For AI vendors

Your claimed numbers, audited independently.

An audited score from your customers’ own systems, a live badge, and a signed attestation for procurement.

Your first audit stays private until you publish it. Free for every vendor today.

Audited by Attify: Claim confirmed · within 5.0 points. Synthetic vendor support agent, synthetic sample. 56.9% audited (95% interval 48.2% to 65.1%), August 2026. Claimed 58.7%: 1.8 points below the claim, confirmed within 5.0 points.

Synthetic vendor, support agent

  • Synthetic demo data
  • Standard #1 · v0.11
  • Pooled across 3 customers

On synthetic data, audited resolution was 56.9% against 58.7% claimed, a 1.8-point gap; 1.8 of those points are conversations counted as resolved where the customer came back within 7 days.

Claimed
58.7%
50.6–66.3
Audited, judge-confirmed
56.9%
48.2–65.1
Gap, points
1.8
0.9–3.6

Claimed, by outcome, % of eligible

  • Audited
  • Repeat contact
  • 95% interval
0%20%40%60%80%58.7%
Synthetic demo data: invented conversations, judged by a scripted demo judge. 2 Aug – 23 Sept 2026. 1,092 audited · 1,056 judged

What you get. A score from your customers’ own data, a live badge, a signed attestation and a traction report.

  • An audited score from your customers’ own data

    Every record of a rolling 30-day period, read from each customer’s own help desk, CRM or calendar with its consent, or streamed by your agent and reconciled by the customer. The claimed rate beside the audited rate, each with its 95% interval.

    The standards
    • Customer A
    • Customer B
    • Customer C

    Audited, pooled

    56.9%

    48.2–65.1 · 58.7% claimed

    Synthetic demo data

  • A live badge and a public profile

    The badge is drawn from your current status each time it loads and links to your profile and its public changelog. A copy hosted anywhere else is not a badge.

    The registry
    Audited by Attify: Claim confirmed · within 5.0 points. Synthetic vendor support agent, synthetic sample. 56.9% audited (95% interval 48.2% to 65.1%), August 2026. Claimed 58.7%: 1.8 points below the claim, confirmed within 5.0 points.

    Drawn from the current status

    • Claim confirmed · within 5.0 points
    • Claim not confirmed
    • Lapsedno fresh data for 14 days
    • Revokedcarries no numbers
  • A signed attestation for procurement

    A statement of your score and status, signed with Ed25519, that anyone can check. It expires 44 days after it is issued, and /verify says when a newer one replaces it.

    Check an attestation

    Attestation

    Synthetic, unsigned
    Vendor
    Synthetic vendor, support agent
    Standard
    #1 · v0.11
    Period
    2 Aug – 23 Sept 2026
    Audited
    56.9% 48.2–65.1
    Claimed
    58.7%
    Status
    Claim confirmed · within 5.0 points

    SignatureEd25519, added when issued

  • A traction report for fundraising

    Usage, retention and audited outcomes over the months your customers’ data covers, signed, with share links that expire and a log of who viewed them. Free when an investor requests it.

    Traction report

    • Usageper customer, per month
    • Retentionactive, new, retained, churned
    • Pilot to productionthe first week at the threshold
    • Concentrationthe largest customer, the largest three
    • Outcomesaudited, with 95% intervals
    • Claims testedup to five, each with its result
The sample attestation in full, as JSON

The score, its period, the standard and its version, the evidence tier, the status and why. This one is built from synthetic data and is not signed.

A real attestation adds its id, when it was issued and when it expires, the profile and badge addresses, and an Ed25519 signature that /verify checks against Attify’s published keys.

