A CANVAS FOR YOUR THINKING
Connect ideas.
Keep the thread.
Bring chats, notes and sources onto one canvas. Let an agent research, connect and develop them with you.
YOUR IDEAS · THEIR SOURCES · THE NEXT QUESTION
- IDEA / 01A better café terraceStart with an observation.
- CHAT / 02“Too hot to sit here.”A clue from a conversation.
- SOURCE / 03Comfort & shadeFollow the evidence.
- QUESTION / 04What changes at noon?An assumption to test.
- DIRECTION / 05Measure local conditionsA next step to review.
An idea has room to grow.
- 01 / COLLECT
Bring it together.
Chats, notes, images and sources. Keep each fragment connected to its context.
- 02 / EXPLORE
Develop it with an agent.
Research connections, question assumptions and follow new directions. You set the scope and limits.
- 03 / SHAPE
Choose what comes next.
Review the evidence. Refine, test again or prepare a specification for your builder.
Follow a question.
See what supports it.
Your agent searches sources and develops the topic within your chosen limits. Findings, questions and decisions stay connected to their evidence.
Keep what helps. Challenge what doesn’t. You decide the direction.
Explore the canvasA little less guessing.
A better place to sit.
A local comfort monitor for small cafés.
A clearer idea.
A useful handoff.
Turn selected research, evidence and decisions into a final package: the scope, deliverables, instructions and checks your work needs.
Review the files, assumptions and open questions. Export a handoff for your builder, research team or service operators.
See example packagesResearch in any direction.
Create a package you can use.
Bring the work together as a scientific, software, hardware, hybrid, financial or services package. Explore what each can contain.
Give it a
physical form.
Define parts, dimensions and connections. Generate drawings with a connected CAD runner.
- 01CAD source & drawings
- 02Electronics specifications
- 03Bill of materials
Hardware exports depend on the connected tools.
Explore hardware examplesComponents → connections → schematic.
Give your builder
a clear start.
Export the reviewed scope, requirements and checks. Your building tool implements the package.
- 01Architecture & interface contracts
- 02Module PRDs & acceptance checks
- 03Build plan & agent instructions
package.md
agents.md
prd/dashboard.md
prd/weather-feed.md
prd/comfort-model.md
A little shade.
A lot more comfortable.
Make the question
testable.
Create a research protocol from your evidence: what to test, how to measure it and how to interpret the results.
- 01Research question & study protocol
- 02Measurement & analysis plan
- 03Evidence, assumptions & review criteria
Does shade change
comfort at the table?
Compare shaded and exposed tables under the same observation schedule.
- 01 / QUESTION
- Local conditions and reported comfort.
- 02 / METHOD
- Record temperature, exposure and observations.
- 03 / ANALYSIS
- Compare the groups. Document uncertainty.
Bring the parts
into one system.
Define how hardware, software and services work together. Create a coordinated package with shared interfaces and integration checks.
- 01Hardware & software specifications
- 02Interfaces & integration plan
- 03System checks & operating handoff
A sensor. A dashboard.
One working service.
- 01 / HARDWARETerrace sensor
Parts, electronics and measurements.
- 02 / SOFTWAREComfort dashboard
Data contracts, views and acceptance checks.
- 03 / SERVICEInstallation & support
Setup, maintenance and responsibilities.
Make the assumptions
visible.
Create a financial decision package with sourced costs, capacity assumptions and scenario ranges. See what changes the case before committing.
- 01Cost & unit economics model
- 02Scenario ranges & break-even analysis
- 03Sources, assumptions & decision brief
What would make
the terrace viable?
A model built around the assumptions you can inspect.
| Driver | Conservative | Base | Upside |
|---|---|---|---|
| Seats used | Lower | Expected | Higher |
| Setup costs | Source-linked estimates | ||
| Break-even | Test against each scenario | ||
- MODEL INPUTS
- Costs · capacity · demand assumptions
- DECISION BRIEF
- Tradeoffs · sensitivities · open questions
Define the service.
Prepare the handover.
Turn the customer journey into an operating package: who does what, how the service runs and what happens when something needs attention.
- 01Service blueprint & customer journey
- 02Operating procedures & staff scripts
- 03Capacity, service levels & launch runbooks
A better terrace.
Every working day.
- 01Set up
- 02Monitor
- 03Support
- CUSTOMER
- Installation → daily use → help when needed
- OPERATIONS
- Setup checklist · staff guidance · maintenance
- HANDOVER
- Owners · service levels · escalation steps
Every package keeps the selected evidence, assumptions and open questions together. Review the deliverables, then export the handoff.
Your thinking.
Your choice.
- Choose your model.
- Connect OpenAI, Anthropic or Gemini, a supported Claude Code plan, or managed access where available.
- Know the cost.
- The canvas is free. Commissioned jobs cost credits. Your provider bills your own model access; managed model usage costs credits separately. Review the price and spending limit before starting. See pricing.
- Your work stays private.
- Investigations, sources and notes stay in your account. Terms & privacy.
START WITH A QUESTION
What’s on
your mind?
One fragment is enough. See where it leads.
What does the agent do?
It researches sources, connects ideas and challenges assumptions within your scope and spending limits. You review its findings and choose what to keep.
Can I build from this?
Where package generation is enabled, create and export packages for software, hardware, hybrid products, scientific work, financial decisions and services. Each brings together the relevant specifications, evidence and handoff instructions. Review the assumptions and open questions before using it. Hardware outputs depend on the configured tools.
Where do I start?
Bring a question, a note or an image. Follow the documentation or explore the cookbook.