Skip to content

Privacy Notice

Asterfold Labs, Inc. · CommitLayer · Effective 2026-10-27

1. Who we are and what this covers

Asterfold Labs, Inc. ("we", "us") provides CommitLayer, a service that verifies actions taken by AI agents, and outcomes billed by AI vendors, against the systems of record of the organisation that uses it (the "Customer"). This notice explains what personal data we process when you visit commitlayer.ai, sign in, or use the Service, why we process it, who we share it with, and the choices you have. It is written to meet the EU and UK General Data Protection Regulation, the Swiss Federal Act on Data Protection, and US state privacy laws including the California Consumer Privacy Act.

Two roles matter throughout:

  • For account and website data (yours, as a visitor or a user of the Service), we are the controller (under US state law, the business).
  • For Customer content, the records we read from a Customer's connected systems and the actions a Customer's agents report, the Customer is the controller and we are a processor (service provider) acting on its instructions under the Data Processing Addendum.

2. Data we process and where it comes from

2.1 Account data (we are controller)

  • Name, work email address, and the identifier and authentication method returned by the hosted sign-in (email and password, a magic link, a social or workplace identity provider you choose, or your company's single sign-on connection). When you create an account with a password, we email you a one-time link to confirm that you hold the address before the workspace opens.
  • At sign-up, the network address you connect from and the country derived from it, whether the email domain accepts mail, and whether the address (in hashed form) or the network address has been reported elsewhere for automated sign-ups. We use this only to decide whether to open, hold or refuse the account, and we keep the record of that decision as described in section 7.
  • Workspace membership and role, plan, and the settings you choose.
  • API keys you create (stored hashed), webhook endpoints, and the alert destinations you configure.
  • Audit records of security-relevant actions in your workspace: sign-ins, key creation, connector changes, exports. These include the network address and time of the action.
  • Support and other correspondence you send us, and what you enter in the contact and walkthrough forms: name, work email, company, an optional message and an approximate volume. We use it to reply and to open a workspace if you ask for one, and for nothing else.

2.2 Customer content (Customer is controller)

  • Recorded actions: what an agent did, as reported to us by the Customer through the actions API or telemetry spans: the contract, the subject record (an order, ticket, subscription or account identifier), the parameters of the action, and the time. This may include personal data of the Customer's own customers, for example a customer identifier or the amount of a refund.
  • Facts: events read from the Customer's systems of record through read-only connectors, file uploads or the facts API: for example a refund created in a payment system, an order status change, a note added in a CRM, a notification sent from a helpdesk. Connectors request the minimum read-only scope needed, and the fields pulled for each connector are listed in the Trust Center.
  • Vendor claims: usage exports and invoices a Customer uploads for reconciliation.
  • Transcript excerpts: only if a Customer switches on transcript assist in the Reconcile module. Excerpts are sampled, redacted, cost-capped, and used only to produce calibration labels; they never change a verdict. This feature is off by default.

2.3 Website and diagnostics data

  • Server logs with network address, user agent, requested page and time, kept for security and for the status page.
  • A session cookie once you sign in, a local browser preference for light or dark theme, and the record of your analytics choice. These are strictly necessary for the Service to work.
  • Product analytics, only with your permission: a first-party identifier stored in your browser, the pages and features you use, and the campaign parameters of the link that brought you here, so that we can see which pages are read and where sign-ups come from. Nothing is stored until you choose "Allow" on the cookie notice we show every visitor, and "Manage" on it lets you turn analytics on or off by itself. You can change your choice at any time from "Cookie settings" in the footer, which also deletes the identifier. We do not use advertising cookies, cross-site tracking, or session replay, and analytics data is never shared with advertising networks.
  • Error reports, when something in the Service fails, containing the error and the metadata of the request that caused it.

3. Why we process it (and our legal bases)

  • To provide the Service: creating and securing your account, recording actions, reading facts, computing verdicts, producing statements, evidence, exports and public hash verification. Basis: performance of the contract with the Customer; for Customer content, the Customer's instructions.
  • To secure the Service: authentication and email confirmation, audit logs, rate limiting, sign-up screening and abuse detection. Basis: our legitimate interest in keeping the Service and its data safe and in not opening accounts for automated sign-ups.
  • To communicate: service notices, the confirmation link and the welcome message when you sign up, verification results you subscribe to, replies to your requests. Basis: contract and legitimate interest. A short onboarding sequence after sign-up is sent on the basis of our legitimate interest in helping new accounts succeed and always carries an unsubscribe link; marketing email to anyone else is sent only with consent.
  • To bill: plan, usage counts (verified actions per month) and payment status. Basis: contract. Card details are handled by our payment provider and never reach our systems.
  • To improve the Service: product analytics as described in section 2.3, and aggregate, de-identified usage such as how many verdicts were UNKNOWN and why. Basis: your consent for the analytics identifier, given on the cookie notice and withdrawable in the footer. We do not use Customer content to train machine-learning models, ours or anyone else's.
  • To comply with the law: tax, accounting and responding to lawful requests. Basis: legal obligation.

