ARSY.AI
TechnologyPricingFind IntegratorCompareCase Studies
ARSY AI SpaceRequest Demo

Security & Deployment

Where your data actually sits

This page is written for the IT and quality functions that have to sign off on a deployment. It states the deployment topologies, what leaves the customer network and what does not, the access controls, the server requirements, and — at the end — the certifications ARSY AI does not hold. Everything here is a property of the product as it ships today, not a roadmap.

Deployment topologies

Managed cloud

Lite · Starter · Professional · Enterprise

ARSY AI runs on our infrastructure. Fastest to stand up, and the only option on the Lite and Starter tiers.

On-premise

Professional · Enterprise

The full stack runs on a server inside the customer network. It operates with no internet connection at all. This option is not available on Lite or Starter.

What stays inside your network on-premise

In the on-premise topology there is no cloud component in the loop. The server sits on the customer LAN and the following never reaches us:

  • Site scans and the spatial maps derived from them.
  • Authored procedures, training scenarios and uploaded documentation.
  • Camera streams from phones and AR glasses.
  • Operator activity, step completion and captured evidence.
  • The language model and vision models, which run locally on the same server.

Network behaviour and latency

ARSY AI uses a persistent WebSocket connection rather than request-response REST calls, which is what makes continuous positioning and streamed AI responses possible. Measured round-trip figures by topology:

On-premise, over LAN
1–5 ms
Managed cloud, WebSocket
5–20 ms
Request-response REST, for comparison
50–200 ms

Access control and audit

Role-based access control is available on the Professional and Enterprise tiers. It is not included on Lite or Starter.

  • Permissions configured per user and per device.
  • Access restricted to specific scenarios, data sets and system functions.
  • Operates entirely within local infrastructure in an on-premise deployment.
  • User and system actions are logged, which is what customers use for their own compliance reporting.

Server requirements for on-premise

Minimum specification for the customer-hosted server. Sizing above these figures is scoped by the integration partner against the number of sites and covered machines.

CPU
8 vCPU / 8 cores minimum
GPU
VRAM 12 GB minimum
RAM
16 GB minimum
Storage
SSD 200 GB minimum
Network
100 Mbit/s, recommended 1 Gbit/s

Service levels

Uptime commitments are tier-bound and we state them exactly: Enterprise carries a 99.5% uptime SLA, Professional carries 99%, and the Lite and Starter tiers carry no uptime SLA at all. Support routing, response times and any custom SLA are set in the subscription agreement and delivered jointly with the certified integration partner.

Ready to Transform Your Operations?

See how ARSY AI can bring embodied AI to your physical environment. Schedule a personalized demo today.

Request DemoContact Sales
ARSY.AI

Embodied AI agents for the physical world. Bringing perception, reasoning, and execution together in a single closed-loop system.

Platform

  • Technology
  • Visual Positioning (VPS)
  • Modules
  • Deployment
  • Security
  • Pricing

Company

  • ARSY AI Space
  • Case Studies
  • Find Integrator
  • Compare
  • Partners
  • Contact

© 2026 ARSY AI. All rights reserved.

Web-AR.Studio Corp

Privacy PolicyTerms of Servicesupport@arsy.ai