Project brain
Ingest briefs, zoning, and site intel into one structured surface—searchable and always current.
- Sync PDFs, emails, and notes
- Link every what-if back to sources
Cocoon Lab is an AI-native design studio rethinking design — starting with Cocoon.
AI copilot for early-stage architectural design.
Cocoon keeps briefs, rules, and AI options in sync so teams choose quickly.
Built with studios designing housing, institutional buildings, and civic spaces.
Demo (opens in new tab)
Powering forward-looking design teams
Studios shipping dense infill, campuses, and civic realm projects.
Problem
Early design work scatters fast.
Inputs fragment across emails, PDFs, and hallway chats before a concept even starts.
- Briefs, PDFs, and markup threads never stay in one place.
- Updates arrive piecemeal, so sketches drift and decisions repeat.
- Misalignment snowballs into days of rework before reviews.
Scattered briefs stall the first option.
We see hours vanish to chasing links, reinterpreting notes, and resyncing approvals before design begins.
Admin + file wrangling
45h
per 100 project hours lost before design
Misalignment rework
8h
per cycle syncing decisions
Concept time at risk
~30h
when context drifts between teams
What Cocoon does
One workspace that keeps early design calm.
Fewer clicks, fewer surprises. Guardrails stay visible while you explore.
Guided options
Generate zoning-aware concepts with constraints upfront. Swap assumptions without restarting.
- Live FAR and coverage checks
- Clear rationale for each envelope
Shared decisions
Comment, pin decisions, and export without digging through threads.
- Threads on briefs and options
- One-click packets for reviews
Demo
See Cocoon keep pace with complex sites.
Press play on your time. Slow connection? Grab the file below.
Prefer a download?
If streaming is slow, open the MP4 directly.
Outcomes
Aligned work moves faster.
Pilot studios trim review loops and keep early hours billable.
Pilot median
~63% faster concepting
Brief to first aligned option.
- Fewer alignment cycles
- Less unpaid early work
Use cases
Where teams run with Cocoon first.
Feasibility studies
Quickly test yield under multiple zoning interpretations and share clear recommendations.
Design reviews
Anchor crits around a shared brief and rationale instead of hunting through decks and emails.
Client workshops
Co-create options live with clients while Cocoon watches constraints and captures decisions.
Founders
People who ship with studios and cities.
Hands-on leaders across architecture, AI research, and urban policy.
Rashid Mushkani
CEO, AI & design researcher
PhD in AI & design from Mila, working at the intersection of architecture, design, and intelligent systems.
Hugo Berard
CTO, AI systems researcher
Built generative tooling with Mila teams and industry pilots.
Shin Koseki
Product, urbanist & professor
Advises cities on resilient public-realm design and policy.
Backed by research-grade AI
Now
Alpha with partner studios.
Beta opens in 2026 with teams already shipping work.
Status
AlphaAlpha with pilot firms. Beta lands 2026.
- Weekly drops with partner teams.
- Beta cohort forming for 2026.
Roadmap
Alpha
Live with pilot firms.
Beta
Opening in 2026.
North America
Expanding 2026–2027.
Global
Rolling out in 2027.
Roadmap
What we’re shipping with partners.
Shipping weekly with pilots; tell us what keeps your projects moving.
Now
- Private pilots with complex urban sites
- City-facing packet exports
- Improved zoning explainability
Next
- Revit + Rhino handoffs
- Daylight + access checks in-line
- Shared assumptions library across studios