MONTREAL — Mistral AI is hiring in Montreal and courting Canadian customers, Arthur Mensch, the firm’s CEO and co-founder, has told The Logic.

“The talent concentration here is high,” Mensch said in an interview ahead of the All In AI conference in Montreal. The firm has made its first hire in Montreal, and plans to recruit engineers as well as sales and marketing staff in the city. 

Mensch also said he met Tuesday with potential clients in the financial services sector, and that he has more conversations lined up on Wednesday; he was speaking upstairs from the Canada Fintech Forum in Montreal, a conference that featured executives from several major Canadian banks and insurance firms.

Mensch also cited Canadian energy, manufacturing, logistics and mining as sectors his firm was targeting. Mistral has clients in those industries in other parts of the world, and “we would love to address [them] in Canada,” he said. Existing customers include the likes of insurance giant Axa, telecom firm Orange and energy firm TotalEnergies.

Mistral is particularly keen on Quebec, where most businesses must operate principally in French, one of several languages in which the firm has trained its models to be conversant.

The firm is one of a handful around the world that trains its own foundation models, which power generative AI tools. It also builds technology to connect those systems to data and make them work in different applications. “Off-the-shelf applications do not really cut it,” Mensch told The Logic. Simply selling a business a chatbot “is disappointing to them.” 

The firm open-sources its models and lets clients run its technology via cloud services, or on their own hardware. It also dispatches staff to help customize and develop applications for clients.

That approach lets Mistral adapt its technology to the needs and customs of particular markets, according to Mensch. “Our ability to customize the models allows us to grasp the cultural nuances,” he said. For example, Mistral has developed tailored versions of its technology for Singapore and for Belgium, where some firms must serve customers in Flemish.

“That’s what we would love to do here in Quebec,” Mensch said, noting the differences between written and spoken French in Quebec and France. “Those cultural nuances we’d love to grasp,” he said, by working with governments and businesses “to pour their own knowledge and orientation into the system.”

Like Mistral, Toronto-based Cohere is also targeting the enterprise market, offers on-premise deployments, works with businesses to customize its technology to their needs and touts its French proficiency. 

Mensch said there’s room for both firms in the market, and Canadian customers may choose to deploy models from each. “We go much more for the technical use cases,” he claimed, citing Mistral’s focus on audio and image applications and the reasoning capabilities of its model.

Mistral already has AI scientists trained at Mila, the Montreal AI research institute, working at its other offices. Mensch himself spent a few months as a research intern at McGill University in early 2014 while doing his master’s degree.

Montreal is home to AI labs from several major tech firms, including Meta, Microsoft, Google’s DeepMind and, more recently, Cohere. Mensch—who’s scheduled to speak at the All In AI conference on Wednesday—is used to having to poach from other companies to build up Mistral’s staff in new cities. “We’ve seen the talent war,” he said. “We are in the middle of it.”