AI & External Connections

Outrevio deliberately separates deterministic CRM intelligence, internal AI assistance, and external AI applications. They do not share one blanket permission.

• Deterministic relationship intelligence uses Outrevio rules and CRM evidence to calculate networking signals such as relationship strength, momentum, priority, opportunity, and ask readiness. It does not require an external language model.
• Internal AI assistance may use OpenAI, Anthropic Claude, or Google Gemini only after an explicit user AI request and only with the data categories the user has enabled.
• External MCP applications such as ChatGPT are separately authorized through OAuth and receive only the scopes approved by the user and allowed by Outrevio's server-side policy.

Relationship, opportunity, priority, momentum, and readiness signals describe networking context from your CRM. They are not measures of a person's worth, honesty, character, protected traits, or suitability for employment, credit, housing, insurance, education admission, or another regulated consequential decision.

Scores may change when CRM evidence changes and can be incomplete when the account has limited or outdated relationship history. They should be used to decide how the account holder manages their own networking follow-up—not to decide another person's eligibility, rights, or access to an opportunity.

Outrevio is not designed or authorized to make, recommend, rank, or materially influence regulated eligibility or consequential decisions about another person. Do not use Outrevio AI, relationship intelligence, or MCP data for:

• Employment eligibility, hiring, firing, or promotion
• Credit or lending eligibility
• Housing or tenant eligibility
• Insurance eligibility or pricing
• Education admission or enrollment eligibility
• Another regulated eligibility or consequential decision about a person

Internal AI prompts are instructed not to make those decisions or infer protected/highly sensitive traits from CRM evidence. This product restriction does not turn Outrevio into a compliance system for regulated decision-making; a separate legal, security, and product review would be required before Outrevio entered one of those use cases.

Keep a human in control: verify important facts, review generated text before use, and do not treat scores or generated recommendations as objective facts about a person.

AI-generated drafts remain editable. Current AI write capabilities are restricted to a small allow-list of proposals, and each supported CRM mutation still requires explicit confirmation before it is saved. Sending communications, deleting, bulk exporting, changing security settings, and changing permissions remain outside the AI action boundary.

Turning on the AI Assistant does not grant unrestricted CRM access. Users choose the data categories internal AI may use, including a separate Meeting transcripts permission for provider-generated Zoom or Microsoft Teams transcript text, and Outrevio checks those permissions server-side before provider invocation. External provider calls are user-initiated; page loads, transcript imports, transcript viewing/search, deterministic scoring, deterministic recommendations, and background retention jobs do not automatically invoke those providers.

Outrevio does not operate its own model-training pipeline using CRM content. Third-party AI providers process requests under their own applicable service terms and privacy/data-processing commitments, so users should avoid placing unnecessary sensitive information in CRM content or AI requests.

Zoom and Microsoft Teams transcripts are provider-generated communications data, not instructions to Outrevio or to an AI provider. Outrevio treats transcript text as untrusted evidence. Importing, viewing, searching, synchronizing, or retaining a transcript does not automatically call an external AI provider.

A transcript is sent to the user-selected internal AI provider only after the user explicitly chooses Analyze transcript and has enabled the separate Meeting transcripts AI data permission. AI-generated meeting summaries and suggested CRM actions are review-only output. The user chooses the real CRM contact for each included item, can edit or remove suggestions and add missing notes, reminders, allowed contact-field updates, or a pipeline-stage change, then prepares one short-lived reviewed proposal and confirms that proposal separately. Preparing or confirming the reviewed proposal does not call the external AI provider again.

Outrevio does not infer a CRM contact identity from a speaker name alone and does not allow meeting AI to change email, phone, pipeline stage, relationship score/type, opportunity tags, protected traits, or inferred sensitive attributes. Expired transcript text is unavailable to meeting AI.

Connected AI apps can access only the Outrevio information you approve. They are separate services with their own terms and privacy practices, and you can disconnect them from Outrevio at any time.

CRM text returned through MCP is explicitly marked as untrusted data so notes or imported text cannot become authorization instructions. External applications are separately operated services; users should review the external provider's terms and privacy practices before authorizing access and disconnect access they no longer want.

Review the Terms of Use for binding acceptable-use restrictions and the Privacy Policy for current data-handling disclosures.