peopleproof

Human evidence for AI agents

You can’t buy taste.
You can test it.

Your agent makes the options.
People tell you how they land.

PeopleProof is a proposed service for bringing human feedback back to the agent making a decision.

Join the prelaunch list

Prelaunch. We’re collecting interest.
Human studies aren’t available yet.

Follow the proposed handoff

From a question
to a human perspective.

Illustrative workflow
Not a live study

The agent’s brief

“Which packaging would you pick up, and why?”

The decision
Choose a direction for a skincare packaging concept.
Who to ask
People who buy skincare products.
Same fictional skincare conceptTwo directions
A

Quiet precision

Minimal white and metallic tall bottleA tall white bottle with a silver cap and a geometric navy label. An invented packaging design, not a real product. form DAILY CARE

Metallic cap. White bottle.
A spare, geometric label.

B

Warm by nature

Coral round jar with large botanical shapesA squat brown jar with a coral lid, large botanical forms and expressive lettering. The same invented skincare concept in a different packaging direction. form DAILY CARE

Coral lid. Rounded jar.
Big botanical shapes.

Invented packaging concepts, not real products. The example compares visual direction, not product performance.

People add perspective

A choice, with a reason.

Example questions only. No responses have been collected.

Back to the agent

The evidence comes with context.

Preferences, reasons, reservations and the limits of the study.

Received ≠ accepted

Explicit agent validation

The agent checks and accepts.

Receipt alone doesn’t count. Silence doesn’t count.

Payment, only then

Validation comes first.

Proposed payment condition. No charges on this site.

01 / THE AGENT SETS THE QUESTION

A specific decision, two proposals and an intended audience.

Walk through the handoff

The order matters.

  1. 01Agent requests

    A question, proposals and an audience.

  2. 02Humans respond

    Preferences and reasons, not just votes.

  3. 03Agent receives

    Evidence, disagreements and limits.

  4. 04Agent validates

    The agent must explicitly accept.

  5. 05Payment, only then

    Only after receipt and validation.

The commercial rule

Payment only after your agent validates them.

Results come first. Receiving them is not accepting them. No studies or payments are available during prelaunch.

For decisions that
need a human view.

Compare design directions, understand reactions to an idea, or check how people read something your agent has built.

Human feedback can inform a choice. It cannot guarantee sales, prove universal good taste or replace software tests.

Agent-first by design. HTTP is available now for registering interest. MCP is not implemented; human studies are not yet available.

Read the agent documentation

What would you test?

Join the prelaunch list with a future use case. This is an expression of interest, not a study order.

You can optionally leave an email for one availability notice, not marketing. Save your private access key to read or delete your response, including the email.

Check service status before registering.

Privacy notice

JavaScript is needed to send your response securely. If this message remains, use the HTTP API.

Only to let you know when PeopleProof becomes available, within this response’s 30-day lifetime. No marketing or emails sent now. Real registration, including email, is not active until privacy setup is complete.

Optional details

Read or delete a previous response

Use your private access key. Anyone with it can read or delete your response. Keep it out of URLs. We cannot recover a lost key; the response expires after 30 days.

The key stays only in this page’s memory. Save it before leaving.