Mistral Challenges OpenAI: Customize Your AI with Forge

Mistral Challenges OpenAI: Customize Your AI with Forge

The French startup Mistral boldly positions itself against AI giants like OpenAI. With its new tool, Forge, Mistral offers companies the opportunity to move beyond standard tools and create their own custom AI models. This approach aims to transform AI from a mere generic tool into a true strategic asset for businesses.

Mistral has chosen not to enter the consumer market, preferring instead to focus on the specific needs of enterprises. At the Nvidia GTC conference, the startup unveiled Forge, a tool that allows companies to develop an AI tailored to their own data. The goal is to make AI an essential business asset rather than just a generic tool.

The Limitations of Generic AIs for Businesses

Current AI models are often trained on publicly available data from the internet, making them effective for general tasks but poorly suited for specific business contexts. This inadequacy is particularly problematic for sectors like banking, industry, or administration, which rely on years of internal documents and specific procedures.

Although solutions like RAG (Retrieval-Augmented Generation) have been developed to integrate internal data, they do not allow for a deep understanding of that data. Mistral therefore proposes a radically different approach with Forge: creating AI models directly trained on companies' internal data, ensuring greater relevance and effectiveness.

Forge: A Custom AI Based on Your Data

Forge is designed as a comprehensive toolkit that allows businesses to design their own AI. By leveraging open models like Mistral's Small 4, companies can develop a system that precisely meets their needs.

This approach enables the management of complex cases, particularly in non-English languages or highly specialized contexts. Companies thus have complete control over their AI's behavior, being able to decide on priorities, elements to ignore, or areas to deepen. Forge also offers the possibility of creating multi-agent systems capable of collaborating and adapting to complex tasks, thereby reducing dependence on external vendors.

A Clear Strategy Against OpenAI and Anthropic

Unlike OpenAI and Anthropic, which dominate the consumer market, Mistral targets a more discreet but potentially more lucrative market: that of enterprises. The startup is on track to achieve one billion dollars in annual recurring revenue, proving that the demand for customized solutions is indeed present.

European companies are increasingly looking to reduce their dependence on American players, and Mistral positions itself as a credible alternative. With Forge, it offers a more manageable AI, aligned with local needs, moving away from the universal model to provide an almost artisanal solution.

A Hybrid Model Between Technology and Consulting

Mistral does not merely provide advanced technology. It also offers human support, with engineers working directly with clients to identify the right data, design assessments, and structure models.

This approach, which combines technology and consulting, is similar to that of companies like Palantir or IBM. It ensures the quality of assessments, a critical factor for the success of an AI project. Mistral's partners, such as Ericsson, the European Space Agency, or ASML, testify to the high-end orientation of the startup.

Conclusion

With Forge, Mistral transforms AI from a consumed model to a built model, offering companies relative autonomy. However, this customization requires time, skills, and resources, and not all companies will be ready to take the plunge.

Mistral could thus encourage the industry to evolve, pushing players like OpenAI and Anthropic to offer more personalized solutions. The one who provides the most control to businesses could very well gain the upper hand.