Large language models have quickly moved from academic experiments into everyday technology. Not long ago, most people only heard about them in research circles. Now they show up in customer support systems, search tools, document analysis platforms, and even internal business software. Companies are starting to treat language models less like futuristic tech and more like practical infrastructure that can support real products.

France has quietly built a strong community around this field. Some teams focus on research and model development, while others concentrate on integrating language models into applications companies already use. The work often involves a mix of engineering, data processing, and product thinking. It is not just about training models, but about making them reliable enough to run inside real systems.

Below are several companies based in France that are working with large language models in different ways. Some build custom solutions for businesses, others develop AI platforms or research-driven tools. Together they give a good picture of how this technology is evolving across the French tech ecosystem.

 

1. OSKI Solutions

At OSKI Solutions, we build and support software systems for companies that rely on modern digital platforms. Our work includes web, mobile, and desktop applications, along with the cloud infrastructure that keeps those systems running. Many projects start with improving existing platforms or building new products that businesses can scale as their operations grow.

Part of our work involves integrating large language models and other natural language processing tools into business software. We connect language models with applications where they can process documents, assist with internal workflows, or support data analysis tasks. We also provide large language model development and integration services for companies in France, helping businesses add language based features to their software systems.

Key Highlights:

• Software development for web, mobile, and desktop platforms
• Integration of large language models and natural language processing tools
• Cloud solutions including computing, storage, and backup systems
• Experience with machine learning and computer vision projects
• Development of business applications and platform integrations

Services:

• Large language model integration into software platforms
• Cloud solutions including computing, storage, and backup
• Machine learning integration
• Natural language processing solutions
• Computer vision development
• Web development
• Mobile application development
• Desktop application development
• Database development and management

Contact Information:

• Website: oski.site
• Address: Kaupmehe tn 7-120, Tallinn, Estonia
• Phone: +48571282759
• E-mail: contact@oski.site 
• LinkedIn: www.linkedin.com/company/oski-solutions 

 

 

 

 

2. Yseop

Yseop works on software that turns structured data into written documents. The company focuses mainly on environments where documents must follow strict formats. Most of the work is connected to healthcare and pharmaceutical reporting. In these areas, teams spend a lot of time preparing documents for regulators, clinical studies, and internal review.

Yseop builds systems that combine language models with document automation tools. The software reads structured information and produces text that follows predefined writing rules. Medical teams and regulatory specialists use these tools to prepare reports, patient narratives, and other technical documents. 

Yseop platforms are designed for environments where accuracy and traceability matter. Language models are used together with rule based logic to ensure documents stay consistent with regulatory guidelines. The goal is not just text generation but controlled document production that fits regulated workflows.

Key Highlights:

• Software focused on automated document creation
• Use of language models for structured report writing
• Platforms designed for regulatory and medical documentation
• Systems that convert data into written narratives
• Tools used in pharmaceutical and life sciences workflows

Services:

• Automated document generation
• Natural language generation for structured data
• Regulatory document writing tools
• AI driven reporting systems
• Language model integration for document workflows

Contact Information:

• Website: yseop.com
• Address: 4-6 Place de la Bourse, 75002 Paris, France
• LinkedIn: www.linkedin.com/company/yseop
• Twitter: x.com/yseopai

 

3. Lettria

Lettria develops software that helps organizations understand large collections of documents. Many companies store thousands of reports, research papers, and internal records that are difficult to analyze manually. Lettria tools are built to extract structure from that information and turn it into something easier to explore.

Lettria focuses on language models that work with knowledge graphs. The platform reads documents, identifies key concepts, and organizes them into structured relationships. This allows analysts to explore information through graphs instead of long text files. It also helps systems answer questions using evidence from real documents.

The technology is often used by teams working with complex technical information. Healthcare researchers, legal teams, and financial analysts rely on large volumes of documentation. Lettria systems help those teams find connections between documents, extract insights, and build structured knowledge bases from raw text.

Key Highlights:

• Language models combined with knowledge graph technology
• Systems that convert text into structured information
• Tools designed to analyze large document collections
• Platforms used in research heavy industries
• Focus on traceable document analysis

Services:

• Document parsing and text analysis
• Knowledge graph creation from documents
• Ontology development for structured knowledge
• Language model tools for document understanding
• AI solutions for research and knowledge management

Contact Information:

• Website: www.lettria.com
• Address: 46, Rue du Val, 77160 Provins
• E-mail: hello@lettria.com
• LinkedIn: www.linkedin.com/company/lettria

 

4. Kyutai

Kyutai is an artificial intelligence research lab based in Paris. The lab studies different areas of AI, including language models, speech technologies, and multimodal systems. The work is focused on research rather than commercial products, and many of the projects are shared publicly through open science initiatives.

