AI in your company.
Not in the cloud.
LMbox is a physical appliance installed on your premises. Your employees ask questions, analyze documents, write content—exactly like with ChatGPT. Except everything stays with you, on your network, under your control.
- · Your queries sent to servers in the United States
- · Your documents analyzed remotely
- · Per-seat subscription, unlimited cost
- · Dependent on internet
- › Your queries stay in your appliance
- › Your documents stay on your premises
- › Unlimited users, fixed cost
- › Works even offline
In plain terms: how does it work?
We deliver an appliance to you
A physical device arrives at your location—compact as an internet router for a small firm, server-format for a larger organization. It is delivered preconfigured. You have nothing to install, nothing to set up.
We connect it to your network
An LMbox engineer comes to your site for 3 days. They connect the appliance to your internal network and integrate it with your tools: email, shared folders, document repositories. You do not touch anything.
Your teams use it from day one
From their usual browser, your staff access a familiar interface. They ask questions, write, analyze. The AI knows your internal documents. Everything stays on your network.
No application to install. No lengthy training. Your teams are operational on the day of deployment.
What happens when your staff member asks a question
- No data leaves your network.
- No external server is contacted.
- LMbox itself does not see your exchanges.
With ChatGPT or Claude, every question sent by your staff member transits through U.S. servers. Your contracts, your client files, your medical data leave your premises.
With LMbox, the appliance is on your premises. The AI runs on your premises. Your data stays on your premises—even if the internet goes down.
Air-gap mode available
The appliance operates 100% disconnected from the internet. Designed for classified or highly confidential environments.
Built for professionals who cannot afford a data breach
Lawyers & notaries
Professional secrecy applies to your digital tools. Analyzing a contract with ChatGPT means sending that contract to a U.S. company. With LMbox, your client files never leave your firm. Your staff have an AI that knows your document repository—with no risk of leakage.
Healthcare & clinics
Health data is among the most protected under French law. LMbox is HDS-compliant—the regulatory framework for digital tools in medical settings. Your practitioners draft reports, analyze protocols, search the literature. No patient data leaves your facility.
Defense & sensitive sectors
Some environments cannot connect to the internet to use AI. LMbox operates in full air-gap mode: the appliance runs in a closed circuit, with no outbound connections. Your teams get professional-grade AI in a fully isolated environment.
Mid-market & SMBs
Your employees already use AI tools—often without your knowledge, often with your data. LMbox gives them an official, secure alternative that you control. Fixed cost, unlimited users, no surprises on the bill.
What arrives at your site—and what happens next
-
The sealed box
Compact, pre-loaded with models, ready to plug in. Serial number + SHA256 signature provided.
-
Cables and PSU
10 GbE network cable, power supply, Thunderbolt cable for initial config.
-
Installation guide in French
Print and digital versions. Step-by-step procedures for the most common configurations. Written to be understood by your teams, not just IT staff.
-
Onboarding session
3 days on-site with a LMbox engineer. User training included for your champions.
-
Ongoing support and updates
New AI models, fixes, new features—deployed automatically to your appliance. Technical support in French included in the annual subscription.
Choose the format that fits your organization
Same software. Same security level. Same support. Only the performance differs.
A compact appliance—fits on a desk or in a network cabinet. No server room required.
- For whom
- Firms of 5 to 15 people
- Concurrent users
- up to 10
Integrates into your existing server rack. Standard rack format.
- For whom
- Organizations of 30 to 80 people with a server room
- Concurrent users
- up to 25
High-performance server for multi-department deployments.
- For whom
- Mid-sized companies of 100 to 300 people, multiple departments or sites, large document base
- Concurrent users
- up to 100
Hardware purchased outright. 36-month warranty. Annual subscription. Unlimited users—no per-seat billing.
What runs inside—without the jargon
LMbox is built on proven, verifiable technologies. No proprietary components. No single vendor lock-in. Everything can be audited by your teams or an independent provider.
Everything runs on your premises
The AI models, the computations, the storage—everything executes on the appliance in your facilities. No cloud subcontractor, no third-party provider processes your data.
Your credentials, your rules
Your employees log in with their usual account (Microsoft, Google, or your internal directory). You define who accesses what. An intern does not see partner files.
You see everything that happens
A dashboard gives you a complete view: who used the AI, when, on which documents. Exportable for your DPO, your internal audit, or the CNIL.
Models improve automatically
When new, more capable models become available, LMbox deploys them to your appliance via a secure update. You always benefit from the best performance—without doing anything.
What our clients observe after 3 months
For the first time, I can let my employees use an AI assistant on client files without worrying about professional secrecy.
Air-gap mode allowed us to deploy LMbox in a clinic without negotiating a security exception with our DPO. That had been a blocker for 18 months.
The questions we're asked most often
Is it as good as ChatGPT? +
For everyday tasks—drafting, analyzing a contract, summarizing a document, searching your archives—yes, the performance is comparable. And because LMbox knows your internal documents, its answers are often more accurate than a generic service. For highly complex tasks requiring very broad general knowledge, large cloud models still have a slight edge—a gap that narrows with each update.
What happens if the internet goes down? +
Nothing. The appliance runs completely autonomously. Your team continues to use AI normally—because everything runs on your premises, with no dependency on an external connection.
How long before our teams are up and running? +
Between 10 and 15 days from order to first real use. Delivery takes 5 to 7 days. Installation and training take 3 days. The rest is spent connecting your existing tools depending on their complexity.
Want to see how it would work in your organization?
30-minute demo on your real use cases. An engineer shows you the interface, answers your questions, and quotes your configuration live.
30-day on-site POC available—deductible from the price if you sign.
You test in your real environment, with your actual documents, before committing.
Let's talk
We reply within one business day.
