AI Impact
AI tool adoption & measurement
Scrum Master Guide: Measuring AI Impact
AI Impact measurement answers three questions: Are teams actually using AI tools? Which tools deliver measurable results? And where should the organization invest next? Jellyfish automatically detects AI usage patterns across GitHub Copilot, Cursor, Claude Code, and other tools — then links adoption data to delivery signals like throughput, cycle time, and code review speed. The goal is objective measurement, not advocacy for any specific tool.
AI Tool Adoption by Team
AdoptionBefore / After AI Adoption
Impact| Team | PRs/wk Before | PRs/wk After | Cycle Time Before | Cycle Time After |
|---|---|---|---|---|
| Platform | 18 | 24 | 4.2d | 3.1d |
| Frontend | 22 | 31 | 3.5d | 2.4d |
| Data | 12 | 16 | 2.8d | 2d |
Green values indicate improvement after AI tool adoption. Cycle time is measured in days from ticket start to merge.
Supported AI Coding Tools
Note: AI Impact data is generated at the platform level by Jellyfish. The Export API v0 does not include AI-specific endpoints — use the Jellyfish dashboard for adoption and impact analytics.
Using AI Data in Ceremonies
If a team adopted an AI tool 2–3 months ago, pull their before/after delivery metrics. If cycle time hasn't improved, investigate whether the tool is being used effectively, whether other bottlenecks dominate, or whether the team needs enablement support.
Factor AI tool impact into capacity assumptions. A team with high Copilot adoption may have higher throughput than historical baselines suggest — adjust commitments accordingly.
Present adoption rates alongside impact data. High adoption with no measurable improvement is a signal to investigate — not to roll back the tool.
Customer outcome: TaskRabbit — teams shipping code faster and delivering twice the value in half the time (jellyfish.co/platform/jellyfish-ai-impact/).