Sample attestation, synthetic and unsigned
{
  "schema": "proven.attestation/1",
  "issuer": "Attify",
  "vendor": {
    "name": "Synthetic vendor"
  },
  "product": "support agent",
  "job_type": "support-resolution",
  "standard": {
    "id": "support-resolution",
    "number": 1,
    "title": "Attify Audited Resolution Standard",
    "version": "0.11"
  },
  "rules_version": "5",
  "status": "claim_confirmed",
  "status_reason": "Audited resolution is 56.9% (95% interval 48.2% to 65.1%), 1.8 points below the claimed 58.7%: claim confirmed within 5.0 points, the gap's 95% interval (0.9 to 3.6 points) staying within the 5.0-point tolerance.",
  "evidence_tier": "judge",
  "period": {
    "start": "2026-08-02",
    "end": "2026-09-23"
  },
  "claimed_rate": {
    "value": 0.5867,
    "lo": 0.5065,
    "hi": 0.6626
  },
  "audited_rate": {
    "value": 0.5688,
    "lo": 0.482,
    "hi": 0.6515
  },
  "verified_rate": {
    "value": 0.5688,
    "lo": 0.482,
    "hi": 0.6515
  },
  "gap": {
    "value": 0.018,
    "lo": 0.0088,
    "hi": 0.0363
  },
  "claim_confirmed": true,
  "claim_tolerance_points": 5,
  "sample": {
    "customers": 3,
    "conversations": 1092
  },
  "summary": "On synthetic data, audited resolution was 56.9% against 58.7% claimed, a 1.8-point gap; 1.8 of those points are conversations counted as resolved where the customer came back within 7 days."
}

Why buyers ask for third-party data.

86% of B2B technology buyers agree or strongly agree that objective, third-party-authored content or data is critical for trusting thought leadership positions or claims.
Forrester Consulting, Content Preferences Study, 2023, 767 technology buying decision-makers at global enterprises. Forrester sells third-party research.

Ways to prove your numbers. Compared on who measures, what it rests on and whether a buyer can check it.

Ways vendors prove their numbers, compared
Self-reported dashboardWho measuresThe vendorWhat it rests onIts own logs, counted by its own rule; under outcome pricing, the same count it billsWho paysPart of the productHow currentLiveCan a buyer check itNo
Commissioned ROI studyWho measuresAn analyst firm’s consulting armWhat it rests onInterviews with customers the vendor names, built into a modelled composite companyWho paysThe vendor commissions itHow currentOnceCan a buyer check itThe report, not the data
Review sitesWho measuresUsers, moderated by the siteWhat it rests onUsers’ own opinions, often rewarded with a gift cardWho paysVendors can fund review campaignsHow currentRollingCan a buyer check itThe reviews, not the outcomes
Case studyWho measuresThe vendor, with a customer’s approvalWhat it rests onResults a customer reports, chosen by the vendorWho paysThe vendorHow currentOnceCan a buyer check itRarely
AttifyWho measuresAn independent auditor, under a published standardWhat it rests onEvery record of the period, from the customers’ own systems, by a published ruleWho paysThe vendor or the buyer subscribes; the price never depends on a resultHow currentEvery 30 days, with a live statusCan a buyer check itYes: a signed attestation anyone checks at /verify

Sources: the disclosure that opens every Forrester Total Economic Impact study (an example); G2’s gift-card policy; the FTC’s Endorsement Guides on reported results.

How it works. Connect the data, audit with your customers, publish when you choose.

  1. Connect the data

    Your customers connect the system your agent works in (Intercom, Zendesk, HubSpot, Google Calendar, Microsoft 365 and more), read-only. Or connect your own agent platform or observability account, or stream each interaction through the Ingest API, an SDK or OpenTelemetry.

    The Ingest API
  2. Attify audits with your customers

    A score needs at least 3 independent customers. Each agrees to exactly what Attify reads and passes four independence checks; witnessed data also needs its consent and a monthly reconciliation. You see a private daily score from the first connection. Your first judged audit is private too: you see every number and choose whether to publish it.

  3. Publish your badge

    Once you publish, the score goes on your profile, the registry, the badge and a signed attestation. A new score is made every 30 days and publishes after a 7-day review. Without fresh data for 14 days, the status reads Lapsed.

    The badge

What we audit today. Three job types, each under its own published standard.