Kyutai explores new ways to design language models that are smaller, more flexible, and easier to adapt. One part of this research looks at modular language models. Instead of training a single large system for everything, the lab experiments with models that can be specialized for certain tasks or languages.

Kyutai also works on voice technologies connected to language models. Some projects focus on speech to speech translation, while others explore real time conversation systems that understand both spoken words and tone. The research often leads to open tools and publications that developers can use when building their own AI systems.

Key Highlights:

• Research lab focused on language and speech technologies
• Work on compact and modular language models
• Development of speech driven dialogue systems
• Projects related to real time translation and voice interaction
• Open research with public tools and publications

Services:

• AI research in language models
• Speech to speech translation systems
• Text to speech and speech recognition technologies
• Multimodal AI experiments combining voice and images
• Open tools for language model development

Contact Information:

• Website: kyutai.org
• E-mail: contact@kyutai.org
• LinkedIn: www.linkedin.com/company/kyutai-labs
• Twitter: x.com/kyutai_labs

 

5. Dust

Dust develops software that helps teams work with AI agents connected to internal company knowledge. The platform acts as a workspace where organizations can create assistants that understand company data and help with everyday tasks.

Dust focuses on systems that combine language models with workplace tools. The platform connects to services such as Slack, Notion, Google Drive, and GitHub. This allows the agents to read internal information, summarize discussions, or answer questions based on company documentation.

Different departments use these agents in different ways. Marketing teams may generate messaging or analyze feedback. Engineers can search technical documentation or review code context. Customer support teams use the system to retrieve knowledge base information when answering tickets. The platform organizes these interactions through specialized agents designed for different tasks.

Key Highlights:

• Platform for creating AI agents connected to company data
• Language model tools integrated with workplace systems
• Support for internal knowledge search and automation
• Agents designed for different business teams
• Infrastructure for managing multiple AI assistants

Services:

• AI agent development platform
• Language model integration with company tools
• Knowledge base search systems
• Workflow automation using AI agents
• Data analysis and reporting tools

Contact Information:

• Website: dust.tt
• E-mail: privacy@dust.tt
• LinkedIn: www.linkedin.com/company/dust-tt
• Twitter: x.com/DustHQ

 

6. LightOn

LightOn develops software that allows companies to run language model systems inside their own infrastructure. Many organizations work with sensitive information and cannot send data to external AI services. LightOn platforms are designed for those environments.

LightOn focuses on systems that combine language models with document retrieval tools. The platform reads large collections of internal documents and allows employees to search them using natural language questions. The software can process different types of content including reports, diagrams, tables, and scanned documents. Companies use the platform to manage internal knowledge, analyze regulatory information, or monitor compliance related materials. A key aspect of the system is that data remains inside the organization’s infrastructure.

Key Highlights:

• Enterprise platforms for private language model deployment
• Systems designed for document search and reasoning
• Tools for analyzing large internal knowledge bases
• Infrastructure that keeps data inside company environments
• Support for compliance and information governance

Services:

• Document search and retrieval systems
• Knowledge management tools powered by language models
• Secure AI deployment inside company infrastructure
• Consulting and implementation support
• Enterprise language model platforms

Contact Information:

• Website: www.lighton.ai
• E-mail: rssi_rgpd@lighton.ai
• LinkedIn: www.linkedin.com/company/lighton
• Twitter: x.com/lightonio

 

7. Mistral AI

Mistral AI works on large language models and the infrastructure needed to build AI applications around them. The company develops its own models and also provides tools that allow organizations to customize and deploy them in different environments.

Mistral AI focuses on building flexible language model systems that developers can adapt to their own data and workflows. The technology can be used for tasks such as document analysis, search, coding assistance, and automated workflows. Companies often integrate these models into their own software rather than using them as standalone tools.

The platform also supports building AI agents and applications on top of these models. Developers can train models with specific datasets, adjust them for specialized tasks, and deploy them in cloud systems or private infrastructure. This allows organizations to build AI systems that fit their own technical environment.

