Privacy Policy
Last updated: July 2026
This Privacy Policy explains what information Origin (“we”) collects, how we use it, and the choices you have. Origin is the evidence layer for high-consequence AI agents. It enforces runtime policy before an AI agent takes a side effect, routes every tool call through a controlled proxy, and produces a tamper-evident, hash-chained evidence package showing what an agent did, why it was allowed, and who owned the risk — so teams can get high-consequence agents through security review. Origin is currently available through a founder-led pilot program. The prototype runs; it is not production SaaS. By using Origin, you agree to this Policy.
Information we collect
- Account information — your name and email when you create an account, and authentication data handled by our identity provider.
- Agent workflow metadata — the agent workflows, policies, scopes, budgets, and approval rules you configure so the Service can enforce what an agent is allowed to do.
- Tool-call requests and responses — in a pilot deployment, the proposed and executed tool calls that pass through a controlled proxy, including their parameters and results, so the Service can evaluate and record each side effect.
- Policy decisions — the verdict returned for each proposed action (allow, cap, require-approval, deny, or pause) and the inputs that produced it.
- Approval metadata — the named risk owner, the approval or denial, and the time it was recorded when a risky action is escalated to a human.
- Audit logs — the append-only, hash-chained record of every proposal, verdict, approval, action, block, and exception that forms the evidence package.
- Customer-system identifiers — references to the tools, systems, and resources your agents connect to (for example resource names or account identifiers) needed to enforce scope and produce the record.
- Usage and security logs — basic logs, IP address, and diagnostics needed to operate and secure the Service, plus aggregated usage analytics.
- Contact and lead information — details you submit through a form or an evidence-review request, such as your name, work email, company, role, and the workflow you describe.
How we use information
- To run the deterministic reference check — binding an agent configuration and returning deterministic verdicts over a scenario battery. (A runtime policy gate evaluating live proposed actions is proposed pilot architecture, not a deployed capability.)
- Where a controlled tool-call proxy is used in a pilot, to record what passed through it. The proxy is proposed pilot architecture and is not deployed as general production enforcement.
- To produce the tamper-evident, hash-chained evidence package showing what an agent did, why it was allowed, and who owned the risk.
- To operate, maintain, and secure the Service and your account.
- To surface human-in-the-loop controls — human approval for risky actions, scope and budget limits, and kill switches to pause a workflow or revoke a scope.
- To communicate with you about your account, security, and changes to the Service.
- To detect, prevent, and respond to abuse, fraud, and security incidents.
Human-in-the-loop controls
Risky actions do not execute silently. When a policy decision returns require-approval, the action is held at the proxy and routed to a named risk owner, whose approval or denial is recorded in the audit log alongside the action. Scope and budget limits bound blast radius before an action runs, and a kill switch can pause a workflow or revoke a scope without redeploying the agent. These controls are part of the record: the evidence package reflects who owned the risk and what they decided.
Append-only, hash-chained audit logs
The audit log is append-only and hash-chained, so the record of what the Service evaluated, allowed, and recorded is tamper-evident and replayable. Its purpose is to provide evidence and accountability for each agent action. Audit access is scoped to the workflow it belongs to and subject to the applicable controls.
How we share information
We do not sell your personal information. We share it only with: (a) sub-processors that help us run the Service (for example, hosting, identity, and infrastructure providers) under appropriate confidentiality and security obligations; (b) authorities when required by law; and (c) a successor in the event of a merger or acquisition, subject to this Policy. A current list of sub-processors is available on request and is referenced in the Data Processing Agreement (DPA).
Security
We use technical and organizational measures designed to protect your information, including least-privilege (need-to-know) access, encryption in transit, and append-only, hash-chained audit logs of policy and tool-call activity. Access to workflow data and audit logs is scoped to the workflow it belongs to and subject to the applicable controls. Origin is responsible for the software controls it provides; you remain responsible for the agents and tools you connect. No method of transmission or storage is perfectly secure. A DPA and a security review are available on request.
Data retention and deletion
We retain information for the life of the engagement and as required by law. Audit logs are append-only and may be retained as the evidentiary record of policy and tool-call activity. Retention windows applied during the pilot are defaults; the finalized retention terms are set in the DPA. To request removal of your account or associated data, contact us using the details below.
Your rights
Depending on your location, you may have rights to access, correct, export, or delete your personal information, and to object to or restrict certain processing. To exercise these rights, contact us using the details below. You may also have the right to lodge a complaint with your local data-protection authority.
Cookies and analytics
We use only essential cookies and local storage needed to keep you signed in and operate the Service. We also use Google Analytics configured in Consent Mode v2 with analytics_storage denied by default, so no non-essential analytics or advertising cookies are set unless you grant consent.
Children
The Service is not directed to children under 16, and we do not knowingly collect their personal information.
Origin's role
Origin provides policy enforcement, evidence, and decision-support records — not legal or compliance certification. We use “tamper-evident” to mean alteration is detectable by replay and digest checks, and “review-ready” to mean built for review rather than already accepted. The customer remains responsible for the agents and tools they connect and for the decisions made using Origin's records.
Changes
We may update this Policy. If we make material changes, we will take reasonable steps to notify you and update the “Last updated” date above.
Contact
Questions about privacy? Reach us at bohueilin@gmail.com or through the contact section of our website.