Security

Enterprise-Grade Security Architecture

United Intelligence is built for organizations where data governance, auditability, and access control are operational requirements — not afterthoughts.

Permission Model

United Intelligence operates with a user-scoped integration model. Every query is executed within the permissions of the requesting user — there are no elevated access patterns, bulk data exports, or administrative shortcuts.

The integration honors Salesforce sharing rules, profile-based visibility, and field-level security. If a user cannot see a record or field in Salesforce, they cannot access it through United Intelligence.

This approach ensures that data access governance already established in your CRM extends seamlessly to AI-assisted analysis. There is no secondary permission layer to manage or audit.

Auditability

Every output generated by United Intelligence includes explicit provenance metadata:

  • Record counts included in the analysis
  • Filters and parameters applied to the query
  • Timestamps of data retrieval and output generation
  • Scope of the query (objects, date ranges, account boundaries)
  • User identity and permission context at time of execution

This metadata is embedded in every response, making outputs suitable for compliance reviews, internal audits, and executive reporting without additional documentation effort.

Audit trails are designed to answer the question: What data was this conclusion based on, who requested it, and when?

AI Boundary Definition

United Intelligence maintains a clear separation between three distinct processing layers:

Deterministic Data Logic

Aggregations, counts, averages, variance calculations, and threshold comparisons. These produce mathematically verifiable results from source records.

AI Summarization Layer

Natural language interpretation of deterministic results. Summaries are generated from verified calculations and always reference their source metrics.

Recommendation Drafting

Suggested actions based on pattern analysis. Recommendations are clearly labeled as AI-generated and separated from factual data summaries.

This separation ensures that stakeholders can distinguish between what the data shows (deterministic) and what the AI interprets or recommends (generated). Each layer is identifiable in the output structure.

Data Handling

United Intelligence is designed around a principle of minimal data retention:

  • No unnecessary storage of Salesforce records outside of active query execution
  • Configurable logging policies that align with client data governance requirements
  • Query results are generated on demand — not pre-cached or warehoused
  • Infrastructure is built to operate within client-controlled environments where required
  • No data is shared across client boundaries or used for model training

For organizations with specific data residency, retention, or sovereignty requirements, United Intelligence can be configured to operate within designated infrastructure boundaries.

Security FAQ

Does United Intelligence require elevated admin access?

No. Every run enforces the requesting user's existing Salesforce profile, sharing rules, and field-level permissions.

Can outputs be used for audit and compliance reviews?

Yes. Each output includes run metadata including record counts, filters, timestamps, scope, and user context.

Is client data used for model training?

No. Workflows are designed with minimal retention and do not share data across client boundaries or model-training pipelines.

Have specific compliance or security requirements to discuss?

Book a 30-Min Workflow Session