Artificial intelligence under your roof

Digital sovereignty, absolute confidentiality, hybrid power.

ProductivIA servers for education, business, and institutions

Our servers assembled in Québec bring the entire ProductivIA suite directly inside your walls, under your rules, your policies, and your firewalls.

Designed for businesses, institutions, and sensitive environments, these servers combine local autonomy with global power: an embedded model for internal requests, and a secure extension to external models when very high reasoning power is required.
Every piece of data verifiably remains stored on your infrastructure, every external call is filtered, masked, or refused according to your policies, and every action is logged locally.

This is more than a technical solution—it’s a choice for digital independence, a vote of confidence in your teams, and a decisive step toward data sovereignty.

Why a local server?

control_camera

To stay in control

Our ProductivIA servers are designed, assembled, and configured in Québec.
They provide the power of modern AI while keeping your data on your own infrastructure, under your full control.

key

No compromise on confidentiality

Locally hosted AI models let you enjoy increased productivity while eliminating the risk of sensitive data leakage.
Your requests, histories, and documents remain within your walls, in an encrypted environment administered by your teams.

flag

Adapted to Québec realities

Our servers meet the requirements of Québec businesses, educational institutions, and public bodies:

Compliance with Law 25
Adherence to public procurement rules (LCOP)
Digital accessibility (RGAA 4.1)
Fully localized support
Deep knowledge of Québec’s cultural and economic context

mode_dual

Two operating modes

Disconnected mode: full autonomy, 100% local responses for maximum security and confidentiality.

Hybrid mode: if authorized, extension to world-class models with filtering and masking of sensitive data, with optional manual validation.

data_exploration

Matania: our sovereign orchestration layer

Each server includes Matania, our orchestration layer developed in Québec, which handles your requests on the front line. Matania is enriched by a large Québec knowledge base fed by the latest public data, official publications, and various open sources relevant to the local context.
When a task requires more power, the server can switch to more robust internal models or, if authorized, to world-class models — in full transparency.
We offer regular updates to expand Matania's capabilities and delegate fewer tasks to external models. This measurable progress ensures growing autonomy over time.

Two power options

editor_choice

Standard Server

The standard server has the power needed to analyze requests, respond when the local language model allows, and call on a world-class service for more complex tasks—once the request has been purged of any potentially confidential elements.

crown

Reinforced Server

The reinforced server can support a larger number of users and process more requests while limiting calls to external services thanks to the integration of a second, more powerful model.
It is the ideal solution for organizations that want to serve many users while keeping the majority—or even all—processing in-house, under their full control.

Intelligence that understands your context

database

Local, cultural knowledge base

Matania ships with a Québec knowledge base, including the latest public government data as well as all publicly available official publications.
The orchestration layer is pre-calibrated to respect Québec norms, practices, and cultural particularities.

arrow_and_edge

Retrieval-augmented data

Servers can integrate:

Your internal documents (PDF, Word, HTML, Confluence, SharePoint, Google Drive)
Public government, economic, and cultural data from Québec
Your domain databases or internal intranets

line_end_diamond

Connection to your directories and tools

Compatible with infrastructure standards:

Active Directory / LDAP
SharePoint
Google Drive / Workspace
Confluence / institutional intranet
Internal servers or shared files

cognition_2

Integrated ProductivIA interface

The server hosts:

The full ProductivIA web application
Configuration and administration modules
No-code modules to design your own internal tools (assistants, generators, smart forms, etc.)

Intelligent processing workflow

arrow_split

Automated decision chain

The request is received by the local server.
Local feasibility analysis.
If possible: instant processing by the local model
Otherwise: transfer to the reinforced model

privacy_tip

Supervised external requests

If the request exceeds local capacity:

Security analysis
Masking / anonymization of confidential data
Human authorization request if necessary
Full history of the request and response

Governance and security

admin_panel_settings

Full control

Access management by role, profile, or department
Fine-grained selection of authorized engines (local / external)
Configurable security rules: absolute, strict, rigorous, flexible
Complete interaction history and detailed logs

enhanced_encryption

Encryption and compliance

Full encryption at rest and in transit (TLS 1.3)
HSM option (dedicated hardware key)
Local management of encryption keys
No data transmitted without explicit authorization
Compliance with Law 25 and government cybersecurity standards

Supervision and administration

view_apps

Complete overview

From the admin interface:

Request history
Engine usage
Security logs and alerts
Tool tracking by profile or department

security

Operational security

Automatic alert in case of anomaly
Export and audit of activities
Restricted and logged admin access

Specifications

Our professional servers are assembled, tested, and configured in Québec with the same care we apply to our software solutions.
Designed for reliability, performance, and sovereignty, they integrate the most advanced technologies to give your organization all the power it needs—while keeping full control of its data and critical operations.

cognition

Standard Server

Ryzen 9 CPU with 64GB RAM and 2TB RAID
AMD EPYC 24 cores
Complete ProductivIA application layer
Large language model with 32K context window
Available for purchase or rental

cognition_2

Reinforced Server

Ryzen 9 CPU with 128GB RAM and 4TB RAID
2x AMD EPYC card (48 cores x2)
Complete ProductivIA application layer
Large language model with 32K context window
Extended language model with 64K context window
Available for purchase or rental

Assistance and support

partner_exchange

Option 1: Full support

Secure remote access by our technical team, 100% based in Québec.
We perform updates, resolve issues, and help connect your tools.

siren_question

Option 2: Guidance without access

We guide you by phone or videoconference for your updates and resolutions, without accessing your confidential environments.

Frequently asked questions

psychology_alt

What's the difference between the standard server and the reinforced server?

The standard server includes Matania and has sufficient power to handle everyday requests and secure those that require redirection.
The reinforced server also includes Matania, plus a much more powerful internal model, allowing it to process more complex requests locally — thus reducing reliance on external services.

database

Do data truly remain on our servers?

Yes. All data—requests, histories, imported documents—remains stored locally on your server, under your exclusive control.
No information is transmitted externally without your explicit authorization, even when an external engine is used.

security

How does ProductivIA ensure information security and confidentiality?

Each server applies multiple layers of protection:

Full data encryption at rest and in transit (TLS 1.3)
Local key management or hardware HSM option
Automatic masking of sensitive data before any external send
Access logs and alerts in case of anomaly
Configurable security settings: absolute, strict, rigorous, flexible.

document_scanner

What types of documents or data can we integrate?

The server can ingest and query your internal files (PDF, Word, HTML, CSV, etc.) as well as your environments:
SharePoint, Confluence, Google Drive, internal directories, or intranets.
This data fuels Retrieval-Augmented Generation (RAG), allowing the model to answer based on your own content.

wifi_lock

Can the servers operate without an Internet connection?

Yes. The ProductivIA server can operate on a closed network (intranet) for highly sensitive environments.
Updates or extensions can then be performed manually according to your security policies.

network_intel_node

What role does the “cultural model” play?

Each server includes a cultural layer integrated into the system prompt:
a set of linguistic, cultural, and contextual adjustments that strengthen understanding of Québec French, local realities, and Québec’s administrative or sectoral vocabulary.

passkey

Can we customize access rules?

Yes. You can:

assign different engines by department or profile,
define authorization levels for external access,
track individual histories and activity logs.

All of this is administered directly from the ProductivIA interface.

expand

Can the server scale with our needs?

Absolutely. The servers are designed to be scalable: you can increase memory, add a GPU, or upgrade the local model to a more powerful version.
Software and model updates are performed without data loss or service interruption.

Discover ProducitivIA for business

Discover ProducitivIA for education

Discover ProducitivIA for institutions