Security for contract data your organization cannot afford to mishandle.
Revver processes vendor agreements, negotiating positions, company standards, decision history, and the institutional knowledge those decisions create. We built the platform around the assumption that this information is confidential.
This page explains the security properties of Revver at a level appropriate for public review, how customer data is separated, how AI is used, how access is governed, what you keep control of, and which assurance work is complete or still in progress.
We label the difference between what exists today and what is still being built.
Last updated August 2026.
The controls that matter first.
Security reviews tend to start with the same questions. These are the answers we can state publicly and precisely today.
| Security property | Status |
|---|---|
| Customer data used to train shared AI models | No |
| Customer and workspace isolation | Available today |
| Role-based access control | Available today |
| Human approval before any vendor-facing output | Available today |
| Customer-data export | Available today |
| Decision attribution and audit record | Available today |
| Encryption in transit and at rest | Available today |
| Single sign-on (SSO / SAML) | Enterprise option |
| Dedicated single-tenant deployment | Enterprise option |
| Independent security assessment | Scheduled |
| SOC 2 examination | Planned |
A contract is only the beginning.
Revver can hold more than the document itself. A customer workspace may contain contract language, commercial terms, findings, remediation, internal negotiating positions, company standards, accepted exceptions, reviewer decisions, corrections, Review Receipts, and the customer-specific institutional knowledge built over time.
Some of that is more commercially sensitive than the contract itself. That’s why the workspace, not the uploaded file, is the security boundary.
The platform is designed to protect the document and the judgment accumulated around it.
Your organization’s knowledge stays yours.
Every request is scoped server-side to an authorized workspace, and reaching a workspace requires a membership. Contracts, findings, standards, decision history, and customer-specific institutional intelligence stay within that customer context, and they never become another customer’s working knowledge.
Revver’s shared System Standards are maintained separately from customer-specific knowledge. Your contracts and decisions don’t become another customer’s standards.
Your organization benefits from what Revver learns about you without teaching Revver how another customer should buy.
Implementation-level isolation and data-flow detail is available during security diligence.
Your data is used to do the work, not to train a shared model.
Revver uses third-party AI models as reasoning infrastructure. Your contracts, findings, decisions, and customer-specific institutional intelligence aren’t used through Revver to train shared foundation models.
What the platform learns from your decisions is stored as customer-specific, human-readable knowledge in your workspace, not trained into model weights. The reasoning model can change without becoming the permanent home of your accumulated judgment.
The model can change. Your organization’s knowledge stays its own.
Revver uses a limited set of cloud, infrastructure, and AI providers. The current subprocessor list, processing purposes, and provider-specific retention detail are available to security reviewers during diligence, because they vary by provider and deployment.
Contracts are data, not authority.
A vendor’s document is untrusted input. Revver evaluates the contract’s language as content, in a separate channel from the system’s own instructions, so text inside an uploaded file doesn’t gain authority over what the platform is allowed to do.
The model’s output has no unrestricted commercial authority either. Revver can analyze, recommend, prepare remediation, and develop negotiating positions, but material commercial decisions stay attributable to people, and nothing reaches a vendor because a model decided it should.
Findings stay tied to the contract language behind them, so the model’s output can be inspected rather than accepted as an unexplained conclusion.
Our AI-security architecture and testing practices can be reviewed with your security team.
Access is explicit, limited, and attributable.
Revver controls access at the account and workspace level. Reaching a workspace requires authorization, and a role determines what a user can do inside it.
Meaningful actions on contracts, evaluations, and decisions stay attributable to the people who performed them, so the record of a review shows whose judgment it reflects.
When Revver personnel or advisors take part in an engagement, they reach the workspace through the same membership boundary that protects everyone else, limited to the work they’re assigned, and their access can be revoked at any time.
Human expertise doesn’t mean unrestricted access.
Customers with specific access-governance requirements can review the permission model with us during diligence.
Encryption, infrastructure, and recovery.
All traffic to Revver is encrypted in transit over TLS, and customer data is encrypted at rest using the controls of the managed cloud infrastructure Revver runs on. That infrastructure is independently examined against SOC 2 Type 2 and operated to recognized cloud-security and privacy standards. Revver’s own independent assessment and SOC 2 status are separate, and stated below.
The platform is operated from the United States and its territories, and customer data is processed there. It’s backed up with point-in-time recovery, so the platform can be restored following disruption.
Detailed infrastructure, key-management, recovery objectives, and deployment architecture are reviewed with qualified customer security teams rather than published as a blueprint.
Security requirements don’t stop at the standard deployment.
Single sign-on (SSO/SAML) through your identity provider is available as an enterprise option, so sign-in follows the authentication controls your organization already enforces, including multi-factor.
For stricter isolation, a dedicated single-tenant deployment, including an AWS environment with model processing through Amazon Bedrock, can be scoped with your security team. We build that environment when your requirements call for it rather than maintaining a one-size demo of it. The exact architecture, model-processing path, and trust boundaries are reviewed with your technical team during diligence.
If your review needs this tier, it becomes part of the engagement.
We separate controls from proof of those controls.
A security program and independent assurance aren’t the same thing. Revver states both, so you can tell what exists today from what has been independently examined.
Independent security assessment. Revver runs internal security reviews of the platform, and an independent, third-party assessment is scheduled. We’ll describe its scope and share what’s appropriate under NDA once it’s complete.
SOC 2. Revver plans to pursue an independent SOC 2 examination. No examination has been completed, and we won’t represent one as complete before it is. This page is updated as the status changes.
Security includes what happens when the relationship ends.
Revver is built so the institutional knowledge you create doesn’t get trapped in the platform. Your contracts, evaluations, Review Receipts, and standards can be exported, and each Review Receipt carries the record of the decisions behind that review. Your customer-specific intelligence stays outside the underlying model’s weights.
Retention, deletion, and termination handling vary by customer, deployment, and legal obligation, and are addressed with your security or privacy team and in the applicable agreement. Documents and workspaces can be deleted outright.
The knowledge is yours either way.
Enough to evaluate us. More when you need to verify us.
A public security page should tell you whether a vendor has thought seriously about your data. It shouldn’t pretend to replace a technical review of your environment, deployment, contractual requirements, and risk profile.
Revver is an early company building for organizations that put a platform through real security review, and we’d rather give you an accurate “in progress” or “not yet” than dress a roadmap item up as a shipped control. This page is kept current as that work progresses.
If your IT, security, privacy, legal, or vendor-risk team needs to go deeper on our architecture, data flows, model processing, subprocessors, access controls, deployment options, assurance work, retention, or secure-development and vulnerability-management practices, reach the people who built the system directly.
We’ll work through the questions with the people responsible for the system, including a direct “not yet” when that’s the honest answer. Security questionnaires are welcome.