AI-Assisted Design and Research Workflows
Industry
Enterprise AI
Client
Accedo
Platforms
LLM workflows, design systems, ResearchOps, custom MCPs
Date
2022 to present
CASE STUDY BRIEF
Capability
AI product strategy, workflow design, design systems
Leadership level
Director
What I led
Built AI-assisted design and research workflows at Accedo, including an LLM-powered design-system linter, knowledge agents, custom MCPs, and contribution flows that made quality checks and insight work faster without treating AI as a substitute for judgment.
Why it matters
Reduced team exploration and validation time by 25% while strengthening the quality and flow of design-system contributions.
DECISION RECORD
Problem and stakes
Design-system contributions and research workflows created repeatable quality and knowledge-management work that could slow teams down as the organization and product surface grew.
Role and scope
Created and introduced AI-assisted design and research workflows at Accedo, spanning contribution validation, intake flows, organizational knowledge agents, and custom MCPs for internal work.
Key decision and trade-off
Used AI for validation, routing, and knowledge retrieval while keeping human judgment in quality-critical decisions and contribution review.
Systems and artifacts
LLM-powered design-system linter, AI-assisted contribution intake, organizational knowledge agents, custom MCPs, AI-assisted research workflows.
Related content
At Accedo, I introduced AI-assisted design and research workflows to reduce the repetitive work around quality checks, contribution intake, knowledge retrieval, and research synthesis. The work included an LLM-powered linter for design-system contributions, AI-assisted intake flows, organizational knowledge agents, and custom MCPs for internal workflows. The key decision was to use AI where it could make a system easier to use and maintain, while keeping human judgment in quality-critical contribution and product decisions. Across the team, exploration and validation time fell by 25%. The work shares the mechanisms and outcome without exposing client or internal operational details.
