FM·PAM AI Companion
An AI copilot for facilities and property management. FM·PAM sorts tenant requests, drafts work-order closeouts, checks vendors, tracks compliance and reads leases, so managers can spend their time on the buildings.
FM·PAM has been selected for the Top 80 of the NVIDIA Inception Grand Challenge 2026. In the next evaluation round we'll present a focused, pilot-ready enterprise proposal built on this platform.
Too much to track, too little time for the buildings
Facilities management and property/asset management teams have too much to keep track of. Tenant requests come in by phone, email and chat. Work orders sit in a CMMS. Vendor credentials expire without anyone noticing. Compliance evidence is spread across folders, and lease deadlines are buried in PDFs. Managers spend hours each week sorting, chasing and reporting. That leaves less time for the buildings themselves.
An AI copilot for building operations
Web Operations Dashboard
Next.js
The command centre for property and facilities managers.
Mobile App
Flutter
Three personas in one app: Client/Tenant, Field Technician/Vendor and Property Manager.
Backend
FastAPI · PostgreSQL · Redis/Celery
Hosts the AI modules and connects to each client's systems.
Ten modules, one copilot
Smart Triage Inbox
Classifies incoming requests by priority, category and confidence, then creates the work order. When it's unsure, it escalates instead of guessing.
Work Order Closeout
Turns a technician's voice notes, photos and notes into a structured closeout summary. A person approves it before anything is written to the CMMS.
Vendor Communication Bot
Drafts dispatch and follow-up messages, and checks the vendor's insurance and certifications on the server before a dispatch is approved.
Compliance Evidence Collector
Tracks inspection readiness across categories and raises alerts as deadlines get closer.
Lease Intelligence Assistant
Extracts expiries, break options, CAM caps and key clauses from leases, then sends 90/60/30-day alerts. A person confirms every extraction.
Portfolio Status Copilot
Answers questions in plain language using only data pulled from the work-order system.
Mobile Field Inspection
Runs checklist-driven inspections. Failed items create urgent work orders automatically.
Exception Detector
Flags repeat repairs on the same asset and emergency backlogs before they grow.
Coaching & Manager Dashboard
Shows AI adoption, actions by module, confidence and time saved, plus weekly coaching tips for each user.
AI Calling Agent
Handles tenant phone calls through Twilio Voice and ElevenLabs. It classifies urgency, logs the call and turns it into a triaged ticket.
Vendor-agnostic by design
Every enterprise runs a different stack. The AI provider (Claude, Azure OpenAI, AWS Bedrock or a private in-house LLM), the CMMS, telephony, document store and identity provider each sit behind their own adapter interface. To connect a new client system, you change configuration instead of code. The whole platform can also run offline in a self-contained demo mode.
Human in the loop
Closeouts, lease extractions and vendor dispatches are drafted by AI and approved by people.
Conservative AI
Triage escalates when confidence is low, and the copilot only reports numbers it actually retrieved.
Auditability
Every AI-assisted action is logged and feeds the adoption and impact reporting.
See it in action
Weekly coaching summary
Each user sees their AI-assisted actions, estimated time saved and modules they haven't tried yet.
Smart triage queue
Incoming requests are ranked by priority, category and AI confidence. Managers can correct any classification with one click.
Exception monitoring
Repeat repairs and emergency backlogs are flagged automatically.
AI calling agent
Each tenant call is logged with its urgency and category tags.
Workflow impact report
Actions, users and average confidence are broken down by module.
Screenshots are from the platform's demo environment and use sample data.
Pilot FM·PAM in your portfolio
For the next round of the NVIDIA Inception Grand Challenge, we're preparing a pilot proposal for one enterprise direction: the client problem, deployment assumptions and measurable pilot outcomes. Our goal is a realistic pilot, measured on real operations data. If you manage facilities or a property portfolio, we'd like to talk.
We'd like to thank the NVIDIA Inception team, and congratulations to our fellow Top 80 startups.