Most AI lists could apply to anyone. Use Case Foundry builds a graph of your company's data, skills, and constraints — ranks defensible opportunities by Impact, Feasibility, Moat, and Quality — and adapts the same portfolio to your stage, mandate, department, data readiness, peer cohort, and who reads the output.
The recommendation board updates as you edit — Analyze needs no API key. Save models to the server when signed in. Generate full plan and dossiers when you're ready.
Workshops and ChatGPT produce long lists of plausible ideas. Almost none explain why you would win — or what to validate first.
The app is one workspace with three steps. Deterministic graph reasoning ranks candidates; the LLM writes plans and dossiers when you ask.
Load an example, auto-map a website, or open a saved model. Set interview focus — department pack, data readiness, peer cohort — plus archetype and resource level. Optional sign-in keeps models private on the server.
Build your evidence map at your pace: offerings, segments, processes, assets, skills, projects, pains, and constraints. Quick-add only what you know; sections appear as you go. Operator coverage shows what's firing vs dormant.
Tune feasibility vs moat and set strategy mandates. Analyze (no API key) returns grouped candidates, operator coverage, and Improve the model gaps. Generate full plan adds LLM write-ups, a strategic bet, and optional dossiers.
Adaptation layers sit on top of the durable company model. They change how opportunities are elicited, ranked, and framed — never what facts you entered. Graph truth always wins.
Startup, scaleup, enterprise, agency, or SMB — plus resource level. Reorders builder sections, profile-aware gap diagnostics, and which operators are relevant for your stage.
Near-term and strategic mandates (cost, productivity, innovation, risk, customer experience) plus budget direction. Re-ranks candidates with a bounded boost — quality stays the gate.
Spreadsheet ops, warehouse-ready, or ML-platform-ready. Evidence checklists, builder hints, and rollout language — e.g. Instrument First before Automate when history is thin.
Finance, revenue ops, operations, support/success, or people/HR. Shapes interview prompts, section hints, and auto-map extraction for the workflow locus you care about.
Auto-match or manually pick reference companies from a peer corpus. Reorders gap questions and may nudge ranking while your graph is still thin — never copies another company's model.
Founder, transformation lead, functional VP, or agency partner. Shapes language and emphasis in the Target Selector and opportunity dossiers without changing scores.
Use Case Foundry keeps every scoring axis transparent. Every generated use case includes the artifacts a skeptical buyer needs.
Start with an example or a single offering. Sections appear as you add facts — no empty forms on first load. Pain interview seeds ranked pains; the board reacts while you work.
Feasibility alone buries defensible plays. Moat scores proprietary data, scarce expertise, abstractions, reuse, and switching costs — with graded I/F/M/Q badges, not booleans.
Operators like "support chatbot" fire weakly. Without grounding, they land in "Dropped as generic." An optional critique pass catches vague LLM output before final ranking.
Every Analyze returns a ranked gap panel — pains, accessible data, abstractions — with exact interview questions. Your multi-day discovery backlog, no API key.
Candidates bucket into prerequisites, quick wins, strategic bets, needs more evidence, and dropped generic — so readers see a portfolio, not a flat list.
Department and data-maturity packs shape prompts and section hints. Peer cohort auto-matches reference companies to reorder gaps and lightly nudge ranking on thin graphs.
Graph triggers, prescores, operator coverage, and gap diagnostics run deterministically. LLM only for auto-map, Generate full plan, critique, and dossiers.
Save to server from Step 1 — models persist across refresh. When auth is enabled, saved models are private to your account. YAML export still works for git or Drive.
Instrument First before Automate. Risk Guardrails before substitution. Dependencies are wired from the graph; related candidates are cross-linked on each card.
Export all candidates or one card as JSON to compare LLM runs. Generate packages quick wins and the strategic bet into persona-tailored Markdown dossiers.
Use Case Foundry is a methodology-powered tool for anyone who needs to defend AI bets with evidence — not enthusiasm.
Run client assessments in days, not weeks. Auto-map cuts modeling time; saved models and exportable JSON support multi-session diligence; explainable I/F/M/Q scores stand up to skeptical buyers.
Build a defensible AI roadmap with prescores a CFO can audit. Set mandates and budget direction in Use Case Foundry to re-rank the same portfolio for this quarter's objectives.
Use Case Foundry lets you start from your department — finance, revenue ops, operations, support, or HR — with tailored interview prompts and rollout framing before you map the full company.
Ask "can this company actually win with AI?" and get moat-grounded answers tied to specific assets and capabilities.
Open the app, expand Getting started for a walkthrough, or read the builder guide. Load an example, auto-map a website, save models when signed in, and Analyze instantly — Generate when the graph is grounded.
python -m engine.server --openpython -m engine.run examples/meridian_fab.yamldocker compose up --build