– AI Feature Onboarding for DevOps

Making Intelligence Visible: Designing for Discoverability in CI Workflows

Role: Lead UX Designer

Duration: Feb – May 2024

Team:  PMs, Frontend Engineers, ML Engineers

Tools: Figma, Hotjar, Amplitude

Background & Challenge

Harness CI markets itself as the world's fastest CI platform, delivering builds 8x faster than competitors. This performance edge relies on four AI-driven "Intelligence" features that optimize builds automatically.

However, user research revealed a critical adoption gap:

Problem Statement

How might we surface high-impact CI intelligence features without disrupting established workflows, while designing a scalable onboarding model that adapts to different user roles and evolving functionality?

SOLUTION

  • Triggered by execution summary
  • Concise panel for enablement
  • No forced page jumps
  • Independent, non-blocking flow

design process

I  used a systematic two-phase approach to balance ideal user experience with enterprise implementation realities:

design process

Design Research
We conducted extensive research on onboarding patterns, prototyped them within our product context, and evaluated each against three criteria: clarity, user comprehension, and scalability.

design process

Evaluation Matrix
Rather than choosing a single pattern, we synthesized V2's scalable architecture with V3's superior clarity and comprehension into a hybrid approach.

design process

CORE PATTERN
Our hybrid solution combined the best of both approaches—triggered contextually by execution summary for consistent access, while remaining non-intrusive and respecting user workflow autonomy.

design process

CONSTRAINT ANALYSIS
To refine this pattern, we conducted in-depth constraint analysis, examining interrelated factors that directly impact user experience.

design process

PATTERN VARIATION
To address these constraints, we explored four design iterations, each prioritizing a different approach: information clarity, configuration-first, hierarchy-based organization, and role-aware adaptation.

design process

EVALUATION MATRIX
Each solution was scored against the six constraints in a weighted evaluation matrix, with points allocated according to constraint criticality.

design process

CHAMPION
The champion goes to our role-aware design, which hits great balance between information clarity, flexibility, and scalability.

impACT & TAKEAWAY

Enterprise feature discovery requires persistent, contextual access to information—not just first-run onboarding. By respecting user autonomy while providing scalable nudges toward adoption, we created a template for introducing AI features within mission-critical workflows. And users loved it!