Frequently asked questions

No. You can start with public context, notes, and partial evidence. The workflow highlights what evidence gaps matter most before major investment.

Yes, in the sense that it automatically gathers and structures evidence from a company URL, documents, spreadsheets, and optional consent-first workflow signals. It then tests that evidence against 52 opportunity patterns and ranks supported candidates. Machine-drafted facts remain proposals until a person reviews them.

No. Advisors, PE/VC portfolio-operations and diligence teams, transformation teams, and functional leaders can all use the same evidence-first workflow with different framing.

General prompting can brainstorm ideas. Use Case Foundry structures evidence, scoring, and sequencing so decisions are easier to defend.

Yes. Teams can use redacted or aggregated evidence and control what facts are accepted before recommendations change.

No. Capture is opt-in per employee, records no screenshots, keystrokes, or file contents, and summarizes activity locally on the employee's machine. Only a redacted summary of recurring workflow patterns is shared, denylists exclude chosen apps, sites, and folders, and nothing enters the company model without human review.

A ranked roadmap that separates quick wins, strategic bets, prerequisites, evidence gaps, and dropped generic ideas. Selected winners can continue into dossiers, agent blueprints, evals, pilot plans, production-handoff packages, and downloadable starter-code scaffolds.

No. It produces an evidence-gated handoff package and starter-code scaffold. It never creates credentials, provisions infrastructure, or connects to customer systems on its own; a technical owner supplies endpoints and credentials, verifies remaining adapters, and decides what goes live.

Analyze and the bundled deployment demos do not need an LLM key. Generate and live use of an exported LLM-backed agent require a configured OpenAI-compatible endpoint. Without a key, Auto-map and CLI generation can emit filled prompts for use elsewhere.

A bundled finished demo opens in minutes. A real company assessment is usually a multi-person, multi-session evidence program: start with a URL or files, analyze immediately, then close the highest-value gaps before generating build artifacts.

Yes. The same server runs locally or in Docker and has documented Cloud Run, Fly.io, and Render deployment paths. Optional authentication adds private saved models; LLM keys remain server-side. Contact the team for hosted or pilot availability.

Yes. Scoring dimensions and rationale are visible so stakeholders can review and challenge assumptions directly.