Monitoring · intervention · interfaces

Every device monitored. Every intervention logged.

CENNTIR connects the computers, screens, rooms and buildings of a site to a single console. The AI assistant diagnoses and proposes; it only acts within the permissions you grant it, and every action is written to the log. Your users, for their part, get interfaces designed for their job.

Hub logExample
  1. Hotel · room 203TV screen: “Sun” theme publishedapproved
  2. Conference room · room PCVideo-conferencing software restartedapproved
  3. Workstation ACC-12 · Windows 11Print service restartedapproved
  4. Main site · networkInventory: 21 devices foundlogged
  5. Cabin 2 · tabletInterface updatedapproved
  6. Building B · CO₂ sensorThreshold exceeded, alert sentto handle
Illustrative example. In CENNTIR, every real action leaves a line: who, what, when, on which machine.
How it works

One intervention, from request to log.

Animated example: an AI assistant restarts a service on a workstation in the fleet, within the permissions granted to its key.

  1. 01Request
  2. 02Permissions checked
  3. 03Run by the agent
  4. 04Result verified
  5. 05Written to the log
You or your AI assistantconsole · command line · assistant

The print service on workstation ACC-12 has stopped. Restart it.

Done: the print service is running again on ACC-12.

CENNTIR hubpermissions · approvals · log
  • Key “support-assistant” recognised
  • “Commands” permission enabled on ACC-12
  • No approval required for this action
Agent · workstation ACC-12Windows 11 · Lyon office
> Restart-Service Spooler✓ Spooler: running
Hub log09:13:30support-assistantcommandACC-12Restart-Service Spoolersucceeded
What CENNTIR does

Observe, intervene, equip.

A lightweight agent on each machine, a hub that enforces permissions and keeps the log, and interfaces for the people who use the place.

Observe

Know what is happening on site without being there.

  • Real-time fleet status
  • Network inventory of a site
  • Alerts on deviations

Intervene

Fix things remotely, within the limits you set.

  • Actions bound by permissions, machine by machine
  • AI assistant that diagnoses and proposes
  • A log entry for every action

Equip

Give users an interface made for them.

  • Tablets and Android TV screens
  • Interfaces in the venue’s colours
  • Remote updates
Evidence

Available capabilities and levels of evidence.

Every capability states its level of evidence. We do not promise universal compatibility: we qualify your hardware during a pilot.

HardwareLevel of evidenceWhat works
Windows 10 and 11 computersAvailableAgent 0.8.5: commands, files, screen and Office.
Windows 7 SP1AvailableDedicated binary, same core capabilities as Windows 10 and 11.
Mac (Apple Silicon)AvailableAgent 0.8.5: commands, files and screen.
Linux servers and workstations, x86 and ARMAvailableAgent 0.8.5: commands and files, no on-screen actions.
Android phones, tablets and TV sticksAvailableApp 0.8.5 running on Xiaomi Mi TV sticks (Android 14), a Lenovo tablet and a Redmi phone, October 2026.
KVM-over-IP boxesAvailableKeyboard, screen and mouse of a computer remotely, including at boot, and a network audit of the site. Running on GL.iNet Comet; JetKVM partially validated; PiKVM to be qualified. Input injection disabled by default.
ESP32-S3 boardsAvailableSensors, pins, e-paper and LCD screens; USB installation, signed updates over Wi-Fi. In service since October 2026.
AV equipment and BMSProven on emulatorSome device profiles are proven on emulator, others are still under study; each on-site integration is validated through a pilot.
iPhone and iPadTrialApp in testing, distributed outside the App Store.

Available: released and documented version. Tested on specific hardware: tried on precise models, to be requalified for another model. Proven on emulator: works against a faithful simulator, not yet on real equipment.

Method

From the first meeting to day-to-day operations.

We start small, on your hardware, with success criteria written down before rolling out.

  1. One-day audit

    Site survey: equipment, network, usage, constraints. You receive a written assessment.

  2. Pilot

    One workstation, one room or one screen, with acceptance criteria agreed together.

  3. Roll-out

    Agent installation, permissions, interfaces and training for your team.

  4. Operations

    Monitoring, updates and logged interventions, with your team or your integrator.

Security and traceability

Nothing happens without permission, and everything can be traced.

Agents connect out to the hub: no port to open on site. The hub enforces permissions, handles the approvals set by the policy and keeps the log.

Read the security page →

Denied by default

Every action and every tool is enabled machine by machine.

Approval in the console

Depending on the policy, some equipment write operations wait for approval, with a one-time code.

Exportable log

Who, what, when, on which machine; CSV export, itself logged.

Masked secrets

The connector engine resolves equipment passwords on the agent and masks them in its results and log.

Demos

Try before we talk.

All of them open in the browser. They are simulations unless stated otherwise: no command is sent to any equipment.

Tablet and TV screenSimulation

Maison Auréa

Welcome, restaurant, spa, room service and events for a fictional hotel, on a tablet and on a TV screen, in English and French.

Open the demo ↗ (new tab)
Room remoteSimulation

Reconfigurable room

A conference room that switches between four layouts: live seating plan, requests to speak, microphones, camera presets and volume, with a demo loop.

Open the demo ↗ (new tab)
Guided animation (in French)Simulation

The AI assistant steps in

From the office, the assistant works on a client site’s network through the agent: it diagnoses, acts, then checks.

Open the demo ↗ (new tab)

All demos →

Frequently asked questions

Short answers.

What is CENNTIR?

A platform and services to monitor, remotely troubleshoot and equip with custom interfaces the computers, screens, rooms and buildings of a site. An AI assistant diagnoses and proposes within the permissions you grant it, and every action is written to the log.

Do we need to open ports on the site firewall?

No. Each agent opens an outbound encrypted connection to the hub on port 443. Corporate HTTP proxies (CONNECT method) are not supported: direct outbound access to the hub must be allowed.

Which devices are supported?

Windows 10 and 11, macOS on Apple Silicon, Linux x86 and ARM, and Android (phones, tablets, TV) are available. Windows 7, KVM-over-IP boxes (GL.iNet Comet) and ESP32-S3 boards are available too; iPhone and iPad are in trial.

Can the AI assistant act on its own?

It only acts within the permissions granted to its key, machine by machine and tool by tool. Depending on the policy, some equipment write operations require approval in the console. Every call is written to the audit log.

Is traffic end-to-end encrypted?

No. Links are encrypted with TLS, but the hub is the control point: it sees the traffic in order to enforce permissions, approvals and the log.

How do we get started?

With a one-day audit or a pilot on one workstation, one room or one screen, with acceptance criteria agreed together.

One site, one use case, one pilot.

Describe your site: we propose a first scope to test on your hardware.

Talk to us about your site →