Now taking early access requests

Show me the receipts.
Not “trust me.”

Receipts turns any investigation or public claim into a room you can open: the claim, the evidence behind it, where that evidence came from, and how confident anyone should be — all laid out like a receipt, because a receipt is proof, not a promise.

Primary + secondary sources logged Corrections are append-only AI inference always disclosed

Claim #0417 · Municipal Budget Office

“The new transit line was completed under budget, at $42M.”

Primary evidence 2 documents

Signed final cost report (City Comptroller, filed 3/12) · Capital ledger export (Transit Authority)

Corroborating evidence 3 independent

Independent auditor letter · Two contractor invoices matching ledger totals · Bond disclosure filing

Contradictory evidence 1 flagged

Change-order memo suggests $1.8M in costs may be booked to a separate fiscal year.

Confidence: corroborated, with caveat Last verified Aug 12, 2026

What we know

  • · Final ledger matches signed cost report
  • · Two independent invoices reconcile

What we don't know

  • · Whether the $1.8M change order lands in this fiscal year
  • · Comptroller has not replied to request for comment
Claim → Evidence → Sourceevery published claim, structured
Append-only correctionsnothing edited away quietly
Evidentiary categories are lockedAI can't upgrade allegation to fact

The PROVE IT flow

One button. The whole chain of evidence, in order.

Every claim on Receipts carries a room behind it. Open it and you move through the same sequence a careful investigator would — evidence first, confidence second, open questions last.

01 / Evidence

Primary → Corroborating → Contradictory

Source documents first, then what backs them up independently, then what pushes back. In that order, every time.

02 / Confidence

A stated confidence level

Not a green checkmark for everything. Confidence is graded — confirmed, corroborated with caveat, disputed, or unverifiable.

03 / What we know

Stated plainly, no hedging

The facts the evidence actually supports, written in plain language — not buried in a document dump.

04 / What we don't know

The open questions, kept visible

Unresolved gaps are a permanent part of the room, with a last-verified date — not smoothed over.

A long thermal-paper receipt printout with itemized entries, checkmarks, and small verification stamps, resting on a linen surface

Feature

The Evidence Nutrition Label

Every claim ships with a standardized quality label — the same way a food product discloses its contents. No claim on Receipts is just a headline; it's a headline with its evidentiary nutrition attached.

  • Primary and secondary source counts, tallied
  • Independent corroborations, listed by count
  • Contradictions and unresolved questions, never hidden
  • Evidence freshness — when it was last checked
  • % of the room drafted by AI inference, disclosed
  • A plain yes/no on human verification

Evidence Facts

Claim #0417 · Municipal Budget Office

Primary sources 2
Secondary sources 4
Independent corroborations 3
Contradictions on record 1
Unresolved questions 2
Evidence freshness 1 day
AI-inferred content 18%
Human-verified Yes

% AI inference reflects drafting and summarization assistance only. Every underlying source document was reviewed by a human editor before publication.

Trust Passport TP-2026-0417-v3
Sources attached 6 attached
Human reviewed Confirmed
AI inference disclosed 18% disclosed
Correction history 1 correction, v2
Current version v3 · Aug 12, 2026

Feature

The Trust Passport

A machine-readable provenance stamp travels with every published claim — so a claim can be checked, embedded, or cited elsewhere without losing its evidentiary chain of custody.

  • Records exactly which sources are attached, and how many
  • States plainly whether a human reviewed the room
  • Discloses AI involvement as a percentage, not a footnote
  • Carries full correction history and current version number

Trust principle

Nothing gets quietly edited away.

Corrections on Receipts are append-only and always visible — the original stands next to the correction, not replaced by it. And the system enforces a hard rule underneath every room: AI is never allowed to silently upgrade an assertion's evidentiary category from allegation to fact. A claim only moves from “alleged” to “confirmed” when a human reviewer says so, on the record.

Aug 12, 2026 — v3 published current
Jul 30, 2026 — change-order memo added as contradictory evidence correction
Jul 2, 2026 — v1 published, confidence: corroborated archived

Built for

For anyone who's tired of being asked to just trust it.

Readers

The general public

Open any claim and see exactly what it's standing on, instead of taking a headline's word for it.

Publishers

Newsrooms & investigators

Attach a credibility layer to reporting without building evidence infrastructure from scratch.

Organizations

Agencies & institutions

Publish claims, metrics, or public statements with a verification room attached from day one.

Researchers

Fact-checkers & analysts

Cite a room instead of a headline — every source, every contradiction, already assembled.

Early access

Be first in the room when Receipts opens.

We're onboarding early publishers and readers in small batches. Leave your email and we'll reach out as rooms open up.

You're on the list. We'll be in touch.

No spam. No backend, no database yet — this is a pre-launch signal only.

You're on the list.