Internal AI project

Office Occupancy Agent

An internal AI agent, built end to end, that answers real questions about how office space is used. Shipped as a reusable consulting use case.

Office Occupancy Agent
Client
Internal AI project
Role
Architect & Builder
Timeline
2025
Scope
Agentic System

The challenge

Occupancy data sits in one system. Identities sit in another. People want answers in the tools they already use. The job was to connect all three into an agent that is secure, reusable, and genuinely useful inside Teams.

The approach

  • Designed the end-to-end architecture of an AI-supported office occupancy agent in Copilot Studio, as a reusable consulting use case.
  • Built a scalable backend on Azure Functions.
  • Integrated real-time occupancy data through the Aruba Controller API and identities through Microsoft Entra ID.
  • Built the Copilot Studio agent with dialog-based topics and flows for occupancy, presence patterns, and trends, delivered through Teams and Copilot.
  • Handled secure distribution of the agent across the Microsoft ecosystem.

The outcome

A working agent that people query in Teams, and a pattern the team can reuse for the next occupancy or infrastructure use case.

Client names are generalized. The work is real; the labels are not.
All works ↗
Computer Vision for Automated Billing
Next — Logistics operator

Computer Vision for Automated Billing

View case study ↗