Key Highlights:

• Development of large language models
• Platforms for building AI applications and agents
• Tools for customizing and deploying language models
• Support for private and cloud based AI systems
• Research and development in language model technology

Services:

• AI application development platforms
• Custom model training and fine tuning
• Large language model development
• AI agents and workflow automation tools
• Deployment of language models in enterprise systems

Contact Information:

• Website: mistral.ai
• E-mail: press@mistral.ai
• LinkedIn: www.linkedin.com/company/mistralai
• Twitter: x.com/mistralai

 

8. Dataiku

Dataiku develops a platform used by companies that work heavily with data. Many organizations have large amounts of information stored in different systems. Dataiku helps teams bring that data together so analysts, engineers, and business specialists can work on the same projects.

Dataiku provides tools that allow teams to experiment with language models alongside traditional machine learning. Inside the platform, users can prepare datasets, train models, and build AI applications that connect directly with business data. Language models are often used for tasks like document analysis, conversational interfaces, and automated reporting.

Dataiku software is designed so different roles inside a company can collaborate. Data engineers prepare the infrastructure. Data scientists train models. Business teams review results and build applications on top of them. The platform acts as a shared workspace for these activities.

Key Highlights:

• Platform used for data science and machine learning projects
• Tools that support language model experiments and applications
• Shared workspace for engineers, analysts, and business teams
• Systems that connect data pipelines with AI models
• Infrastructure for deploying AI solutions in production

Services:

• Data science and machine learning platforms
• Language model integration in analytics workflows
• AI application development environments
• Data preparation and model training tools
• Enterprise AI deployment and monitoring

Contact Information:

• Website: www.dataiku.com
• Address: 201-203 rue de Bercy 75012, Paris
• E-mail: security@dataiku.com
• LinkedIn: www.linkedin.com/company/dataiku
• Twitter: x.com/dataiku
• Instagram: www.instagram.com/dataiku

 

9. Giskard

Giskard focuses on testing and monitoring AI systems before they are used in real applications. When companies build products with language models, unexpected behavior can appear. Models might produce incorrect answers or reveal sensitive information. Giskard software is designed to detect those problems early.

Giskard builds tools that evaluate how language models behave in different scenarios. The platform simulates interactions with users and checks the results. It looks for issues such as hallucinated responses, prompt manipulation, or inconsistent answers.

Engineering teams often add these tests to their development pipelines. That way the system runs checks every time a model is updated. Instead of waiting until problems appear in production, teams can identify risks while the system is still being developed.

Key Highlights:

• Platforms designed to detect risks in AI systems
• Automated testing scenarios for model evaluation
• Systems used by engineering and security teams
• Integration with development pipelines
• Tools for testing language models and AI agents

Services:

• AI agent vulnerability detection
• Automated model testing workflows
• Monitoring tools for AI system behavior
• LLM testing and evaluation tools
• Security testing for AI applications

Contact Information:

• Website: www.giskard.ai
• LinkedIn: www.linkedin.com/company/giskard-ai

 

10. SESAMm

SESAMm develops software that studies large amounts of online text and turns it into insights about companies and markets. Financial institutions often monitor news articles, blogs, and social media when evaluating investments. SESAMm systems analyze this information automatically.

SESAMm builds language processing tools that read large volumes of public content and detect signals related to sustainability, risk, or corporate activity. The platform identifies companies mentioned in articles and analyzes the context around those mentions.

Investment teams use the platform to track controversies, public sentiment, and ESG related topics connected to companies. The system gathers information from many sources and organizes it so analysts can review potential risks or emerging trends more easily.

Key Highlights:

• Language processing tools for financial and ESG analysis
• Systems that monitor news, blogs, and social media sources
• Platforms that analyze large collections of online text
• AI tools used by investment firms and analysts
• Multilingual text analysis technology

Services:

• ESG monitoring and controversy detection tools
• Text analysis for investment intelligence
• Data delivery through dashboards and APIs
• AI systems for monitoring global news sources
• Natural language processing for financial research

Contact Information:

• Website: www.sesamm.com
• Address: 7 rue de Madrid, 75008 Paris, France
• LinkedIn: www.linkedin.com/company/sesamm-sas
• Twitter: x.com/sesamm_nlp

 

11. Outmind

Outmind develops software that helps employees find information inside company documents. In many organizations, files are spread across servers, email systems, and collaboration tools. Searching through those sources manually can take a lot of time. Outmind was built to make that process easier.

