AI Value Measurement with Apache DevLake

Project: AI Value Measurement with Apache DevLake

Situation

The client needed evidence for evaluating AI-assisted software development. AI-tool usage, delivery data, developer feedback, quality indicators, and cost information existed in separate places and could not be assessed as one evaluation.

Task

Complete a time-boxed measurement project that connected software-delivery data with AI-tool telemetry, developer surveys, privacy controls, and a cost model.

Action

Configured Apache DevLake and Grafana as the software-delivery measurement foundation
Connected GitLab, CI/CD, ticketing, and relevant delivery data
Added project-specific integration for AI-tool telemetry, survey responses, and cost inputs
Defined adoption, usage, delivery, quality, and cost measures with explicit interpretation boundaries
Mapped identities and teams across the participating systems
Added data-quality checks for completeness, freshness, duplicates, and unmatched identities
Defined aggregation, access, retention, and privacy requirements for team-level evaluation
Prepared dashboards, exports, documentation, and an effort estimate for later scaling

Result

Delivered a defensible technical foundation for evaluating AI-assisted development without treating tool activity, self-reported time savings, or delivery changes as interchangeable evidence.

Technologies

Apache DevLake, Grafana, GitLab, CI/CD integrations, Jira, AI-tool telemetry, survey data, cost modelling