These guides describe the full platform, which opens in a few days. Today you can try Instant Chat.
All documentationWork with the app
AI Use Case Canvas: The 11 Sections to Fill In
- 1Frame the need
- 2Describe the proposed work
- 3Review value and challenges
The authored Canvas sections stay connected to the same opportunity.
The Agentic AI Canvas is an AI use case canvas with 11 authored sections, ordered so that later technical and value questions can use the business evidence established earlier. It is the primary builder and editor for the operating model.
The 11 authored sections
- Business Overview — the organization, market, customers, and operating context.
- Use Case — the specific repeatable process or decision being improved.
- Problem Statement — the current pain, its causes, and who experiences it.
- Goals — the objectives the initiative should advance.
- Target Outcome — measurable outcomes and success indicators.
- Key Stakeholders & Actors — owners, participants, reviewers, and affected groups.
- How AI Can Help? — appropriate AI contributions within the defined problem.
- Systems & Architecture — current systems, systems of record, integrations, and data flows.
- Challenges & Constraints — technical, operational, financial, regulatory, and delivery limits.
- Business Impact & ROI — expected impact, KPIs, cost assumptions, and value evidence.
- Agentic Workflow Implementation — triggers, actions, systems, human handoffs, exceptions, governance, and outcomes.
AI Readiness is computed
AI Readiness is no longer an authored Canvas section. The Brain computes it across data readiness, process readiness, decision clarity, stakeholder alignment, and governance readiness using evidence from the authored sections. Older Canvases may still contain a legacy readiness section, but new work should use the dedicated AI Readiness view.
Business Impact & ROI versus Agentic ROI
The Business Impact & ROI section is where the team writes what it believes: value drivers, baselines, KPIs, costs, benefits, and assumptions. Agentic ROI is the computed Brain view that tests those claims together with readiness and the rest of the Canvas.
Improving the section can strengthen the assessment, but filling the section does not guarantee a strong case. See Agentic ROI for its scoring and baseline rules.
Filled is not the same as supported
A filled section has content. A supported section has enough relevant evidence for the next decision. A long AI-generated draft may still be weak if its claims are inferred, stale, or missing owners and baselines.
Build mode reports how many authored sections contain content. AI Readiness and Agentic ROI report different forms of evidence confidence. Use all three signals rather than treating any one score as overall project truth.
Editing with Brain
Section chat focuses on the active section while retaining relevant Canvas context. Chat mode presents changes for individual review. Run Guided prepares one highest-impact proposal and waits for Apply Action — nothing changes until you accept it. Run Agent authorizes one bounded plan-improve-stop pass. See Using the Agentic Brain for the detailed approval model.
Review and sharing views
The Canvas hub also includes:
- Overview for reviewing the complete Canvas at a glance;
- Summary for a clean stakeholder-oriented rendering;
- Slides for a presentation projection; and
- Export for the available downloads, copy and print options, including audience-focused bundles.
These views project the same Canvas; they do not become independent sources of truth.
Choose sharing access deliberately
Where sharing is available, Public preview creates a read-only link to the results you select; recipients do not need an account. Review that selection before distributing the link. Invite people is a separate collaboration path: invitees sign in and receive view or edit access. A public preview is not an invitation to edit the Canvas. Choose the audience and access level for the review you need, and check generated content for freshness before handing it over.
Once the sections are filled in, see Agentic Outputs for the files and diagrams the Canvas produces.
Before filling it in, you can run the agentic AI readiness checklist on the workflow. The client intake worksheet is a short way to start, sources and memory explains how supporting material is reviewed, and the agentic AI implementation plan shows what to do with a finished Canvas.
Frequently asked questions
What sections does an AI project canvas have?
The Canvas has 11 authored sections: Business Overview, Use Case, Problem Statement, Goals, Target Outcome, Key Stakeholders & Actors, How AI Can Help?, Systems & Architecture, Challenges & Constraints, Business Impact & ROI and Agentic Workflow Implementation. They run in that order so later technical and value questions can build on the business evidence written earlier.
Is AI Readiness a Canvas section?
No. The Brain computes AI Readiness from the evidence in the authored sections, across data readiness, process readiness, decision clarity, stakeholder alignment and governance readiness. Older Canvases may still contain a legacy readiness section, but new work should use the AI Readiness view.
Does a filled section mean the project is ready?
No. A filled section has content; a supported section has enough relevant evidence for the next decision. A long AI-generated draft can still be weak if its claims are inferred, stale, or missing owners and baselines.
Can I share a Canvas with someone who has no account?
Where sharing is available, a public preview creates a read-only link to the results you select, and recipients do not need an account. Inviting people is a separate path: invitees sign in and receive view or edit access.
What goes in an AI use case canvas?
The business context first, then the people, workflow and systems involved, then the technical and value questions that depend on them, ending with agentic workflow implementation. The 11 sections above set out the order.
Is there an AI use case template I can use?
The 11 sections are the template: work through them in order and mark what you do not know as open. If you want a shorter starting point to send to someone else, use the client intake worksheet.
Next: Agentic ROI.