Clear controls for customer data and AI actions.
Review what a connected assistant can do, which records an analysis can reveal, and how your team can stop access. These are the controls in place today.
Read the security and administration guide →Reviewed October 2026
Scoped, auditable AI connections
MCP connections use explicit scopes for CRM, tasks, meetings, knowledge, and search. OAuth grants act as the approving member, and API keys act as their creator. Vesso checks the credential, workspace, and actor before running tools.
Members can revoke their own OAuth connections; workspace administrators can revoke connections across the workspace. API keys are revoked separately. Tool calls and OAuth lifecycle events are audited, with redacted argument summaries instead of free-text payloads.
MCP scopes, audits, and revocation →Hierarchy enforced in Data Explorer
Data Explorer requires effective read permission and applies ownership and reporting hierarchy to query results. Related records inherit the visibility of their parent or account; workspace-shared sources are classified separately.
These checks apply to visual queries, AI-generated queries, source statistics, and Custom Reports. A source without a visibility classification is withheld from restricted users. Query cost limits also bound joins and row processing.
Data Explorer visibility →Sensitive analytics fields require an administrator
Within Data Explorer and its reporting surfaces, meeting transcripts and summaries, recording URLs, event descriptions, task comment content, and form responses require workspace administrator access.
These fields are removed from non-admin schemas and results. Explicit references are rejected in queries and report grouping or detail requests. Access in the original product surface is governed by that surface’s own permissions.
Sensitive-field boundaries →Separate source content from AI instructions
AI workflows that process agreement text, web research, workspace search passages, and external agent triggers label source material as untrusted data and instruct the model to ignore embedded commands.
Document and trigger boundaries are escaped or sanitized in the applicable prompts. Action guardrails and server-side permissions provide additional checks. These layers reduce prompt-injection risk; they do not make AI output infallible. Review proposed changes and outbound content before acting.
Honor recipient preferences across the workspace
Full email opt-outs are shared across the workspace, so changing reps does not bypass suppression. Category preferences are checked for matching sends, and full unsubscribes end active cadence enrollments for the address.
Supported outreach sends include signed preference links and one-click unsubscribe headers. Administrative removal from the unsubscribe list requires a workspace administrator. These controls support a team’s compliance process without replacing its responsibility for consent, content, and applicable rules.
Unsubscribe and suppression controls →Security fundamentals and review
Vesso uses TLS in transit and encryption at rest in AWS-managed storage. Workspace roles, multi-factor authentication, and administrator exports support everyday control. Enterprise plans add OIDC single sign-on; SAML and SCIM are not supported today.
The SOC 2 audit is not yet complete. Contact us to discuss current controls, plan availability, and your security review requirements.
Discuss a security review →