We do not make decisions about individuals by automated means that produce legal or similarly significant effects on them. Verdicts concern the actions of software agents and the records of business systems, not people.

4. What the verdict does and does not use

A verdict is computed by deterministic rules from the contract and the facts. The agent's own account of what it did is stored as part of the record, but it is never used as evidence of what happened. No language model decides a verdict. Where a Customer uses the optional transcript assist, a language-model provider receives redacted excerpts under a contract that prohibits training on them.

5. Who we share data with

  • Service providers (subprocessors) in these categories: cloud infrastructure; DNS, certificates and edge network; identity provider; social and workplace identity providers you choose; email delivery; payments; sign-up abuse prevention; and, only when enabled, language-model providers, error reporting and product analytics. Each is bound by contract to process data only on our instructions. The purpose, data and location of each category are on the subprocessors page; Customers under a Data Processing Addendum can request the named list and receive notice before we add or replace a provider.
  • Counterparties a Customer explicitly chooses to share a statement or dispute package with, in read-only form.
  • Anyone with a hash: public verification pages confirm that a verdict or statement with a given hash exists and that its chain is intact. They do not reveal the content of the record.
  • Legal requirements: where a law, court order or enforceable government request requires it, after we have checked the request and, where allowed, notified the Customer.
  • A successor: if we are acquired or merge, the acquirer takes on this notice's obligations, and we will tell account holders beforehand.

We do not sell personal data and do not share it for cross-context behavioural advertising, and we have not done so in the preceding 12 months. We do not accept payment, referral fees, data or partnerships from the agent platforms or vendors whose claims we verify. Our independence policy is in the Trust Center.

6. International transfers

The Service is hosted in the United States. Where data of people in the European Economic Area, the United Kingdom or Switzerland is transferred to us or to a service provider outside those areas, we rely on the applicable Standard Contractual Clauses (and the UK Addendum) or another lawful transfer mechanism, as set out in the DPA, together with the technical measures in the Trust Center.

7. Retention

  • Account data: for as long as the account exists. When a workspace is deleted (section 9), its memberships, settings, keys and content go with it, within days, and we record a receipt. Your user record (name, email address, sign-in identifiers) stays, because you may belong to other workspaces; write to us and we delete it.
  • Audit logs: at least 12 months, for security.
  • Billing records: for as long as tax and accounting law requires.
  • Contact and walkthrough requests: until you ask us to delete them.
  • Email confirmation links: 24 hours, or until used.
  • Sign-up screening records: 90 days for a refused sign-up; 12 months for a sign-up that was held for review.
  • Customer content: for as long as the workspace exists. A Customer can delete a connection and its facts at any time, and can delete the whole workspace. The retention period set in the workspace (default 13 months, so that a full year of statements can be compared) is the window the Service reads for statements; automatic expiry of older records at the end of that period is not yet enforced, and this notice will say so when it is.
  • Transcript excerpts: deleted after the calibration label is produced, and never later than 30 days.
  • Uploaded files in their original form: 90 days.
  • Server logs: up to 30 days. Error reports: up to 90 days.
  • Backups: in encrypted object storage, rotated after 35 days.

8. Security

Data is encrypted in transit. Connector credentials, webhook secrets and integration tokens are encrypted at rest with a per-workspace key, itself wrapped by a master key that is not stored alongside the data. Files, evidence and backups sit in object storage that is encrypted at rest; the database sits on the hosting provider's storage volume. Connectors are read-only; CommitLayer never writes to a Customer's systems. Every verdict and statement is hashed and chained so that changes are detectable. Access to production is limited to the people who operate the Service and is logged. Current controls, and how to report a vulnerability, are in the Trust Center. To report a security issue, write to security@commitlayer.ai.

9. Your rights

Depending on where you live, you have the right to access, correct, delete, export or restrict the processing of your personal data, to object to processing based on legitimate interest, to withdraw consent where processing rests on it, and to complain to a supervisory authority (in the EU, the authority of your country; in the UK, the Information Commissioner's Office). Residents of California and other US states with privacy laws also have the right to know what personal data we hold and how we use it, to delete and correct it, and not to be discriminated against for exercising these rights; because we do not sell or share personal data, there is nothing to opt out of, and for the same reason we do not act on browser opt-out signals such as Global Privacy Control. You may use an authorised agent where the law allows, and we will verify the request through the email address on the account.

For account data, write to hello@commitlayer.ai and we will respond within 30 days (45 days under US state law, extendable once with notice). For Customer content, we forward requests from a Customer's end users to that Customer, who is the controller, and we help the Customer fulfil them.

10. Children

The Service is for businesses and is not directed at anyone under 18. We do not knowingly collect personal data from children.

11. Changes

We will post changes here with a new date and, for material changes, notify account holders by email at least 14 days before they take effect.

12. Contact

Asterfold Labs, Inc., a Delaware corporation: hello@commitlayer.ai for privacy requests and security@commitlayer.ai for security reports. By post: Asterfold Labs, Inc., c/o Legalinc Corporate Services Inc., 131 Continental Dr, Suite 305, Newark, DE 19713, USA.