IBM AI Hackathon · Track 3 · Rule of Three
DesignFlow.
A living brain inside Figma.
An AI-powered design workflow agent built as a Figma plugin — bringing comments, tasks, status, and design decisions together in one place.
- Project
- DesignFlow — AI Figma plugin
- Context
- IBM AI Hackathon · Track 3, Design Deliverables
- Team
- Rule of Three — Samriddhi Vijay, Rujuta Sanghvi, Sharvani KB
- My role
- Product Design · Research · Agentic behaviour design
- Stack
- IBM ICA 2.0 · Figma API · IBM Bob & Claude · Carbon UI
- Duration
- Hackathon sprint
- Result
- Top 6 selected teams at IBM AI Hackathon
Designers spend more time managing work than actually designing.
Design collaboration is scattered across disconnected tools — designers, developers and PMs each track progress, feedback and decisions on separate platforms.
Tracking
Manually recreating tasks & user stories from Figma inside Azure or Jira.
Progress
Design feedback gets buried in comments, causing repeated stand-up discussions.
Collaborating
Constant back and forth with devs and PMs who need context throughout.
Documenting
Designers manually document decisions so everyone understands the rationale.
“Designers spend hours organising files, translating work for developers, updating managers, and later trying to remember why decisions were made.”
Maya · Senior UX Designer, IBM — from 6 designer interviews
An agent that observes, reasons and suggests — with humans in control.
The loop: the agent observes the file → reasons → suggests → the human confirms or declines → the plan updates on the canvas. The agent constantly watches and learns.
01
Intent
The real goal, not the surface request — centralise collaboration in Figma and preserve rationale.
02
Observation
The agent reads frames, sections, comments, previous planners and prototyped flows.
03
Reasoning
It estimates stories, screens, timelines and milestones, and detects missing requirements.
04
Suggestion
Nothing is applied silently. Every plan change is proposed for confirmation.
05
Human control
Approvals and “is this done?” stay with people. The designer always stays in control.
One living source of truth, right inside Figma.
User stories become the primary object in navigation — each flow carries its frames, tasks, comments, status and the “why” behind every decision.
The impact
Built in a hackathon sprint. Validated like a product.
84%
AI precision — agent recommendations accepted with minimal intervention
71%
Faster design delivery & task completion for designers
68%
Average time saved for developers
72%
Faster understanding of design status for PMs
4.3/5
Average user confidence & satisfaction score
91%
Participants eager to deploy the tool organisation-wide