Keep the content. Give customers something they can independently verify.
Cloud Proof creates evidence for consequential agent actions without sending the underlying content to Tallwright. Your environment holds the records. Tallwright holds the proof. Anyone you hand a record to can check it without trusting your systems, or ours.
First three Environments free. No per-record pricing.
The boundary.
Your environment holds the content. Tallwright holds the proof. The line between them is the product.
- The records your agents produce
- Payloads, subject identifiers, amounts, decisions
- Your evidence log, in your systems
- Commitments, computed here
Nothing in this box is sent.
- Witnessed checkpoints of your evidence log
- Proof history, one year included
- The public verification service
- Payloads
- Subject identifiers
Enough to confirm a record. Never enough to read one.
What actually leaves your environment.
Categories, not values. A commitment is a fixed-size summary of a record that cannot be turned back into the record.
One per record, derived in your environment. Lets a verifier confirm a record is the same, never what it contains.
Which Environment, which position in the evidence log, when it was committed.
The state of your evidence log at the moment Tallwright witnesses it.
The record itself: what the agent did, for whom, with which values.
Customer, account, order or user references.
Notes, transcripts, messages. Tallwright has no field for them.
How Cloud Proof fits.
Four places, one direction. Content stops at the second. A witnessed checkpoint is a point at which Tallwright witnesses the current state of your evidence log.
Takes a consequential action.
Records it. Computes the commitment. Keeps the content.
Witnesses checkpoints of your evidence log. Keeps the proof.
Checks the record against the witnessed state. No account, no access to you.
What your customer gets.
A verifiable record, or for a dispute an evidence package: the records you present, the checkpoint that covers them, and a check anyone can run. Tallwright does not interpret it. It states whether the records are what they were.
How verification works for the other sideAll three records match the witnessed state and the evidence chain is consistent. Nothing here reveals their content.
From integration to verification.
One call after a consequential action.
The SDK appends the record to your evidence log and computes the commitment locally. Minutes, not a project.
DevelopersCheckpoints are witnessed as you go.
25,000 witnessed checkpoints per Environment a month are included. Beyond that, usage is billed, never blocked.
PricingA public page or a CLI, no account.
The other side checks the record against the witnessed state. The result is one word and a date.
VerifyOne Environment per product, customer, deployment or workload.
Each has its own identity and its own evidence log; evidence never mixes across them. The first three are free. Identity is managed by Tallwright under Cloud Proof; Own Identity, with your keys, is coming soon.
What Cloud Proof establishes.
A customer or an independent third party can check that the record you present still matches the state Tallwright witnessed earlier, and that any later change would show up in verification. They do it without your systems, your dashboard or your word.
Cloud Proof establishes the integrity of records, not the truth of what happened. The full scope and its limits are set out in the Verification and Evidence Policy. Read it