TLDRocket
Sign in

Inside Mirakl's agentic commerce vision

OpenAI

Mirakl is rebuilding its business tools around AI agents using ChatGPT Enterprise. It's a bet that shopping and B2B commerce will soon be run by bots talking to bots, not people clicking around.

Based on reporting by OpenAI — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

Mirakl, the French marketplace-software company that helps big retailers and distributors run third-party seller platforms, is now leaning hard into agentic AI. According to OpenAI's own writeup, Mirakl has rolled out ChatGPT Enterprise across its workforce and is using it to speed up internal documentation and sharpen customer support, the kind of unglamorous operational work that usually eats months of employee time.

But the more interesting piece is where this is headed. Mirakl is building something called Nexus, described as infrastructure for what the company calls agent-native commerce, essentially a marketplace architecture designed from the ground up for AI agents to browse, compare, and transact on behalf of humans rather than humans doing it themselves. That is a meaningfully different bet than just bolting a chatbot onto an existing storefront.

The timing tracks with a broader shift OpenAI has been pushing all year, moving ChatGPT from a research and writing tool into something that acts inside business software, files documents, checks inventory, maybe even completes purchases. Mirakl, sitting on top of relationships with retailers and B2B distributors, is a useful proving ground: its whole existence is about connecting buyers and sellers at scale, so if agents are going to do commerce, this is exactly the kind of plumbing they will need.

What is missing from OpenAI's telling, predictably, is much detail on error rates, cost, or how Mirakl handles the messier parts of enterprise adoption, things like data governance across multiple retail partners or what happens when an agent gets a product spec wrong in a live marketplace. Those are the parts that decide whether agent-native commerce is a real category or just a compelling case study.

My take — AI-written commentary, not fact-checked reporting

I like the ambition here more than I trust the hype cycle around it. Case studies published by the AI vendor itself always read like highlight reels, and agentic commerce touching live money and live inventory is exactly the place where quiet failures get expensive fast. Worth watching, not worth believing wholesale yet.

Read more about this at: OpenAI

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.