CHOOSE YOUR SITUATION

Back to customer servicesHealth-consumer products and wellbeing providers

Your product is preparing for agent-assisted procurement, but its health claims lack a verifiable evidence dossier.

A nutrition, sleep, emotional-wellbeing or wellness brand is considering a new purchasing channel. Claims are scattered across websites, studies and internal reports, making applicability hard to assess.

How does this scope address your concerns?

Marketing copy alone cannot explain the evidence in a comparison. Ingredient studies may be mistaken for finished-product effects, or subjective experiences presented as treatment outcomes.

When this scope fits

For authorised health brands, suppliers and digital-health services with a defined product or service version, target population and claims to check.

When the scope needs adjusting

Not for predetermined positive conclusions, invented efficacy endorsements or unauthorised disclosure. This work does not guarantee platform listing, agent recommendations, sales growth, regulatory approval or protocol certification. Live APIs, automated contracting and payments are not deployed.

REAL SOURCES · ILLUSTRATIVE REQUEST · DOWNLOADABLE SAMPLE

Sleep-service dossier: the same facts for people and machines

Download sample packZIP

Three claim types, three evidence chains

01User report

Assumed feedback only; comparator, measure and complete version missing

02Objective sleep

Relevant measurement not supplied; status: missing

03Insomnia treatment

Disease-specific study absent; cannot upgrade from feedback

04Disclosure rights

Separate public-summary permission from restricted attachments

This is an assessment-ready dossier sample, not established efficacy, purchasing-agent integration, a live API or automated trading.

Check	Finding / sample content	Use
Sample fields	claim_id / product_version / population / outcome / evidence_status / source / reviewed_at	Represent missing fields explicitly with null and missing
Review consistency	JSON and readable report point to the same claim record	After review, output the same version within permissions

SOURCES USED IN THIS EXAMPLE

Prepared 5 Oct 2026 · Selected references, not an exhaustive systematic search

READABLE FOR PEOPLE AND AGENTS

Agents should read the limits as well as the claims.

A dossier identifies the product version, population, conditions of use, comparator and outcomes. It retains sources, negative findings, gaps, interests and update dates, and flags comparisons that cannot be made.

Brand or supplier

Provide authorised product details, claims and sources; confirm which fields can be published or read under permission.

Client and procurement agent

Draft a comparison for the client-approved question, retaining gaps and uncertainty. Orders require separate authority after professional review.

Illustrative JSON structure · not a product assessment · no live API

{
  "schema_version": "eos.product-evidence.proposed.v0",
  "status": "illustration_not_product_assessment",
  "product": {
    "product_id": "illustrative-service",
    "version": "example-version",
    "category": "sleep_wellbeing",
    "market": null
  },
  "claims": [
    {
      "claim_id": "claim-example",
      "claim_text": "Illustrative claim: improved self-reported sleep quality",
      "population": {
        "target": "to_be_agreed",
        "studied": null,
        "applicability_limits": [
          "Population applicability not assessed"
        ]
      },
      "endpoint": {
        "name": "self_reported_sleep_quality",
        "measure": null,
        "time_window": null
      },
      "comparator": null,
      "evidence": {
        "type": null,
        "strength": "not_assessed",
        "assessment_method": null,
        "source_ids": [],
        "source_locations": [],
        "favourable_findings": [],
        "negative_findings": [],
        "uncertain_findings": [],
        "missing_evidence": [
          "No product assessment performed in this example"
        ]
      },
      "conflicts_of_interest": {
        "funding": "not_assessed",
        "disclosures": []
      },
      "review": {
        "status": "not_reviewed",
        "reviewer_role": null,
        "reviewed_at": null
      }
    }
  ],
  "access": {
    "public_fields": [
      "illustrative_payload"
    ],
    "restricted_sources": [],
    "rights_status": "illustration_only",
    "permitted_audience": "example_readers",
    "onward_sharing": "subject_to_source_permissions"
  },
  "validity": {
    "evidence_cutoff": null,
    "issued_at": null,
    "valid_as_of": null,
    "update_triggers": [
      "product_or_service_version_changed",
      "new_relevant_evidence",
      "correction_or_retraction",
      "population_or_market_changed",
      "source_rights_changed"
    ],
    "supersedes": null
  },
  "decision_boundaries": {
    "order_authorisation": "required_separately",
    "automated_payment": false,
    "live_api": false,
    "protocol_certification": "none",
    "health_effect_validated": false
  }
}

This is a deliverable plan and field demonstration, with no assessment of a real product, live API, automatic procurement or payment.

