FM Genie

An AI-powered assistant that turns FM documents, SLAs, and operational data into actionable insights and automated decisions.

Comment

FM Genie — Autonomous Facilities Management Agent

The Problem

Facilities management requests are often handled through forms, emails, phone calls, and manual follow-ups. A simple request such as “The AC isn't cooling in Meeting Room B” can require the employee to identify the right category, provide location details, wait for assignment, track the SLA, and repeatedly follow up for updates.

This creates unnecessary administrative work for employees and facilities teams while making it difficult to proactively identify SLA risks and escalations.

We asked:

What if an employee could simply describe the problem in their own words, and an AI agent could take care of the entire service journey?

Our Solution

FM Genie is an autonomous AI-powered Facilities Management agent that converts natural-language workplace issues into actionable service requests and manages them through their lifecycle.

The employee does not need to know the technical category, responsible team, or applicable SLA.

They simply say what is happening.

For example:

“The water tap is leaking in the 2nd floor washroom.”

FM Genie can understand the issue, categorize it as Plumbing, identify the location, determine the applicable SLA, create the service request, and route it toward the appropriate facilities workflow.

The agent can then continue supporting the request through its lifecycle — including status updates, SLA monitoring, escalation handling, notifications, resolution, and reopening of eligible tickets.

What Makes FM Genie Agentic?

FM Genie is designed as an action-taking agent, not simply a conversational chatbot.

It can:

  • Understand facilities issues expressed naturally.
  • Create service requests from conversational input.
  • Automatically categorize requests.
  • Determine applicable SLA requirements.
  • Route requests toward the appropriate facilities team.
  • Retrieve live ticket status from the ticket database.
  • Update ticket lifecycle status.
  • Monitor SLA conditions.
  • Identify SLA breaches and generate escalation notifications.
  • Handle safety-critical incidents with appropriate escalation.
  • Generate lifecycle notifications autonomously.
  • Maintain notification and activity history for each ticket.
  • Provide users with live service-request tracking.
  • Provide administrators with operational KPIs directly from the ticket database.

The user does not have to explicitly instruct the agent to perform every step of the workflow.

Autonomous Notification & Activity History

One of the key capabilities demonstrated in FM Genie is autonomous notification generation.

When a ticket reaches an important lifecycle event — such as resolution or an SLA breach — FM Genie creates a notification record associated with the ticket.

Users can see these updates in two ways:

  1. Global notification bell — showing the latest FM Genie update across all service requests.
  2. Ticket Notification & Activity History — showing the notification events associated with an individual ticket.

This provides a visible audit trail demonstrating that the agent is taking actions beyond simply responding to the user.

Safety-Critical Handling

FM Genie also distinguishes normal facilities requests from safety-critical incidents.

For example:

“There are sparks coming from an electrical panel on the 3rd floor.”

The agent can identify the request as an electrical safety incident and trigger the appropriate emergency escalation workflow instead of treating it as a normal maintenance request.

This allows the agent to adapt its behavior based on the severity and nature of the issue.

Admin Intelligence

Facilities administrators have a dedicated operational dashboard powered directly from the live ticket database.

The dashboard provides visibility into:

  • Total tickets
  • Active tickets
  • Ticket status distribution
  • Category workload
  • Team workload
  • SLA breaches
  • SLA approaching deadlines
  • Safety-critical incidents
  • Resolved and closed requests

This transforms FM Genie from a ticket-creation interface into an operational assistant for the facilities team.

How We Built It

FM Genie combines:

  • Google Gemini for natural-language understanding and agent reasoning.
  • Google ADK for building the action-taking agent and tool-based workflows.
  • Google Cloud Run for deployment.
  • Google Cloud Firestore for live ticket and notification data.
  • Python for the agent tools and backend services.
  • Flask for the application/API layer.
  • HTML, CSS and JavaScript for the interactive user interface.

The agent is connected to tools for ticket creation, ticket retrieval, ticket lifecycle management, reopening, SLA monitoring, notifications, and operational dashboard data.

What We Learned

The biggest lesson was that building an agentic application is not just about adding an LLM to an existing interface.

The important part is giving the agent meaningful tools and allowing it to interact with real application data.

We focused on making every important action verifiable through the underlying system:

User request → Agent reasoning → Tool action → Firestore data → User-visible result

This helped us avoid a common problem with AI demos where the agent appears to perform an action but the underlying system has not actually changed.

Challenges

One of our biggest challenges was maintaining consistency between the conversational agent, ticket lifecycle, SLA logic, notifications, and user interface.

We also needed to ensure that operational information such as ticket status, notification history, and dashboard KPIs came from live database records rather than static demo data.

This led us to build direct Firestore-backed views for the service-request history, notification history, global notification bell, and administrator dashboard.

The Result

FM Genie demonstrates how an AI agent can move beyond answering questions and become an autonomous operational layer for Facilities Management.

Instead of:

Employee → Form → Category → Team → Follow-up → Status Check → Escalation

FM Genie aims to make the experience:

Employee → Describe the problem → FM Genie manages the workflow

The goal is simple:

Let employees describe the problem. Let the agent manage the journey.

Built With

Share this project:

Updates

Submission history