About this article

• Audience: Operators, integrators, and partners evaluating the central VMS
• Goal: After reading, you understand Phantom Vision VMS responsibilities, benefits, and how it pairs with Shadow devices
• Type: Reference

Summary

Phantom Vision VMS is the central fleet back end: registry, MQTT observability, justified clip playback, readiness / remediation, MCP forwarding, and owned cloud video escalation. Edge Shadows stay sovereign (no cloud API keys on-camera). The VMS is the single place operators pull recordings and escalate justified clips for richer temporal analysis.

Product videos

Product benefits

Loading stream…

Operator tour (~89s)

See Phantom Vision in action — fleet health, Specter actions, live detections, forensic pull, and safe repair on real product UI.

Loading stream…

Full film pack: Product promo videos.

What the VMS owns

Surface	Role
Fleet registry	Shadow identity and role labels (site-role-a / site-role-b)
MQTT tap	Latest telemetry and health aggregation
Playback-ready	Justified chunk discovery and cached MP4 play URLs
Pull / transcode	Secure pull of edge recordings for review or cloud upload
Readiness	Module grades and approved remediation playbooks
MCP / A2A	Tool forwarding and peer mediation behind zero-trust boundaries
Cloud escalation	Gemini File API and Video Intelligence — credentials stay on the proxy host

Default operator base URL pattern: https://<phantom-proxy>:8788 (see API reference).

Benefits (aligned with the video)

1. Sovereign edge, capable center — Cameras never hold cloud secrets; VMS owns uploads and API keys.
2. Justified-only cloud cost — Escalation runs on justified ~60 s clips, not continuous raw uplink.
3. Precomp first — On-device RF-DETR remains authoritative; cloud adds temporal NL and structured labels.
4. Operator automation — Specter agents, readiness remediations, and fleet grade gates sit on the host stack.
5. Fleet-wide observability — One proxy for registry, MQTT, playback, and TAP-friendly curls.

Architecture position

Edge Shadows (Shadow 1 / Shadow 2)
  → justified H.265 + MQTT
  → Phantom Vision VMS (this product)
  → optional cloud video APIs
  → fused results to MQTT / TAP / A2A

Concept pages:

• Architecture overview
• Cloud video escalation
• Video understanding (operator)
• Phantom / Cost Sentinel usage

Operator entry points

Task	Start here
Host stack up / down	Host / Phantom stack
Fleet soaks and gates	Soaks & automation
Playback and prompts	Video samples
On-demand recovery	VMS on-demand
Shadow hardware roles	Shadow Edge Devices

Requirements (VMS host)

Public expectations for a healthy VMS deployment:

Area	Expectation
Network	Reachable HTTPS proxy for operators; MQTT broker for fleet bus
Storage	Cache for recordings-preview / pull artifacts with retention policy
Secrets	Cloud API and storage credentials only on the VMS host
Zero trust	Prefer Access / tunnel for remote operators; no inbound shadow SSH from internet
Backups	Config and playbook catalog under version control

Next steps

• Shadow Edge Devices — device product video and roles
• Architecture overview — full tier diagram
• API & surfaces — proxy endpoints
• Quickstart — first shadow through the VMS path

Operator depth

Live readiness grades, playbook execution IDs, and soak evidence live in private operator handoff (not published).