Scope & acceptance

Work covered by the quote

Future deliverable plan · quoted by scope

Confirm product or service versions, claim count, populations, markets, outcomes and comparators.
Agree search coverage, full-text volume, human-review depth, JSON fields and readable formats.
Define access, citation, onward sharing and retention rights for public and restricted materials, together with acceptance leads, milestones and update frequency.

A formal quote and milestones follow scope confirmation. New formulations, versions, markets or claims require agreement on additional work; no order or price is assumed.

Finished-product trials, recruitment and measurement, formal submissions, advertising-law review, channel integrations, live APIs and payment systems are not automatically included.
Unlicensed full texts, trade secrets and personal information do not enter public files.

Acceptance criteria

Every claim records population, product version, outcome, comparison conditions, evidence cutoff and missing fields. JSON and readable materials agree.
Each evidence item can be located in its source. Ingredient, nonclinical, finished-product human and experiential evidence are separated.
Negative and uncertain results, extrapolation limits, funding and conflicts are retained. Missing evidence is not labelled effectiveness or ineffectiveness.
Public and restricted content have explicit permissions and audiences. Reviewers, version changes, effective dates and update triggers are recorded.
Agreed examples verify parseable JSON and field meanings. This is file-delivery acceptance, not platform certification, agent ranking or automated-transaction acceptance.

Checking the evidence

A proposed future service and fictional situation, not a completed engagement, platform certification or deployed agent interface.
Google AP2 addresses transaction authority, authenticity and accountability. Current OpenAI ACP documentation also covers structured product catalogs and discovery. Health claims require a separate evidence assessment.
Dossier structure, professionals, permissions, pricing, updates and responsibilities must be agreed before a formal engagement.

Workflow and inputs

From inputs to delivery

Explore the workflow and deliverable structure. This preview does not run searches, read real records or produce medical findings.

01 / WORKFLOW PREVIEW

Define the product and claims

Agree product or service version, market, populations, health claims and permissions. Define disease-treatment, general-wellbeing and experiential claims separately.

Methods, sources and professional responsibilities

THE BASIS FOR THIS SCOPE

Why this approach? Inspect the basis.

Check the match between product, population, use conditions and outcomes before defining a supported claim.

OFFICIAL GUIDANCE

FTC · Health Products

Federal Trade Commission · US

Does the research match the product claim?

Scientific support for health-product claims should match the product and claim, rather than merely a similar ingredient.

Scope & access conditions

This is US guidance. Other markets require separate assessment; it is not an approval or compliance conclusion.

Open official source

PUBLIC DATABASE

ClinicalTrials.gov

NLM / NIH

Check studies and reported results

Review study objectives, eligibility, locations, status and reported results, then identify questions for the clinical team.

Scope & access conditions

Sponsors or investigators submit the records. A listing is not government approval of safety or effectiveness, or a determination that you should participate.

Read the official description

Open official source

METHOD REFERENCE

GRADE · Cochrane Handbook

Cochrane · Chapter 14

Show effect size and confidence separately

Present important outcomes, relative and absolute effects, available data and certainty, with reasons for downgrading.

Scope & access conditions

GRADE evaluates a body of evidence for a particular outcome. Database inclusion or one study does not automatically establish high certainty.

Open official source

These references support problem definition and evidence appraisal. Your conclusions still require project-specific sources and professional review.

The value of this commissionEOS plans same-source human/machine dossiers with versions, claims, populations, endpoints, sources and unknowns for authorised comparison.

Agent-readable evidence and procurement comparisonCompare alternatives and value

A procurement agent should see evidence and limitations alongside marketing.

What to assess	OpenAI / Google Cloud	Evidence OS · Commissioned scope
Service focus	ACP supports discovery; AP2 establishes transaction authority, a different function.	EOS plans same-source human/machine dossiers with versions, claims, populations, endpoints, sources and unknowns for authorised comparison.
When to choose	Listing and payment integration follow platform requirements; integration is not effect validation.	Structure demo: one version, up to three claims, JSON validation, source links, rights and update triggers.

