Mistral didn’t roll their own agent runtime. They wrapped Temporal — the same durable execution engine Netflix and Stripe use — and pointed it at LLM workflows. Public preview hit late April 2026, and ASML, ABANCA, France Travail, and CMA-CGM are pushing millions of executions through it daily.
What it actually is
An orchestration product for long-running AI processes, not a chat framework. If a workflow crashes mid-step, it resumes from the last checkpoint. Every step gets logged, traced, and retried. Mistral hosts the control plane. Workers run inside your own Kubernetes via Helm, so data never leaves your perimeter.
The API and SDK
Python SDK v3.0 is live. Decorate functions to define steps. Retries, timeouts, rate limits, tracing — single-line configs. Human-in-the-loop is one decorator: the workflow pauses without burning compute, reviewer signs off in Le Chat or via webhook, execution resumes. Publish to Le Chat and any non-engineer can trigger it.
Why it matters
LangGraph keeps reinventing what Temporal solved a decade ago. Mistral’s bet: enterprises don’t want a research-grade agent framework, they want the boring battle-tested stuff with LLM hooks bolted on. Given the customer list, that bet looks right.
You Might Also Like
- Stripe Coinbase and Ramp Built Internal Coding Agents Langchain Open swe Gives you the Same Architecture for Free
- Stripe Minions 1300 Pull Requests a Week Zero Human Written Code
- Temporal Technologies Just Raised 300m and Yeah it Makes Sense
- Adobe Photoshop ai Assistant Just Dropped in Public Beta Edit Photos by Talking to Them
- 600 Million for 16 Engineers why Netflix Made its Second Largest Acquisition on ben Afflecks ai Startup
Top AI Product 
Leave a comment