In partnership with
News of the day
1. Mistral CEO Arthur Mensch warns companies about proprietary AI models, highlighting data privacy risks and potential competition from AI labs. → Read more
2. Skello secures €200M to fuel European expansion, focusing on AI integration and new market penetration. → Read more
3. Google's new ad humorously depicts the Founding Fathers using AI and Workspace tools to draft the Declaration of Independence, sparking mixed reactions. → Read more
4. NVIDIA introduces HORIZON, a hands-free agent for hardware design that treats it as repository-level code evolution, achieving 100% RTL benchmark completion. → Read more
Our take
Hi Dotikers!
A few days ago, we watched Microsoft copy the Palantir playbook: engineers embedded on the client's premises to wire AI into the core of business processes. This week, Arthur Mensch takes the opposite tack. In a lengthy LinkedIn post, the Mistral boss warns companies that leaning on closed models means handing the big labs a front-row seat to your internal operations.
And he has a point. Proprietary model providers pile up your data, learn from how you use their tools, and gain real leverage over your business. Mensch pushes it a notch further, pointing to labs that supposedly "have a track record of going after their most successful customers." The worry is legitimate, even if, in this case, the proof has yet to surface.
Still, the message fits the in-house catalogue rather neatly. Mistral happens to sell Studio and Forge, everything you need to train your own models and keep your data at home. Fair enough, and honestly fairly consistent: Mensch practises what he preaches.
The wink is elsewhere. This plea for European independence comes from a company nearly 30 percent owned by American investors, one that hasn't yet closed its gap on GPT-5.6 or Fable 5, and that borrowed its famous secret weapon straight from Palantir, engineers deployed on site. You can champion the alternative and take a page from the best; the two aren't mutually exclusive.
A recent experiment actually backs him up: a Qwen model fine-tuned by the Bridgewater fund beat the frontier models on financial documents, 85 percent accuracy against 78 percent, at fourteen times lower cost. The case holds, even if OpenAI or Anthropic could buy that data tomorrow.
The point is solid, and coming from a European, you almost want to buy it. It just happens that the messenger runs a shop.
G.
Scale AI support on AWS, see how July 9
Customer expectations keep rising. Support budgets don't. On July 9, Fin and AWS are hosting a live executive session on how leading enterprises close that gap: scaling AI-powered support while simplifying how they buy it.
You'll see how to resolve an average 76% of conversations with Fin on AWS enterprise-grade infrastructure, procure through AWS Marketplace to put committed cloud spend to work, and turn the Fin and AWS collaboration into lower support costs. Register for the live session to see how.
Meme of the day