Customer support

Standard #1 v0.11

Read from
Intercom, Zendesk, HubSpot, Salesforce, Gorgias, Freshdesk
Or from the agent itself
Its platform, or our API and OpenTelemetry
Free check
An export file, no account

Sales meetings

Standard #2 v0.6

Read from
HubSpot, Salesforce, Google Calendar, Microsoft 365, Calendly, Cal.com
Or from the agent itself
Its platform, or our API and OpenTelemetry

Bookings

Standard #3 v0.6

Read from
Google Calendar, Microsoft 365, Calendly, Cal.com
Or from the agent itself
Its platform, or our API and OpenTelemetry
What each standard checks

Did it stick?Did a person have to redo it?Did it happen?

Customer support

  • Did the customer come back within 7 days?
  • Did support redirect instead of answering?
  • Does the judge find the issue resolved in the conversation?

Sales meetings

  • Is it on the customer's calendar, booked by the AI?
  • Was the person new to the customer when it was booked?
  • Did a person on the customer's team have to step in?
  • Was it cancelled, or not held within 14 days of its first time?
  • Did someone from the prospect's side attend?
  • Does the judge find a sales meeting a potential buyer agreed to?

Bookings

  • Is it in the business's system, booked by the AI?
  • Was it a new appointment, not a move or a duplicate?
  • Did staff have to change it within 48 hours?
  • Was it cancelled, or not kept within 14 days of its first time?
  • Did the customer come?
  • Does the judge find the booking matches what the customer agreed to?
  • Connected

    Attify reads every record of the period from the customer’s own systems, with its consent. It supports a verification.

  • Witnessed

    The vendor’s agent streams each interaction to Attify as it happens. It counts once the customer consents and reconciles it each month.

  • Uploaded

    An export you send. A self-reported check, never a verification.

The founding cohort. Free for every vendor today.

Start free verification

Free

Every vendor is free today. The first 10 vendors to open a program also keep it free for 12 months from their first published score once pricing starts, then 50% off, recorded on their account.

In return: Feedback when asked.

Once vendor pricing starts

  • Starter · up to 10 customers$500 a month
  • Growth · up to 50 customers$1,500 a month
  • EnterpriseCustom
  • Traction report, seed$2,500
  • Traction report, Series A and later$5,000
  • Pricing

Questions

What if our score is lower than we claim?

You see where you stand before anything is public: a private daily score from your first connected customer, a private report while you are below the publishing minimums, and a private first audit. Nothing publishes until you choose to. Once it does, the status reads Claim confirmed · within 5.0 points when the audit shows, with 95% confidence, that your count overstates audited outcomes by at most 5.0 points, and Claim not confirmed otherwise.

How the claimed rate is measured
Who sees our results, and when?

Until you publish, only you and Attify’s reviewers. Each customer sees its own results. You see each customer’s results and consent, never their conversations. Once published, the score, its status and the changelog are on your profile, behind your badge and in the attestation.

What happens if our score drops later?

Once you publish, every later score publishes too, whatever it shows. Each stays private for 7 days, where you can add one public comment, then replaces the last one; a published score cannot be hidden or edited. With no fresh data for 14 days the status reads Lapsed. If you stop verification, your profile reads Lapsed and keeps its full history. Each change is dated in the public changelog, and the badge always shows the current status.

Do our customers have to do anything?

Each one accepts your invitation, reads exactly what Attify will read, and connects its system read-only, which some companies put through a security review first. For an agent on your own platform, they consent and reconcile each month: their own total, then a spot check of 10 interactions. Attify also asks some of them directly whether they use your product.

How long does it take?

Counted from your third independent customer connecting: its first sync reads 90 days back at once, the independence checks run (a customer Attify asks directly has 14 days to answer), then your first judged audit, private until you publish it. About one to two weeks, given at least 3 customers and 500 AI-handled conversations in the 30-day period (sales meetings: 150 meetings over 90 days).

Is it independent if we pay?

The price never depends on a result: buyers pay by audited AI interactions and vendors by customers covered, never by a score. The standards and the judge’s instructions are published. Attify never advises a vendor on how to raise its score. Customers you invite must pass four independence checks, and a customer proven fake revokes the program in public.

Pricing

Your claimed numbers, audited independently.