TL;DR Summary
Friction Shift
Collaboration → handoff + process gaps
Tooling Maturity
Major platform improvements shipped (Figma-to-code, inspect, public sharing)
New Tension
Prototype-first creates no Figma files — but eng handoff still requires them
Real Data, Real Signal
Piping in real data exposed product gaps that mocks wouldn't have surfaced
Biggest Ask
Async feedback collection on prototypes
Core takeaway: The tools are maturing fast, but the process around prototype-first — handoff, documentation, and feedback loops — is where the real design work now lives
Pilot Evolution
How the bottleneck has shifted each week
1
Week 1
Setup & tooling friction
2
Week 2
Collaboration & infra friction
3
Week 3
Handoff & process friction
Quant Snapshot
Effectiveness still split
3.0
out of 5 · range: 2–4
Split mirrors different project phases: one team near handoff, one still iterating
n=2 early responses — additional reflections may shift the picture
AI Comfort ↑ converging
4.0
out of 5 · range: 4–4
AI comfort converging upward — the tool learning curve is flattening
n=2 early responses — additional reflections may shift the picture
Interpretation
- AI comfort is no longer the blocker — both respondents at 4/5, up from Week 2's bimodal 2–5 spread
- Effectiveness variance now tied to project phase, not environment quality (the Week 2 driver)
- The team near eng handoff scored higher but hit process friction; the iterating team scored lower due to visual fidelity and sharing pain
Week 1: variance = onboarding. Week 2: variance = infrastructure. Week 3: variance = project phase and process maturity.
Key Themes — Week 3
Platform Tooling Shipped Fast
- /figma-implement-design skill for tighter Figma-to-code workflow
- HTML inspect tool: select a region, copy HTML, paste into Claude
- Project details panel with problem/hypothesis and QR codes
- Public sharing via /pilot-deploy to GitHub Pages (no login needed)
Erik's improvements directly addressed Week 2's top pain points — fast feedback loop in action
The Handoff Gap Emerged
- Prototype-first produced no Figma files — but eng handoff requires them
- One team scrambled to recreate Figma designs after alignment moved fast
- Prototype was the alignment tool, but not the handoff artifact
Prototype-first needs a clear answer to: what does eng receive when it's time to build?
Real Data Is Exposing Real Product Gaps
- Team uploaded actual category spreadsheets and RF questions into prototypes
- Piping in real data surfaced ideas "not fully thought out" in abstract planning
- Led to a product pivot — initial plan introduced too much customer friction
This is the prototype-first thesis in action: learn faster by making it real sooner
Async Feedback Is the Missing Layer
- Multiple requests for async feedback collection on prototypes
- Design studio revealed gap: no way to leave text comments on own time
- Praveen suggested embedding feedback directly in the prototype itself
Without async feedback, prototype sharing defaults to synchronous meetings — a scaling bottleneck
Roles Are Blurring Deeper
- "Lots of coding, trying to figure out how our project slots into NHX/NDFP — typically handled by a PM or TPM"
- People doing work outside typical roles: coding, project scoping, infra decisions
- Missing project documentation outside the prototype itself
Role blurring is still net positive but needs guardrails — especially around documentation and accountability
Visual Fidelity Still Hard
- Visual tweaks remain the most time-consuming part of prototyping
- New Figma-to-code skill and selected frames improving but not 1:1 yet
- Placeholder images poorly represent intent; real category images being added
Visual fidelity is a craft problem, not a tool problem — templates and starting points help most
In-the-Moment Feedback
What's Working
- Alignment around prototype direction happening quickly
- Real data in prototypes exposing product assumptions
- Platform improvements shipping fast based on feedback
- Public sharing via GitHub Pages (no auth required)
- Project details panel adding context without cluttering prototype
What's Breaking
- No Figma files for eng handoff when working prototype-first
- Visual tweaks still slow and time-consuming
- Getting feedback is still a learning process
- Project documentation doesn't live anywhere outside the prototype
- Some participants had limited time due to off-site / competing priorities
High-Value Signals
- Need async feedback collection directly on prototypes
- Need per-view problem/hypothesis details, not just project-level
- Need QR codes for sharing web prototypes on mobile
- Need clearer handoff artifacts (prototype → what eng receives)
- Need project documentation layer outside the prototype itself
eng handoff gap
async feedback
visual fidelity
project documentation
real data in prototypes
role blurring
Figma-to-code fidelity
competing priorities
Key Tensions
Prototype-first
No Figma Files
vs
Eng handoff
Needs Figma Files
Prototype-first skips traditional Figma design. But engineering workflows still expect Figma as the handoff artifact. Someone has to backfill — or we need a new handoff model.
