In this live demo, Nico showcased the seamless integration of Proxima as an AI application server with MCP (Model Context Protocol), highlighting how role-based data access and guardrails can be enforced effortlessly within enterprise systems.
Setting Up Proxima with MCP
He started by configuring Cursor, the front-end client, to use Proxima as the MCP service. With a simple JSON file containing the endpoint URI and API key, he connected Proxima and immediately began querying enterprise data with the tool get answer for user.
Smart Data Access Based on User Roles
- Bob (Executive Access): When Bob asked for business performance metrics and corporate strategy, Proxima correctly recognized his executive privileges and returned the requested confidential data.
- Alice (Individual Contributor): The same query from Alice was denied, showing that guardrails were working as expected to restrict access based on user roles.
Fixing Data Entitlement Gaps
To simulate a common enterprise error, Alice was given access to a sensitive SharePoint file. When she queried for the highest revenue customer, she received the answer, revealing a policy gap.
Proxima’s policy engine quickly remedied this:
- A new rule was applied to block access to PII and financial data for all individual contributors.
- Repeating the same query now resulted in a denial, even though the file access still existed.
- This demonstrated Proxima’s ability to override faulty entitlements with data-level guardrails.
Built-In Audit Trails
Proxima automatically records:
- The user prompts
- The output generated
- The data lineage (for example, which SharePoint PDF was accessed)
This makes compliance and security audits seamless and transparent.
Final Thoughts
This demo illustrated how Proxima and MCP work together to:
- Connect AI tools to enterprise data
- Enforce identity-aware access policies
- Provide complete audit visibility
And the most powerful part? It all works through intuitive configurations without complex development.