Factory

Run every Agent as one fleet

Factory runs a Kerberos Agent for every camera on your own Kubernetes cluster. Add a camera with its RTSP stream or discover it through ONVIF, and Factory creates the deployment, service, and configuration while Kubernetes keeps it running on the right node. Storage, Hub, MQTT, and WebRTC settings are defined once for the whole fleet, and upgrades, reboots, logs, and resource usage are handled from one place.

  • One Agent per camera
  • Runs on your Kubernetes
  • Fleet-wide upgrades

Cameras

Scale your Agents

StatusCreatedNameImageReplicas
Online2026/08/27agent-shop-floorkerberos/agent:v3.5.21 / 1
Deployment:
agent-shop-floor
Pod:
agent-shop-floor-7f9c6d5b8d-x2kqp
Node:
edge-ghent-01
Image:
kerberos/agent:v3.5.2
Memory limit:
512Mi
IP address:
10.20.0.11
agent-shop-floor-7f9c6d5b8d-x2kqpStreaming
  1. INFO Running Kerberos Agent v3.5.2
  2. INFO RunAgent: opening the RTSP stream
  3. INFO OpenRTSP: H264 1920x1080 at 15 fps
  4. INFO HandleHeartBeat: heartbeat sent to Hub
  5. INFO ProcessMotion: motion detected in region 1
Online2026/08/27agent-entrancekerberos/agent:v3.5.21 / 1
Online2026/09/10agent-loading-baykerberos/agent:v3.5.21 / 1
Online2026/08/05agent-parking-lotkerberos/agent:v3.5.11 / 1
Online2026/09/25agent-warehouse-aislekerberos/agent:v3.5.21 / 1
Online2026/10/02agent-receptionkerberos/agent:v3.5.21 / 1
Interactive product preview

The Factory interface as operators use it: an Agent for every camera, live node and pod usage, ConfigMaps, and global settings. Expand a camera to stream its logs, or open Add Agent and Upgrade Agents to try a rollout.

Core capabilities

Your Agent fleet, backed by an open platform.

Each capability below maps to a current product surface or runtime flow. Use the source and documentation to go deeper or talk to us about your deployment.

01

Onboard any camera

Add a camera with its RTSP stream or find it through ONVIF discovery, and Factory turns it into its own Agent deployment.

02

Kubernetes keeps it running

Every Agent gets its own deployment, service, and ConfigMap. Kubernetes places it on an available node and restarts it when needed.

03

Global settings, local overrides

Define timezone, MQTT, STUN/TURN, encryption, Hub, and Vault persistence once. Every Agent inherits them and can still override what it needs.

04

Fleet-wide upgrades

Pick an Agent version and roll it out to one camera or the whole fleet. Agents already on that version are left untouched.

05

Live node and pod usage

Follow CPU and memory per node and per Agent, live from the Kubernetes metrics API, and see where every camera is scheduled.

06

Diagnose without SSH

Stream logs, open a terminal, reboot an Agent, or adjust its memory limit straight from the camera list.

How it fits

From configure to action.

Factory is useful on its own and becomes more capable when connected to the rest of the UUG.AI stack.

Read the documentation
  1. 01

    Configure

    Install Factory in your cluster and set the storage, Hub, MQTT, and WebRTC settings every Agent inherits.

  2. 02

    Onboard

    Add cameras by RTSP or ONVIF. Each one becomes an Agent that Kubernetes schedules on an available node.

  3. 03

    Operate

    Watch nodes and Agents live, roll out new versions, and reboot or inspect any Agent from one place.

Build with UUG.AI

Shape the right video stack for your operation.

Start a conversation