WhiteGlove AV — Overview

WhiteGlove AV is a conference room that recognizes who is present, reverse-looks-up their Microsoft Entra identity, reads their department and job title, and auto-loads the matching AV scene, turning command-based control into context-aware assistance. We let AI see, and we let the automation platform decide and act: recognition is the only probabilistic input, and everything downstream is a reproducible, fully-audited automated workflow, so every choice the room makes carries a replayable audit trail.

The platform underneath is swamp. swamp runs the room's logic as declarative, step-by-step workflow definitions, executes them deterministically, and records every step so each decision can be replayed and accounted for after the fact.

What it does

A person walks into a room, and the room sets itself up for the work in front of them without anyone touching a panel. A sales account executive gets a video-call scene with speaker framing and warm lighting; a marketing manager gets a presentation scene with wide framing and content on the main display; an executive gets a briefing layout. The same thing happens unattended when a meeting is booked on the room, because a Microsoft Graph calendar change drives the identical decision policy that a person at the door would.

How to think about it

The architecture page walks through the full flow from recognition to audited action, and the integrations page covers each platform the room depends on. The short version is that recognition sees, swamp correlates and decides, and the AV control layer acts, with swamp holding the decision and the audit in the middle so the room's behavior is deterministic and provable rather than a model's guess.

The deterministic spine is the part that matters. Given an asserted identity, swamp does a live Microsoft Entra lookup, decides a scene, enacts it on a Cisco endpoint, and writes the audit, on every run. The always-on calendar path runs the same way end to end, driven by a Microsoft Graph change notification rather than a person at the door. Recognition runs on a browser face bridge alongside badge and QR input, all resolving to the same kind of identifier, so the room reacts the same way no matter how a person is recognized.

The documents and the deck

The architecture and operations write-ups are published alongside this overview on the Resources page, and the full presentation deck is available there as a download. Read the deck for the story, the architecture for the system, and the operations page for how it stays dependable.