
Engineering managers have no reliable, continuous way to assess developer performance. Code reviews are reactive, quarterly reviews are subjective, and raw git metrics (commit count, lines added) tell you nothing meaningful about code quality or team growth. Insignia solves this by connecting to your GitHub repositories and analyzing every pull request — the code diff, PR description, reviewer feedback, and CI/delivery signals — to generate structured, AI-powered assessments for each developer on your team. Each developer has an admin-configured profile: seniority level, tech stack, role, and custom context. Each project has its own profile too: architecture patterns, business objectives, naming conventions, and custom practices. Insignia uses these to make assessments contextual — a senior backend engineer is judged differently than a junior frontend developer, and against the standards that actually matter to your organization. Assessments are scored across three weighted dimensions: objective code quality (universal programming best practices), subjective fit (architecture compliance, business logic alignment, level appropriateness), and surrounding signals (PR quality, review outcomes, CI results). Each dimension is evaluated by IBM watsonx.ai Granite models in separate prompts, then synthesized into a final score, verdict (Stellar / Great / Functioning / Detrimental), and actionable improvement suggestions. Built with IBM Bob IDE as the core development partner and powered by IBM watsonx.ai Granite models on the backend — Insignia turns raw git activity into structured, fair, and actionable engineering intelligence.
17 May 2026