About
Monali Samarth is a UX Lead and Design Systems Architect with 22+ years bridging product strategy, design, and front-end engineering across enterprise SaaS, edtech, fintech, and healthcare. At RIB Software, she architected a tokenized component library that reduced UI inconsistencies by 80% and directly enabled AI-assisted design-to-code pipelines across five products. She has personally shipped the systems this talk is about and lived through the failures that inspired it. Monali mentors designers and developers on design systems, accessibility, and scalable UX ops, and brings rare dual fluency in Figma and Angular to every handoff conversation.
Workshop details
The last mile of UX - protecting design intent in production
About this workshop
This workshop started from a recurring frustration: designers invest heavily in craft, then lose control of the output the moment it leaves their hands. The arc moves deliberately from awareness (where intent breaks) to skill-building (how to prompt AI precisely) to diagnosis (red-teaming real outputs) to systems-thinking (building personal guardrails). The three labs are the load-bearing structure without them, this is just another talk about handoff problems. Key trade-off: 2.5 hours is tight for the ambition. Phase 2 (COAT framework) can run long if the group is new to structured prompting; facilitators should time-box Lab 1 firmly at 15 minutes or Phase 4 gets compressed. The red-team lab (Lab 3) consistently produces the most insight but needs pre-prepared broken designs don't improvise these on the day.
Workshop key takeaways
- Diagnose exactly where design intent erodes in your pipeline from handoff notes to implementation using a shared vocabulary for the six most common production failure modes.
- Prompt AI design tools with precision using the COAT framework (Context, Outcome, Audience, Tone), producing outputs that are specific enough to survive the handoff without human translation.
- Red-team any AI-generated design output against a 15-point production readiness checklist covering contrast, component states, design tokens, copy edge cases, and responsive gaps.
- Design and document your own AI-augmented workflow, with clear boundaries between what AI owns, what gets reviewed together, and what stays in human hands.
- Build the team agreements and review gates that protect design intent in production not as a policing mechanism, but as shared professional practice.