What the fee pays for: Planned service quoted by scope; file delivery, maintenance and real API integration are separate projects.
Value beyond this task: Human summaries and JSON reuse sources; new channels do not need a separately invented medical claim set.
How to assess value: Measure readability, source parsing and manual-check time first; conversion and distribution require real integration.

No live integration, certification or automatic orders; validate recipient fields, source rights and human authorisation first.

CHECK THE OFFICIAL SERVICE SCOPEOpenAI · Powering Product Discovery in ChatGPT Google Cloud · Announcing Agent Payments Protocol (AP2)

References for research definitions and review methods

These sources inform product, population, use-condition and review definitions. Findings about a particular product require its original research; claims and submission requirements are checked for the target market.

Reuse and updates after delivery

The same authorised evidence baseline can support customer questions, channel due diligence and later study planning, reducing repeated explanations across teams. Savings and conversion effects require measurement; the page does not invent ROI.

Agree reuse rights, retention and export periods, and future update fees when scoping.

Questions before commissioning

Can we start when evidence is limited?

Yes, with an inventory. Evidence not found, inaccessible materials and mismatched product versions are labelled before considering narrower claims, new research or stopping a claim. Supportive findings are not guaranteed.

Will JSON make shopping agents recommend us?

No recommendation follows automatically. JSON is a readable deliverable format; channel rules, integration, ranking and purchasing authority are separate. This page demonstrates a future deliverable plan, with no deployed live API or platform certification.

Will confidential materials or paid full texts be public?

Rights are confirmed first. Public dossiers contain only permitted disclosures; restricted evidence is supplied for authorised audiences and purposes. Machine readability does not expand rights. Put only a nonconfidential overview in the current request form.

Define the delivery for your actual question.

Add your goal, available inputs and timing. Agree scope, review responsibilities and a quote before commissioning.

Request scope & quote

Explore other situations

Health-consumer products and wellbeing providers

You have ingredient papers. What do they actually show about your finished product?

Receive an ingredient-to-product evidence table, claim boundaries, conflicting findings and gaps, with options to obtain missing records, narrow the claim or study the finished product.

Quoted by scopeSee the deliverable

Health-consumer products and wellbeing providers

Is the new formula worth another study? Define the one thing it must show first.

Receive a version-specific gap list, priority research question, primary outcome and measurement schedule, comparator options and staged budget inputs.

Quoted by scopeSee the deliverable

Health-consumer products and wellbeing providers

Evidence for meditation is not automatically evidence for your app. What is yours missing?

Receive a version-and-content evidence map, measurable support goals, population boundaries, gaps and study-design elements that also track adverse experiences.

Quoted by scopeSee the deliverable

Health-consumer products and wellbeing providers

An employer wants your stress-support service. Can you show what it changes?

Receive a version-specific evaluation design covering users, primary outcome, comparator, timing, service dose and delivery records, with findings separated from purchasing assumptions.

Quoted by scopeSee the deliverable

Health-consumer products and wellbeing providers

You changed the wellbeing device. Does the old evidence still apply?

Receive a version-change table, affected claim evidence, reasoned transferability assessment and next steps for bench, usability, experience or human comparison studies.

Quoted by scopeSee the deliverable

Health-consumer products and wellbeing providers

People love the experience. What does your wellbeing service actually provide?

Receive a version-specific question, measurable experience and support outcomes, relevant and inapplicable research, risk/adverse-feedback records and a staged evaluation plan.

Quoted by scopeSee the deliverable

Health-consumer products and wellbeing providers

Several health products claim to work. Before procurement, you need to know whose evidence fits your customers.

Receive a supplier-evidence comparison, source and missing-data records, comparison JSON, disagreement notes and a study-gap list. Agreed humans review the materials; authorised staff separately decide purchasing conditions, ranking criteria and orders.

Quote after scope confirmationSee the deliverable

Other tasks in this area

Food, supplements and functional claims Mental support, meditation, sleep and stress Wellbeing experiences, relaxation devices and everyday health Product evidence and research priorities Clinical evaluation and product changes Evidence modules and support capacity