Outmind combines language models with enterprise search technology. Users can ask questions about projects or documents in simple language. The system scans connected sources and returns relevant information from internal files.

Outmind works with many document formats such as PDFs, emails, presentations, and spreadsheets. The platform connects to common business tools like file servers or collaboration platforms. 

Key Highlights:

• Enterprise search assistant for internal documents
• Language model based question answering for company data
• Systems that connect multiple document sources
• Tools designed for knowledge management
• Support for many document formats and archives

Services:

• Enterprise document search systems
• Language model powered knowledge assistants
• Document summarization and analysis tools
• Integration with internal file systems and collaboration tools
• AI platforms for knowledge management

Contact Information:

• Website: en.outmind.ai
• Phone: +33 6 51 62 60 73
• E-mail: hello@outmind.fr
• LinkedIn: www.linkedin.com/company/outmind-app
• Twitter: x.com/outmind_ai

 

12. Shift Technology

Shift Technology develops software used by insurance companies to review claims and analyze policy data. Insurance organizations handle large volumes of documents and records every day. Many of these records contain text descriptions that need to be reviewed and categorized.

Shift Technology applies language processing systems to insurance data. The platform reads claim reports, policy descriptions, and investigation documents. Machine learning models help identify patterns that may indicate risk or irregular activity.

Insurance teams use these tools to support claim reviews and investigations. The system processes large collections of documents and highlights information that may require closer examination. This helps analysts focus on complex cases instead of reviewing every file manually.

Key Highlights:

• Software designed for insurance data analysis
• Language processing tools for claim and policy documents
• Systems used in fraud detection and risk analysis
• Platforms used by insurance companies worldwide
• AI systems integrated into insurance workflows

Services:

• Insurance fraud detection systems
• Claim analysis platforms
• AI tools for underwriting review
• Document processing for insurance records
• Risk monitoring and investigation support tools

Contact Information:

• Website: www.shift-technology.com
• E-mail: marketing@shift-technology.com
• LinkedIn: www.linkedin.com/company/shift-technology
• Twitter: x.com/shiftechnology
• Facebook: www.facebook.com/ShifTechnology

 

13. Hugging Face

Hugging Face works as a platform where developers, researchers, and companies share machine learning models and datasets. The company built a large online hub where people can upload models, test them, and use them inside their own applications. Many projects related to language models are published and maintained through this platform.

In the context of LLM companies in France, Hugging Face plays a different role compared with traditional software vendors. Instead of building a single product, Hugging Face created an ecosystem where language models and other machine learning systems are developed collaboratively. Developers can download models, fine tune them, or deploy them in their own systems.

The platform includes tools for working with language models, datasets, and applications. Engineers use libraries and frameworks maintained by Hugging Face to train and run models in research environments or production systems. The hub also hosts community projects where developers share experiments and applications built on top of machine learning models.

Key Highlights:

• Platform used by developers working with machine learning models
• Hub where models, datasets, and applications are shared
• Tools used in many language model research projects
• Libraries for training and deploying language models
• Community driven ecosystem around machine learning tools

Services:

• Hosting platform for machine learning models
• Tools for training and deploying language models
• Dataset hosting and management tools
• Open source machine learning libraries
• Infrastructure for running AI applications

Contact Information:

• Website: huggingface.co
• E-mail: privacy@huggingface.co
• LinkedIn: www.linkedin.com/company/huggingface
• Twitter: x.com/huggingface

Conclusion

The ecosystem of LLM companies in France is still evolving, but it already covers a wide range of fields. Some teams focus on language models themselves, while others apply these systems inside specific industries such as healthcare, insurance, or scientific research. In practice, this means language technology in France is not limited to chat interfaces or consumer tools. It often appears inside complex platforms that analyze documents, study biological data, or support large research projects.

Another noticeable pattern is how closely many of these companies work with real data from their industries. Medical research groups collaborate with hospitals. Insurance platforms process claim documents. Scientific teams study biological datasets and medical images. Language models become part of a larger system rather than a standalone product.

France also benefits from a strong research culture. Many organizations building LLM related tools have roots in universities, research labs, or scientific communities. That background shows in the way these systems are developed - often with a focus on technical depth and long term research rather than quick consumer applications.

Looking ahead, this environment will likely continue to grow. As more industries experiment with language models, the role of these companies may expand as well. For now, the French landscape already shows something interesting - language technology being used not just for conversation, but for solving very specific problems in science, business, and research.