Manulife Hackathon · Case study
Design work,
with a copilot.
I explored how AI could make the design workflow easier: helping designers review a screen, understand the evidence behind a suggestion, and turn a decision into clear next steps.
2nd place · Manulife Hackathon
- My role
- Product design lead
- Team
- Individual Contributor
- Timeline
- 6 Weeks
- Format
- AI-powered Figma plugin
- 01Design review in context
- 02Evidence you can inspect
- 03Human led decisions
- 042nd place at the hackathon
Prototype status
Synthetic examples and deterministic mock responses; no live model or enterprise data integrations are claimed.
The challenge
Good design decisions take more than a good eye.
Design feedback, customer research, reusable patterns, and delivery tickets often live in different places. Designers spend time reconnecting the context and teammates may not see what a recommendation is based on.
01
Feedback without context
A comment on a screen does not always explain the customer problem or why it matters.
02
Research is hard to bring into the moment
Designers can lose time searching for the finding that supports or challenges a decision.
03
Known patterns are easy to miss
Teams may solve the same interaction more than once when existing patterns are hard to discover.
04
Handoffs lose the “why”
When reasoning is separated from the task, delivery partners have to reconstruct the intent.
“Could an AI copilot help me understand what to improve, why it matters, and what to do next without taking the decision out of my hands?”
The question behind the concept
The concept
One helpful loop, right inside Figma.
Design Evidence Copilot is a side panel concept that connects a screen to a recommendation, its supporting evidence, a relevant pattern, and a reviewable handoff.
Designed for
Built for the people shaping better products.
Designed for the broader team, the copilot connects design decisions to supporting evidence, reusable patterns, and clear next steps.
Product designers
Run audits and make evidence-backed design decisions.
UX researchers
Bring traceable findings into the design context.
Design-system teams
Identify consistency gaps and reusable patterns.
Design operations
Improve governance, traceability, and reuse across teams.
Product managers
Connect prioritized user impact to roadmap decisions.
Developers
Receive clearer acceptance criteria and design context.
AI with a human in the loop
Helpful suggestions. Visible reasoning. Designer in charge.
The product idea puts AI to work on tedious parts of a design review surfacing possible issues, finding relevant context, and preparing a first draft while leaving interpretation and approval with the people closest to the work.
AI can assist
Bring the pieces together
- Spot possible clarity, usability, accessibility, or trust issues.
- Retrieve approved research that may support or challenge a finding.
- Find comparable patterns and explain the match.
- Turn a reviewed recommendation into an editable ticket draft.
People stay in control
Trust is part of the workflow
- The designer decides whether a finding applies.
- Evidence is available to inspect before acting.
- Recommendations and ticket copy remain editable.
- Any future Jira creation requires a person’s approval.
Prototype status: The demo uses synthetic examples and deterministic mock responses to communicate the AI assisted workflow.
The experience
From “something feels off” to a next step.
The sample walkthrough uses a coverage selection screen. The copilot flags unclear trade offs, opens synthetic evidence, points to a reusable pattern, and prepares a draft ticket.
Product leadership
I made the workflow and the guardrails part of the product.
A lead level solution is a clear decision about what to build now, how to earn trust, and what must be true before scaling.
Decision 01
Prove one complete journey
Focus the scope on frame → finding → evidence → pattern → draft. A coherent experience is easier to learn from.
Decision 02
Make provenance visible
Place evidence metadata beside the recommendation so users can judge relevance and confidence.
Decision 03
Keep the prototype dependable
Use synthetic seed data and deterministic services, with future integrations behind clear boundaries.
Decision 04
Keep consequential actions reviewable
Keep recommendations editable, tickets in draft, and designers in charge of the final decision.
2nd
Manulife Hackathon · The concept earned second place.
How it was shaped
A lightweight build with a path to real integrations.
The concept architecture separates the Figma interface from services and data. That keeps the demo simple while giving a future team a sensible place to add approved AI and enterprise connections.
Figma Plugin API
Selected frame context and side panel experience.
TypeScript + React
Typed interaction states and focused interface.
Vite + mock services
Packaged UI and predictable synthetic responses.
From hackathon to enterprise
A workflow worth exploring across Manulife.
The long term opportunity is an evidence backed design capability that can scale across Manulife starting with governed pilots, then learning what should be shared across Manulife Insurance and Manulife Bank.
Shared workflow. Business line context.
Manulife Insurance
Connect the workflow to approved insurance research, patterns, permissions, and domain specific guidance.
Manulife Bank
Apply the same core experience to banking journeys, grounded in Bank’s own approved sources and review practices.
Hackathon prototype
Figma workflow, synthetic findings, evidence cards, pattern suggestions, ticket draft.
Business line pilot
One approved AI service and permission aware evidence source. Evaluate relevance, trust, and time saved.
Manulife wide capability
Shared platform controls for governance, security, source ownership, and model evaluation.
Reflection
The best copilot helps people think not just move faster.
This project reinforced that useful AI depends on the experience around the model: suggestions need context, evidence must be checkable, and people need a clear say in what happens next.
Second place was a milestone. The next proof would be adoption: test the workflow with designers, then pilot one governed evidence source before scaling across Insurance and Bank.
Feedback
“Well deserved recognition for Fenil, whose PIXEL PILOT solution earned 2nd place at the AI x Craft Hackathon. What stood out was the practical application of AI to a real challenge in our design process: turning UX audits and research into structured insights, recommendations and reports while keeping human judgment and accountability at the centre. The idea demonstrated how AI can reduce time spent documenting findings and increase time where it matters most, acting on them. A well executed solution with clear potential to help teams work smarter with a focus on high value work. Well done!”
“Congratulations on achieving 2nd Place at the AI x Craft Hackathon! This recognition celebrates the team's creativity, problem solving, and dedication in delivering a strong and impactful solution. Thank you for contributing your talent and energy to making the event a success.”
“Congrats Fenil on a job very well done! Your concept focused on a real pain point in the design workflow and one that particularly resonated with me due to its strategic implications! Well done!!”
“Congratulations on earning Second Place in the AI x Craft Hackathon Challenge! Your submission stood out for its creativity, strong execution, and clear demonstration of AI's potential to solve real world problems. As a member of the judging panel, I was impressed by the quality and impact of your work. Thank you for bringing your talent and innovation to the competition. This recognition is well deserved, and I look forward to seeing what you create next.”
“Congratulations to Fenil on earning 2nd place at the AI x Craft Hackathon for PIXEL PILOT. Design Evidence Copilot demonstrated a thoughtful approach to applying AI in the design process by transforming UX audits and research into structured findings, ratings, recommendations, and reports while maintaining human review and accountability. The judges appreciated the practical value of the solution and its potential to help teams spend less time on reporting and more time acting on insights. Thank you for bringing such an innovative and well executed idea to the event.”
“Congratulations to Fenil on taking 2nd place in the AI x Craft Hackathon! An incredible achievement in building an innovative Figma plugin that transforms UX audits and research into actionable findings, recommendations, and reports, helping designers move from insight to action faster. Congratulations on this well deserved recognition!”
“Congrats to Fenil on his work and presentation at the AI x Craft Hackathon!! I appreciated the thoughtfulness to the UX and accessibility audit use case, feature to explain the why behind severity and allow designers to accept, refine, or dismiss, and generate report. Grateful for the contribution of innovative ideas to the design community!”