Mistral Adds Managed Agents and Apps to Its Studio Enterprise AI Platform

The update puts Mistral up against the agent-building platforms that OpenAI, Anthropic, Google and the big cloud providers are selling to large companies. Mistral is competing on where the software runs. The company said Studio can be deployed in the cloud, in a customer's own virtual private cloud, on-premises, or across a mix of these, so sensitive data and the processes built on it can stay inside a company's own systems. That matters most to banks, governments and other regulated buyers that are reluctant to send data to US-hosted services.

Mistral said customers and partners including ABANCA, Accenture, Alten, BNP Paribas, Capgemini, CMA CGM and Tata Consultancy Services already use Studio in production.

Studio is built around Mistral Large 4, which the company nicknames "Le Chonk". Mistral released the open-weight model as a preview in October. The company says it is strong at coding, defensive cybersecurity and reasoning over complex documents. Developers connect the model to company data and systems through connectors and servers that use the Model Context Protocol, the open standard for linking AI models to tools.

The announcement splits the platform into three parts. Mistral said in its announcement that an agent "investigates and acts", a workflow coordinates execution and approvals, and an app gives employees a way to use these systems.

The new managed agents can carry out multi-step tasks. Developers can give them skills that encode a team's procedures, sandboxes in which to run code, and subagents to hand off parts of a job to. Hooks let developers insert their own logic at set points while an agent runs. Mistral said Studio runs the agents on the same infrastructure that supports millions of sessions a week of Mistral Vibe, its coding agent. Developers can limit which tools, data and actions each agent can reach.

Workflows set out which steps follow fixed rules, where a model reasons and where a person makes the decision, and an interrupted run can pick up where it stopped. Teams can describe a process in Mistral Vibe to generate, test and deploy a workflow as code they can edit. The release adds event triggers from enterprise systems and a visual graph of each workflow's logic.

For apps, Studio supplies hosting, authentication and access controls, as well as pre-approved components meant to cut the amount of new code that needs a security review. Using a new Mistral server for the Model Context Protocol, developers can build with Mistral Vibe or another coding agent, test locally and deploy with a single command.

Mistral gave examples from its own teams. It said an agent that starts investigating production incidents, following the runbook and asking for approval before sensitive actions, saves its engineers hundreds of hours a month. A workflow that sorts internal ideas and customer feedback saves its product operations team about 35 hours a week.

The company also said Studio produces OpenTelemetry traces of every step in a run, including model calls, tool use and approvals, and lets teams test changes to prompts, models or tools against their own datasets before release. The traces and evaluation results stay with the customer.

The announcement did not include pricing or say whether the new features are available in every region.