend attribution connects AI costs with initiatives, deliverables, and roadmap items, while cost benchmarks compare spending and outcomes with other companies using the platform. Jellyfish can also reconcile costs reported through APIs with costs calculated from usage telemetry.