Benefit
Speed to Signal
vs
Cost
Skipped Due Diligence
Moving fast with prototypes exposes real product gaps sooner. But it also means teams may skip the upfront due diligence they'd normally do before hi-fi design.
Prototype as
Alignment Tool
vs
Prototype as
Documentation
Prototypes are great for driving alignment conversations. But they don't serve as project documentation — problem statements, hypotheses, and decisions need a home too.
Expanding
Role Blurring
vs
Need for
Clear Ownership
People doing work outside their typical roles is a positive signal. But without clear ownership models, it creates confusion about who is accountable for what.
Emerging System Insight
The pilot is now testing the process, not the tools
Week 1 tested whether people could use AI tools. Week 2 tested whether the infrastructure could handle parallel work. Week 3 is testing something harder: what does a prototype-first workflow actually look like end-to-end — from idea to alignment to user signal to eng handoff?
- Handoff model — what replaces Figma files when prototypes lead?
- Documentation layer — where do problem statements, hypotheses, and decisions live?
- Feedback infrastructure — how do stakeholders give async input on prototypes?
- Due diligence sequencing — what upfront work should still happen before prototyping?
The question has shifted from "can we prototype?" to "what does the full product cycle look like when prototypes lead?"
Slack Channel Signals (#prototype-first-pilot)
🚀 Platform Improvements
- /figma-implement-design skill merged — selected Figma frames now produce tighter code
- HTML inspect tool: click any region, copy HTML, paste to Claude for targeted edits
- Project details panel with problem, hypothesis, notes, and QR code per project
- Public deployment via /pilot-deploy to GitHub Pages — shareable without login
- Real category images replacing placeholder images for better prototype fidelity
"Panel overlay addition is amazing! For subsequent versions, wonder if this could be used for Joe's use case in sharing project context up front" — Dave
💬 Feature Requests & Gaps
- Andre requested per-view notes (not just project-level) for different hypotheses per prototype
- Andre built standalone QR code feature for web prototypes viewable on mobile
- Praveen asked about embedding feedback collection directly in the prototype
- Craig built problem/hypothesis into the iframe itself because the panel wasn't available yet
- Discussion on whether tappable areas need visual highlighting for research
"Have also considered adding option to collect feedback in the prototype? This could be a big unlock" — Praveen
🔮 Emerging Patterns
- Teams self-building features they need (QR codes, hypothesis panels, feedback tools)
- Erik prototyping the prototype playground itself — "we're prototyping the prototype"
- Dave engaged hands-on, testing prototypes on mobile, exploring interactive messaging
- Valerie's research insight: hotspot highlighting not needed — "I always turn them off for research"
- Chris (returning from OOO): "really fun to see all the progress" — visible momentum
"Even being able to type into a focused state on mobile and have it respond was magic" — Dave
Week 4 Focus (Final Week)
Top Priority
01
Define the Handoff Model
The #1 open question. When prototypes lead, what does engineering receive? Resolve the Figma-file gap before the pilot ends.
- Can prototypes serve as the handoff artifact directly?
- What's the minimum Figma needed for eng specs?
- Document the recommended handoff path
02
Enable Async Feedback
Prototype sharing defaults to synchronous meetings without a feedback layer. This is the most requested feature.
- Explore lightweight commenting on prototypes
- Consider existing tools (Loom, annotation layers)
- Critical for research and stakeholder review scaling
03
Capture Pilot Learnings for Playbook
The pilot ends April 17. Start synthesizing learnings into the playbook that will guide EPAD rollout.
- What worked, what didn't, what needs investment
- Recommended workflow stages and decision points
- Role expectations in a prototype-first world
04
Add Documentation Layer
Prototypes are alignment tools, not documentation. Problem statements, hypotheses, and decisions need a place to live.
- Build on Erik's details panel — extend to per-view scope
- Connect to project documentation outside the prototype
- Consider what replaces PRDs in this workflow
05
Get Remaining User Signal
Teams should aim to get prototypes in front of real users before the pilot ends. Public sharing is now unblocked.
- Use /pilot-deploy for shareable research links
- At least one round of user signal per team
- Document what was learned and what influenced decisions