In-House Graph Broker for ABM SharePoint AI
Built ABM's internal AI governance program and an in-house Graph Broker in front of SharePoint AI tooling, so the Apps team could ship without native Microsoft Graph permissions that would have over-permissioned them.
The Challenge
ABM wanted SharePoint AI tooling in production, but native Microsoft Graph permissions would have given the Apps team more access than Security Operations could accept.
- 1SharePoint AI and Copilot-style tooling needed Graph access to be useful
- 2Native Microsoft Graph permission models were too coarse for the Apps team
- 3Blanket Graph access would have over-permissioned application owners
- 4Internal AI work needed a governance program, not a policy slide
- 5Security Operations needed a control plane they could operate
Our Approach
Stood up an internal AI governance program and built a tailored in-house application that sits in front of SharePoint AI as a risk-adjusted Graph Broker.
Internal AI Governance
Defined who can use SharePoint AI, what data it may touch, and how Security Operations reviews new AI use before it reaches production.
Graph Permission Risk
Mapped the native Microsoft Graph scopes the Apps team would have received and marked the ones that over-permissioned application owners.
In-House Graph Broker
Built a tailored application that brokers Graph calls for SharePoint AI, granting only the risk-adjusted access each workload needs.
Security Operations Handoff
Put the broker in the path that internal SecOps can run, so AI tooling ships without handing the Apps team a blanket Graph token.
The Results
SharePoint AI tooling shipped behind a risk-adjusted Graph Broker. The Apps team was no longer over-permissioned, and internal Security Operations had a control plane instead of a slide deck.
Adil rebuilt our AI governance program and shipped an in-house Graph Broker in front of our SharePoint AI tooling. Native Microsoft permissions would have over-permissioned the Apps team; that was a real win for Security Operations.
Related Case Studies
Building an AI Governance Program from Zero
Discovered 47 shadow AI tools, implemented NIST AI RMF alignment, achieved ISO 42001 readiness, and reduced AI-related incidents by 85%, helping secure a $180M Series D.
Securing Autonomous AI Agents
Architected runtime guardrails and zero-trust identity for autonomous agents, enabling aggressive AI adoption while preventing model hallucination and goal hijacking.
Intune MAM for 5,000+ Devices During the Landmark and Optum Integration
Designed, implemented, tested, and rolled out Microsoft Intune mobile application management for 5,000+ Landmark devices during the Optum Health integration, so corporate apps could be used without opening unmanaged BYOD risk.
Ready to Achieve Similar Results?
Thirty minutes is enough to see if the same approach fits your next audit, raise, or AI rollout.
See if I should be in the room