# Customer Discovery Prover MCP for AI Agents AI Agent Connect > Customer Discovery Prover forces your AI to stop making assumptions about your target market. Instead of letting your agent guess who your customers are, this Connector requires it to cite specific interview data, real pain points, and actual commitment signals like deposits or letters of intent. It stops the 'everyone wants this' trap by enforcing Mom Test principles and segment separation. ## Overview - **Category:** productivity - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_FcoU6IdAx3q5sWDJpDPGsPfVRCZ7YEkM4nbyPExm/ai-agent-connect - **Tags:** customer-discovery, product-market-fit, mom-test, icp, persona, willingness-to-pay, startup, customer-development ## Description Customer Discovery Prover acts as a rigorous gatekeeper for your product-market fit research. Most startups fail because they build something nobody wants, often because they let their AI agents hallucinate a 'busy professional' persona or assume 'everyone struggles with time management.' This Connector stops that cycle by forcing the agent to prove the ground truth. When you use it, your agent can't just give you a generic marketing plan; it has to distinguish between a 5-person design agency and a 200-person manufacturer, even though both might be called SMBs. It also stops the agent from accepting 'they said they would pay' as a win, instead demanding actual commitment signals like pilot dates or letters of intent. By using this through the Vinkius catalog, you're giving your AI a YC partner's level of scrutiny. It catches biased questions, identifies when segments are being conflated, and ensures that every claim about a customer's pain is backed by a specific quote or observed behavior. It turns your discovery process from a guessing game into a verifiable audit trail. ## Tools ### validate_customer_discovery Runs a structured reflection on your discovery data to check for persona grounding, evidence, bias, segment separation, and WTP. It forces the agent to cite specific interviews and commitments. ## Prompt Examples **Prompt:** ``` I want to target small business owners who need help with their inventory. ``` **Response:** ``` That's a conflated segment. A 5-person boutique shop and a 100-person warehouse have completely different budgets and buying processes. To pass validation, you need to separate them by company size and specific regulatory requirements. Also, 'need help' is a vague pain. Can you provide specific quotes about how much they currently spend on manual entry? ``` **Prompt:** ``` Our customers said they would definitely pay for this tool if we launched it. ``` **Response:** ``` That is verbal interest, not a commitment signal. To be 'Discovery Proven,' you need to show actual commitment. Have you secured any of the following? * Signed Letters of Intent (LOIs) * Deposits or pre-payments * Scheduled pilot dates * Referrals to other decision makers ``` **Prompt:** ``` Review my persona: Busy professionals aged 25-45 who value productivity. ``` **Response:** ``` This is a demographic, not a persona. It fails the grounding test. To fix this, you need to provide: * Specific job titles (e.g., VP of Engineering) * Company context (e.g., 100-person fintech) * Observed behaviors (e.g., spending 4 hours a week on manual data entry) * Direct quotes from your interviews. ``` ## Capabilities ### Validate customer personas Forces the agent to cite specific roles and observed behaviors from real interviews. ### Extract pain point evidence Requires the agent to provide specific quotes and costs regarding current workarounds. ### Screen for biased validation Checks for leading questions and ensures the agent follows The Mom Test methodology. ### Separate distinct buyer segments Identifies unique budgets and buying processes for different groups to avoid conflation. ### Verify willingness-to-pay Demands tangible commitment signals like deposits or signed letters of intent. ## Use Cases ### Validating a new SaaS idea A founder asks the agent to review their ICP. The tool flags 'SMBs' as too broad and forces a breakdown by company size and budget. ### Preparing for a pitch A founder wants to prove market need. The tool identifies that their 'willingness to pay' is just verbal and demands evidence of deposits or signed letters. ### Refining a product roadmap A PM wants to know why a feature is a priority. The tool ensures the pain is evidenced by specific quotes from a specific segment. ### Auditing existing research A consultant runs your 23 interviews through the tool to see if the discovery actually holds up to investor scrutiny. ## Benefits - Stop building for 'busy professionals' by using validate_customer_discovery to force the agent to name specific roles and behaviors. - Eliminate biased validation by ensuring the agent asks about past behaviors instead of future promises. - Prevent segment conflation by forcing the agent to separate different buyer groups with unique budgets and triggers. - Move past 'verbal interest' by requiring actual commitment signals like LOIs or pilot dates for every WTP claim. - Catch 'hallucinated' problems by making the agent cite specific quotes and costs from your actual interviews. - Prepare for investor scrutiny by turning vague market interest into a verifiable audit trail of customer commitment. ## How It Works The bottom line is you get a rigorous audit of your market research that proves you aren't building in a vacuum. 1. Input your current persona, problem statement, or segment analysis into your AI client. 2. The agent runs the data through the validation logic to check for assumptions and biases. 3. You get a verdict matrix showing exactly where your discovery is grounded and where it is just hand-waving. ## Frequently Asked Questions **How does Customer Discovery Prover stop me from making bad assumptions?** It forces your AI client to cite specific evidence from your interviews. If the agent can't provide a quote or a specific role, the tool flags the discovery as 'Assumed' or 'Invented.' **What is 'The Mom Test' and how does this Connector use it?** The Mom Test is a methodology for getting honest feedback by asking about past behavior instead of future promises. This Connector ensures your AI doesn't ask leading questions like 'Would you pay for this?' **Can I use Customer Discovery Prover to validate my ICP?** Yes. It specifically checks if your Ideal Customer Profile is grounded in real data or if it's just a list of demographics like age and interests. **Will this tool help me get more funding?** It helps you prepare for funding by ensuring your discovery survives investor scrutiny. It moves you from 'we think people want this' to 'we have evidence that this specific segment will pay.' **Does this Connector work for both B2B and B2C startups?** It works for any business model. It's particularly effective for B2B because it helps separate complex buyer segments that often get conflated as 'enterprise' or 'SMB.' **What happens if my discovery is rejected by the tool?** The tool provides a verdict matrix. It tells you exactly which pivot failed, whether it was biased validation, conflated segments, or lack of commitment signals, so you know exactly what to fix. **Does it conduct interviews?** No. It validates that your discovery process is grounded in real data — interview evidence, unbiased methodology, separated segments, and commitment-based WTP signals. It does not replace conversations with customers. It forces you to prove you had them. **What is The Mom Test?** A framework by Rob Fitzpatrick for conducting customer interviews that produce truthful data. The core rule: never ask leading questions about the future ('Would you use this?'). Instead, ask about past behavior ('When did you last encounter X? What did you do?'). People lie about future behavior — past behavior is a reliable signal. **Can pre-product startups use this?** Yes — it is designed for pre-product discovery. WTP signals for pre-product include: signed LOIs, paid design partnerships, deposits against future delivery, time commitments (agreed to a weekly feedback session), and reputation commitments (willing to be named as a design partner). You do not need a product to test willingness-to-pay.