What This Means
Most design teams bolt AI onto existing processes and call it innovation. That creates tool sprawl, inconsistent outputs, and no measurable improvement. I restructure the workflow itself, embedding AI where it eliminates friction and replacing manual steps that no longer justify their cost. The result is a design operation that runs faster, scales without headcount, and produces consistent output across distributed teams.
How I Approach It
Map the workflow end-to-end
Map current design workflows end-to-end, identifying manual bottlenecks, redundant handoffs, and steps where AI tooling delivers immediate ROI
Make vibe coding a production practice
Introduce vibe coding as a production practice, not an experiment. Designers describe intent in natural language; AI generates functional UI, components, and prototypes that feed directly into the build pipeline
Deploy agentic build systems
Deploy agentic build systems where AI handles component generation, documentation, QA checks, and design-to-code translation autonomously
Redesign legacy workflows
Redesign legacy workflows to integrate AI tooling without requiring team restructuring or new hires. Existing roles absorb new capabilities through structured enablement
Establish governance
Establish governance frameworks that define where AI outputs require human review and where they ship directly
When You Need This
- Your team is using AI tools individually but there's no standardised workflow, no governance, and no way to measure impact
- Design cycle times haven't improved despite adopting new tools because the underlying process is unchanged
- You're scaling output but can't justify proportional headcount increases
- Handoffs between design and development remain manual, slow, and error-prone
- Leadership is asking for an AI integration strategy and you need something concrete, not a slide deck
Expected Outcomes
- Design cycle times reduced by restructuring workflows around AI-assisted prototyping and automated component delivery
- Repeatable agentic pipelines that handle documentation, QA, and design-to-code translation without manual intervention
- Clear governance defining AI tool usage, review gates, and quality thresholds across the team
- A measurable reduction in time spent on production tasks, reallocating design capacity toward strategic and research work
- Team-wide adoption through structured enablement rather than ad-hoc experimentation
Expected Challenges
- AI-generated outputs require calibration. Early iterations will produce inconsistent quality until review gates and prompt standards are established. Budget two to three sprint cycles for tuning before expecting production-grade output
- Designers accustomed to manual workflows will resist adoption if AI tooling feels imposed rather than earned. Structured enablement works; mandates don't. Expect a transition period where parallel workflows run simultaneously
- Governance gaps surface quickly. Without clear rules on where AI outputs require human review, teams either over-review everything (negating speed gains) or under-review (shipping defects). Getting this boundary right takes deliberate iteration
- Tool fragmentation is a real risk. The AI tooling landscape shifts monthly. Committing too early to a single stack creates lock-in; staying too broad creates chaos. The strategy needs built-in flexibility with defined evaluation cycles
- Measuring impact is harder than it looks. Cycle time reduction is straightforward to track; quality improvement and creative output are not. Metrics frameworks need to account for both quantitative and qualitative signals from the start
What I Bring To This
At Johnson Controls, I directed 39 designers across multiple regions and cut operational costs by 28% through workflow restructuring. At SwissRe, I reduced design defects by 37% through systematic process redesign. I now apply the same operational rigour to AI integration, using vibe coding and agentic tooling in my own consultancy practice daily. This isn't theoretical. I build with these tools, ship with these tools, and know where they break.