AI Week

Design Manager — rapid AI prototyping across CDRO and Unify surfaces

Role: Design Manager

product-designdesign-leadershipux-researchai

Problem

CDRO's feature complexity had a high cost: understanding pipeline status required significant product knowledge, and building new pipelines demanded deep YAML familiarity. AI Week was the opportunity to explore whether AI could lower that entry cost — making the product more accessible without changing its underlying capabilities.

Solution

Prototyped two AI-native concepts in a sprint format: BeeBot/ChatDash, an AI chatbot embedded in the CDRO dashboard for natural language querying of CI/CD status and reporting; and Pipelines AI Assist, a natural language interface for generating pipeline configurations. Both were sprint-quality prototypes — enough to demonstrate the concept and inform whether either thread warranted a formal product investment.

Rapid AI prototyping across CDRO surfaces

AI Week

AI Week compressed product exploration into a sprint format — no formal discovery, no spec gates, just teams building against a hypothesis. For CDRO, the question was straightforward: where does AI actually help in a CI/CD workflow? Two threads emerged: helping users understand what's happening across their pipelines through natural language, and helping them build those pipelines with less friction.

001

BeeBot / ChatDash — AI chatbot for CDRO reporting and CI/CD status queries

Problem

CDRO dashboards required manual widget configuration — users had to know which metrics existed and how to surface them. Getting meaningful pipeline visibility required significant product knowledge before the dashboard could deliver value.

Solution

A conversational AI layer that interpreted natural language requests — 'what was our release frequency this month vs last month?', 'add failed runs', 'create a DORA metrics group' — and built the corresponding widget layout in real time, without requiring users to navigate the widget library.

002

Natural language pipeline creation

AI Workflow Assist

The Pipelines AI Assist explored a different angle: using natural language input to generate pipeline configuration. Rather than building pipelines through a form-driven UI, users could describe what they needed and let the assistant draft a starting point. The concept demonstrated how AI could reduce the steep learning curve of CDRO's YAML pipeline model.

003

Part of the CloudBees selected work →