Software development
- The latest open-source models make the alternative credible
- Integration modules for your teams’ IDEs
- Your source code never leaves your perimeter
Enterprise AI
that stays
A sovereign platform to industrialise the automation of your business processes, without giving up your data and while keeping your bills under control.
Ask your sovereign AI…
Your data stays inside your perimeter · Inference on your GPUs
- ISO 27001
- HDS
- Diffusion Restreinte
- GDPR
- Zero per-token billing
- Inference traceability
Why ?
Neither hand everything to a public AI nor build it all yourself: AntemetAI gives you the speed of one and the control of the other.
Your data stays inside your perimeter
Public AI
Sent to the provider
In-house build
On your premises
Fast to deploy
Public AI
Immediate
In-house build
Months of project work
Predictable cost
Public AI
Per-token billing
In-house build
Team and maintenance to fund
Control over models and updates
Public AI
Set by the vendor
In-house build
Full
GDPR, HDS, Diffusion Restreinte compliance
Public AI
Hard to guarantee
In-house build
Your responsibility
Operations and support
Public AI
By the vendor
In-house build
Your responsibility
Already using Microsoft 365 Copilot?AntemetAI complements it as soon as your data is sensitive or per-token billing spirals.Learn more
Budget predictability
- Subscribe to the solution, acquire the appliance: no tokens, no pay-per-execution billing
- Infrastructure sized to your actual needs
- A clear cost for your CIO, CISO and CFO
Data control
- Hosted on your premises or in AntemetA datacentres: ISO 27001, HDS, Diffusion Restreinte
- In-house RAG, guardrails from ingestion onwards
- Full inference traceability
One platform,
Agents, automation, office suite and data management: the SmarT portal brings all your use cases together in a single environment, administered by your teams and open to your business applications.
- Agent StudiO
- RAG
- MCP
- Guardrails
- OpenAI-compatible API
- Optimised prompts
- SSO Entra ID / LDAP
- White label
- Enterprise applications
- Partner marketplace
ChaT
AI agents and conversations
Questions, files, documents, dashboards and use cases in one place, grounded in your knowledge.
TexT
Text document generation
Tender responses, NIS2 or DORA compliance documents, on-brand formatting.
PreZ
Presentation generation
On-brand PowerPoint presentations with this month’s KPIs, in seconds.
SurveY
Questionnaires and Excel forms
Automatic completion of Excel forms running to tens or hundreds of rows.
RecorD
Meeting recording and summaries
Your meetings transcribed and summarised: decisions, key points and follow-up actions.
TranslaT
Text and document translation
Translation of text and .docx files using your company glossary.
AssisT · key differentiatorAutomate your business processes,
AssisT orchestrates your AI agents, connectors and business rules in visual workflows. Describe your goal in natural language: AI builds the workflow, you approve it, and it runs on your infrastructure.
catalogue nodes
native connectors
triggers
Daily support digestDescribe your goal and let AI build the workflow
Ready-to-use connectors and logic nodes
- SharePoint
- Teams
- Outlook
- ServiceNow
- Zendesk
- Splunk
- SQL
- HTTP / API
- Webhook
- Amazon S3
- Cloudian S3
- CIFS / SMB
- NFS
- FTP / SFTP
- Word
- Excel
- RSS
- Plakar
- Sub-workflow
- Condition
- Switch
- Loop
- Wait
- JS code
- Variables
- Data table
- XML → JSON
- Form
- Human approval
CoWorK · business use casesAI for every team,
CoWorK gives each team its own library of use cases: responding to a tender, screening applications, extracting invoices… Your people fill in a simple form, and AI agents and workflows do the rest.
CoWorKMy use cases
HRFinanceLegalSupport
- Sales
- Prepare a client meeting
- HR
- Finance
- Legal
- Support
SalesSales proposals
Respond to a tender
A technical proposal written from the tender documents.
Tender documents
Contract type
PublicPrivate
Deadline
ready-to-use use cases
business functions covered
A ready-to-use library, plus your own use cases
Connect a form to an agent or an AssisT workflow: it immediately shows up for the teams concerned.
- Sales
- Marketing
- HR
- Finance
- Procurement
- Legal
- IT
- Customer support
- Management
- R&D
- Healthcare
- Transport & logistics
ClassifY · key differentiatorPrepare your data for AI
AI is only as good as the data it reads. ClassifY brings together, classifies and cleans your documents from every source to make them AI-ready, securely and in full compliance.
1. Connect
Your data sources
SharePoint
Google Drive
OneDrive
S3 Protect
SFTP server
CIFS / SMB
NFS share
Knowledge base
Plakar
Files
Read-only: your original files are never modified.
2. Standardise
Prepare & clean
Categorisation
Entities
Cleaning
Anonymisation
Backup
Workflow
Renaming, metadata, deduplication, GDPR anonymisation.
3. AI-ready
Usable data
Normalised, enriched and secure documents, ready to feed your models.
A clean, structured and compliant corpus for your AI use cases.
Confidentiality × audience
Cross-checks how sensitive files are against how many people can open them.
| 1–5 | 6–20 | 21–100 | 101–500 | > 500 | Public link | |
|---|---|---|---|---|---|---|
| Secret | 12 | 4 | 2 | 0 | 0 | 0 |
| Confidential | 148 | 96 | 31 | 9 | 3 | 1 |
| Internal | 820 | 1,340 | 2,210 | 980 | 4,120 | 6 |
| Public | 40 | 22 | 18 | 65 | 310 | 54 |
CompliantTo watchInconsistent
Multi-label categorisation
Your own business classification scheme, a confidence score per label and detection of uncertain labels.
