Preuve AI helps founders investigate a startup idea before committing to build it. It uses AI agents to research demand, competitors, and pricing, then presents a viability score with claims linked to their sources. The workflow can go beyond evaluation: Smart Pivots suggest ways to improve an idea, and a build kit turns the research into a structured plan for a coding agent. It is aimed at solopreneurs, founders, and agencies that need to assess ideas or prepare client research.
Key features
Sourced market research
A deep report uses ten AI agents to investigate an idea across more than 60 live sources, including Crunchbase, Google Trends, Reddit, Product Hunt, and G2. Reports can include up to 20 sections, and users can open the source behind individual claims. The report also maps up to 15 competitors, with information such as funding, pricing, threat level, and ways to compete. A viability score out of 100 summarizes the assessment and indicates whether the idea clears the product’s launch-ready bar.
Preuve AI also offers Starter Scans for quickly checking new angles. The page describes these as a roughly 60-second way to shortlist ideas; a deep report takes several minutes and provides more detailed research, Smart Pivots, and a build kit.
Improve an idea, not just score it
Smart Pivots identify what is limiting a score and rank suggested changes by expected impact. Each suggestion includes effort, what would change, what would stay, supporting evidence, and a trade-off. Applying a pivot triggers a new deep report and score for the revised idea. Users can also rewrite an idea themselves and rescore it, as described in the product FAQ.
Research-to-build handoff
Each deep report can generate a kit containing an AGENTS.md, a five-milestone PLAN.md, and documentation about buyers, scope, competitors, pricing, and risks. The kit is designed for coding agents including Claude Code, Cursor, Codex, and Antigravity. The product page also describes starting a project in Lovable and accessing research through MCP from tools such as Claude, ChatGPT, or Cursor. A launch plan for finding the first 100 customers is another part of the report workflow.
Agency workflow
For agencies, Preuve AI describes a branded intake form and dashboard for screening prospects, plus controls for choosing which report sections clients can see. Client-facing links can use an agency’s name and visual identity; reports can be exported as Markdown or JSON. The page identifies the agency workspace and commercial-use features as intended for client work.
Who it is for
Preuve AI is most relevant to founders who want evidence to inform whether and how to pursue an idea, especially those using AI coding tools to build an MVP. It may also suit consultants or agencies that evaluate client concepts and need to present research in a branded format. Its combination of competitor research, idea scoring, and a build plan is more focused on early-stage validation than general-purpose AI chat.
Limitations
A Starter Scan is for shortlisting, not a substitute for a deep report; more detailed analysis takes several minutes. The page describes research and scoring, but it does not establish that a score predicts commercial success. Results should therefore be treated as research inputs rather than a guarantee that an idea will find customers. The depth of the agency workflow also depends on access to the agency-specific features described on the page.
Verdict
Preuve AI brings market research, sourced evidence, idea iteration, and MVP planning into one workflow. Its strongest distinction is the handoff from a report to a structured plan for coding agents, with Smart Pivots providing a way to investigate alternatives instead of stopping at a score. It is worth considering for founders who want to pressure-test ideas before building, while keeping in mind that the research supports judgment rather than replacing it.


