Skip to content

Get started

Enterprise AI

Govern models, connectors, workflows and data controls across Pervaziv AI.

Enterprise AI connects organizational information and approved actions to the Cortex experience. It brings engineering systems, collaboration tools, cloud services and business records into work that begins in Developer Tools, Discover, Browser Extensions, Mobile or CLI.

Review the supported MCP connectors, authorize a service through connector setup, and confirm the permitted operations. Administrators can then configure roles and governed actions for the intended users.

Enterprise AI exposes ten MCP connector families. Each connection makes supported service tools available to an eligible session using the approved identity and scopes.

Product area Connections Context and work
Engineering and delivery GitHub, Atlassian, Azure DevOps Source, pull requests, issues, Jira planning, Confluence knowledge and delivery information
Collaboration and productivity Slack, Microsoft 365, Google Workspace Conversations, mail, calendars, documents and files
Cloud operations AWS, Microsoft Azure, Google Cloud Authorized infrastructure, resource and operational context
Customer and business context Salesforce Accounts, contacts, opportunities, cases and supported record operations

The connector reference describes the services behind each family. Available tools depend on service authorization, resource permissions, account entitlements and the active workflow.

An engineering task often crosses systems. A Jira ticket may need Confluence requirements and a GitHub change; an incident may need a Slack discussion and cloud resource information; a Salesforce case may need to be related to existing engineering work.

Enterprise AI lets those sources contribute to the same objective. Cortex can retrieve eligible context, explain it and invoke approved operations. Reads and writes are controlled separately, so a service can be useful for research even when changes require additional approval.

Model, search and skill routing help match the request to the available intelligence, current information and engineering guidance. Planning identifies the steps and approval points. Cortex Connect preserves supported continuity between clients, and Cortex Cloud supplies managed execution for eligible project work.

See how Enterprise AI works for the relationship between context, connected tools, planning and results.

Organizations can govern identity, roles, entitlements, model profiles, connector access, retention, execution targets and approvals. The effective options shown to a user reflect the controls for that account and task. Service credentials authorize operations without becoming conversation text.

Provider permissions still apply to each resource. A connector does not grant every user access to every repository, channel, mailbox or cloud account. Review security and privacy when choosing a connection and retention policy.

Connect customer, engineering and operational work

Section titled “Connect customer, engineering and operational work”

Enterprise AI is valuable when the answer spans more than one system. A support case can describe customer impact, a Jira issue can contain delivery intent, Confluence can explain the requirement, and a GitHub pull request can show the proposed implementation. Cortex can use the authorized portions of those sources in one task rather than asking the user to reconstruct them manually.

For Salesforce, supported operations include discovering available objects and fields, querying permitted records and preparing explicitly requested record changes. Custom CRM schemas can be explored before a focused query is constructed. This helps relate cases, accounts and opportunities to engineering work without treating customer access as unrestricted database access.

Collaboration and productivity connections add discussions, documents, mail and calendars where permitted. Cloud connections add operational resources. The ten connector families are separate authorizations: connecting GitHub does not connect Slack, Salesforce or a cloud account automatically.

Coordinate expertise without managing every model

Section titled “Coordinate expertise without managing every model”

Model Routing matches eligible requests to allowed intelligence. Search Routing can bring current public information into research, such as recent SDK guidance or an advisory. Skill Routing contributes engineering practices for testing, implementation, security analysis, security validation and verification.

Cortex Planner structures larger objectives around dependencies, acceptance conditions and review points. A request to fix an issue may require discovery, implementation, regression tests, security review and verification. The plan makes those responsibilities visible; it does not grant new service permissions.

Long-running work accumulates files, tool results, decisions and validation records. Cortex’s context management is designed to preserve current instructions and important constraints while reducing eligible repeated material and retaining access to authoritative sources.

When a decision changes, the latest direction matters. A compacted summary or earlier conversation is not authority to revive a discarded requirement. New service state, changed source and permission changes can require fresh context. Users should inspect the supporting source when a decision depends on exact details.

Start with the services needed for a concrete workflow and verify read access. Then decide which writes are permitted, which require approval and who can authorize them. Review data handling and retention before bringing sensitive records into a conversation or export.

Use organization roles and effective account settings to define access rather than expecting a connector to bypass provider controls. For managed project execution, also review the allowed destination, capacity and validation scope. This separates connected information, authorized external action and project execution into decisions administrators can evaluate.

Enterprise AI provides the connected service and governance capabilities used across Cortex clients. Developer Tools focus on the project, Browser Extensions bring in web context, and Mobile provides conversation and task review away from the desk. DevSecOps provides the project security and delivery console.