Personal data
People, emails, phone numbers, addresses: the share of files affected and a GDPR check.
Pseudonymisation
EMAIL1, EMAIL2… or anonymisation by type: the processed document can be exported as a PDF.
Data cleansing
Duplicates and files untouched for over a year are flagged: the potential savings are quantified.
Naming conventions
[DATE] - [LANG] - [TITLE].[EXT]: a new name suggested for every file.
Key metadata
Author, sensitivity label, encrypted, protected or signed files.
Available standalone or built into the platform. A dedicated ClassifY offering for data preparation, with its own glossary and sources, that can trigger custom AssisT workflows.
Audit my dataWhen public AI services are essential,
Every outgoing request is analysed against your data policy: it goes through, it is pseudonymised, or it never leaves your infrastructure.
Users and applications
Requests in OpenAI API format
AllowedPassed on unchanged
PseudonymisedSensitive data replaced before sending
BlockedThe request never leaves the infrastructure
Public AI services
- • OpenAI
- • Gemini
- • Meta
Pseudonymisation example
User request
“Summarise the dispute between Dupont SA and Mr Martin (martin@dupont.fr)”
Request sent to the public AI
Awaiting analysis…
Three use cases to get started
Visible gains from the very first weeks, on tasks where data is sensitive and per-token bills spiral.
Document production
- Responses to tenders and RFPs
- Technical documentation
- Analysis of contracts and quotes
1
2
3
4
5
AI inside Office
- Modules built into your office applications
- Automation of repetitive actions
- Features missing or poorly covered by generic add-ons
AI solutions for your most specific needs
Where general-purpose AI falls short, AntemetA’s software partners bring specialised solutions, designed for a specific industry and need.
Omundu· Unity Forge
AI built for industrial maintenance
Unity Forge is an AI platform designed to tackle the major challenges of industrial maintenance: reducing downtime, improving access to information and optimising interventions.
Visit website Omundu (opens in a new tab)Wintics
Video analytics for everyone who shapes our cities and regions
Wintics turns existing cameras into management tools that power bold initiatives for more comfortable, safer and more sustainable cities and infrastructure. Wintics gives everyone who shapes a region effective solutions to make it an attractive and pleasant place to live.
Visit website Wintics (opens in a new tab)SafenAI
The AI solution to manage, control and monitor your operational data models
SafenAI provides solutions that secure the use of artificial intelligence, with a focus on compliance, risk management and data protection in AI systems.
Visit website SafenAI (opens in a new tab)Marvin Systems
Your documents pseudonymised in one click
Marvin Systems is a French SaaS platform that pseudonymises and anonymises documents so that regulated professions (lawyers, doctors, auditors…) can use AI without breaching professional secrecy or the GDPR. It automatically detects and masks more than 30 types of sensitive data.
Visit website Marvin Systems (opens in a new tab)Magic LEMP
AI that enlightens without dazzling
Magic LEMP supports you from identifying AI opportunities to putting custom AI solutions into production: under control, with no magic tricks.
Visit website Magic LEMP (opens in a new tab)Three operating models
On your premises or in AntemetA data centres, dedicated or shared: you choose where your models run and who operates them.
AntemetAI Private Dedicated
Infrastructure installed on the client’s premises
AntemetA
- Single or multi-GPU infrastructure, even multi-appliance
- Managed by the client’s managed service provider
AntemetAI Private Dedicated
Dedicated infrastructure at AntemetA
Client
- Single or multi-GPU infrastructure, even multi-appliance
- AntemetA managed services
AntemetAI Private
Shared infrastructure at AntemetA
Clients
- Shared Multi-GPU infrastructure
- Shared models
- Platform shared between several clients
Included in all 3 models
Full operations
Monitoring, backup, AntemetAI support
Commitments and oversight
SDM follow-up with quarterly reports
24/7 support and maintenance
Two major updates a year
Built with the
AntemetAI appliances run on hardware and GPU acceleration from leading partners. AntemetA integrates them, operates them and keeps them in operational condition.
NVIDIANVIDIA
GB10, RTX 4500 and RTX 6000 Blackwell, H200, GB200 / GB300 GPUs, and NIM inference microservices.
Dell Technologies
Server platforms for inference appliances, on-premises or hosted, and Dell Pro Max GB10: a compact AI workstation built on NVIDIA GB10.
Hewlett Packard Enterprise
HPE servers for inference appliances, on-premises or hosted.
Lenovo
Server platforms for inference appliances, on-premises or hosted, and ThinkStation PGX: a compact AI workstation built on NVIDIA GB10.
Backed by the in digital performance and resilience
For a sovereign, available and secure information system and enhanced productivity, with a full catalogue of professional and managed services, from audit to administration.
Since
revenue
clients
employees
servers · +220 racks
data centres · 6 regions · 8 branches
of protected data
trees planted in agroforestry
Infrastructure & sovereign cloud
Supply, integration and maintenance of infrastructure built on the latest technological advances, to host your production tools.
Cyber-resilience
Proactive, consistent 360° protection: cybersecurity, data protection and business continuity.
Everything about
Let’s build your
An initial conversation with our AI architects to scope your use cases, size the appliance and choose your deployment model.
- Platform demo on your own use cases
- Audit of your data with ClassifY
- Costed proposal: a subscription, not a meter