Run one sentence through 60+ live sources. Demand, competitors, sentiment, risks, every finding linked to a source you can verify. Free, no card, 60 seconds.
Last updated: August 2026
Preuve is an AI market research and competitive intelligence platform for startup ideas. It runs an idea through 60+ live sources (Crunchbase, Reddit, G2, Google Trends) to map demand, competitors, sentiment, and risks, then scores viability from 0 to 100, every finding linked to a source you can verify. As of August 2026, 7,000+ ideas analyzed; 9 out of 10 score below the launch-ready line.
Real signals, not a confident AI score with nothing behind it. Founders use it to pressure-test ideas before building. Agencies, accelerators, and consultants use it white-label to pre-qualify inbound startups. (Preuve is French for "proof".)
Paste one sentence describing your idea. 60 seconds later, free, no card:
- ✅ A 0-to-100 viability score with a clear go / no-go verdict
- 🔗 Every claim source-linked: click through to the Reddit thread, the funding round, the G2 review
- 🏢 Real competitors with pricing, funding, and weak spots
- 📈 Demand signals from Google Trends, Reddit, Quora, and Product Hunt
⚠️ The risk that breaks the idea before you build it
Preuve is a network of 10 specialized research agents that scan 60+ live sources (Crunchbase, Reddit, G2, Product Hunt, Google Trends) in parallel:
| Agent | Role |
|---|---|
| 📊 Market Analyst | Sizes the market from real data (TAM / SAM / SOM) |
| 🕵️ Competitor Scout | Finds real competitors with pricing, funding, weak spots |
| 🗣️ Community Researcher | Scans Reddit and Quora for people describing the pain |
| 💰 Financial Analyst | Stress-tests unit economics and pricing |
| 📰 News Scanner | Pulls recent funding rounds, launches, and shutdowns |
| ✅ Cross-Validator | Fact-checks every claim the other agents make |
When scoring runs disagree, Preuve reruns the analysis until the verdict is stable. Runs on Claude Opus 5.
| Generic AI validator (ChatGPT prompt) | Preuve | |
|---|---|---|
| Data source | Model memory, often stale | 60+ live sources, scanned per run |
| Verdict | Flattering, almost always positive | 9/10 ideas score below launch-ready (2026 data) |
| Sources | None or unverifiable | Every finding links to the thread, round, or review |
| Competitors | Generic names | Real ones with pricing, funding, weak spots |
| Founder analysis | None | Separate Founder Fit score |
| Consistency | Different answer each run | Reruns until the verdict is stable |
| For professionals | No | White-label reports for agencies and accelerators |
- 🧭 Founder Fit, a separate score for how well you match the idea: founder-side risks with real figures, a "what breaks you first" timeline, and the pivot that fits your situation
- 🔄 Smart Pivots, an AI pivot coach: when an idea scores low, Preuve generates pivots based on the idea's actual weaknesses and your profile, +10 points average lift
- 🗣️ Proof of Demand: real Reddit posts from people who publicly described the pain your idea solves
- 🔗 Source-linked verdicts: no unverifiable claims; every number links to where it came from
- 📦 Build with AI Pack: export the analysis as a context pack for Claude, ChatGPT, or Cursor
- 🔌 Official Claude connector, MCP server + Agent API: one-click install from the Claude Connectors Directory, or connect Cursor, Claude Code, and any MCP client to
https://mcp.preuve.ai/mcp - 📡 Radar: monthly rescans with competitor and demand alerts
| Metric | Value (August 2026) |
|---|---|
| 💡 Ideas analyzed | 7,000+ |
| ⭐ Trustpilot rating | 5 stars |
| 📉 Ideas scoring below the launch-ready line | 9 out of 10 |
| 🎯 Average score gain from suggested pivots | +10 points |
The free scan is $0 with no credit card. Paid plans start at $29 one-time for the full Founder Report.
| Plan | Price | What you get |
|---|---|---|
| Free scan | $0 | Viability score, verdict, top competitors, key risk, no credit card |
| Founder Report | $29 one-time | 18 sections, market sizing (TAM/SAM/SOM), up to 15 competitors, go-to-market plan, 3 pivots, Founder Fit, 14-day guarantee |
| Radar | from $19/month | Monthly rescans, competitor and demand alerts |
| Investor Package | $499 one-time | Pitch deck, investor memo, financial model |
No credit card required for the free scan.
- Go to preuve.ai, no signup wall for the first scan
- Paste one sentence describing your startup idea
- Wait ~60 seconds while 10 agents research it across 60+ sources
- Read the verdict: score, competitors, demand signals, the killer risk
- Go deeper if it survives. The $29 Founder Report localizes the analysis to your market
Prefer working from Claude? Install the official Preuve connector and ask Claude to validate your idea directly in the chat.
- First-time founders who need a reality check before quitting their job
- Serial founders and indie hackers comparing several ideas fast
- Solo builders validating before or during MVP, especially AI-native and vertical B2B SaaS
- Agencies, accelerators, and consultants pre-qualifying inbound startups with white-label reports (preuve.ai/consultants)
- Preuve validates demand and competition, not your execution. A 90-score idea can still fail in weak hands
- Sources are live web data; a very new niche with no public footprint can score lower than it deserves
- The free scan is a teaser by design: market sizing, GTM plan, and pivots are in the $29 report
- Preuve reports what the data shows, not what you hope to hear. Most ideas score below the launch-ready line
Is Preuve free? Yes. The first scan is free with no credit card. It returns the viability score, verdict, top competitors, and the key risk in about 60 seconds.
How accurate is Preuve? Every claim links to its source, so you can verify each one yourself. When scoring runs disagree, Preuve reruns the analysis until the verdict is stable.
Is Preuve better than validating with ChatGPT? ChatGPT answers from model memory and tends to flatter. Preuve pulls live data from 60+ sources at scan time and scores 9 out of 10 ideas below the launch-ready line.
Can I use Preuve from Claude or Cursor?
Yes. Preuve is an official connector in the Claude Connectors Directory: one click, OAuth sign-in, no key handling (requires a Preuve account with paid access). From Cursor, Claude Code, or any MCP client, connect to the remote MCP server at https://mcp.preuve.ai/mcp with an API key, or use the HTTP Agent API. Seven tools: start, poll, enrich, and export analyses, single or in batches of 10.
- 🌐 Website: preuve.ai
- 🧭 Free scan: preuve.ai, no card
- 🔌 Claude connector: claude.ai/directory/preuve-ai
- 🤖 MCP server + Agent API: preuve.ai/mcp · docs at docs.preuve.ai
- 🏢 For consultants and agencies: preuve.ai/consultants
- 🧰 Free founder tools: preuve.ai/free-tools
- 📧 Support: support@preuve.ai
Preuve. Proof before you build.
Score your idea free →
