DataCamp

It's the end of my 2nd week at Bonx, and the early stage energy is giving me Dataiku circa 2016 vibes all over again (IYKYK). Here's what I'm uncomfortably excited about: 1. The product vision is bold. At Bonx, we're building the AI-native manufacturing ERP of the future. Not a system of record that waits for humans to execute every decision, but a system of action. Not an agentic layer on top of legacy platforms, but a whole new way of doing things. 2. The problem is massive and deeply underserved. The average 100-person manufacturing company loses 15,000+ hours a year fighting with their ERP. That's thousands of hours spent NOT running their business. LLMs have the ability to address the fundamental rigidity that has made ERPs so painful for decades. 3. Innovative European manufacturers, including La Maillecotech , Amantys Joaillerie , SomethingAdded , Féroce , L'Atelier du Ferment , NERVURES , already love Bonx (bold product vision, big problem AND 50+ happy customers = a marketer's dream 😍 ). 4. The team is hungry and moving fast (my kind of people!). Thanks to Alexandre Barroux and Rémi Bèges for the incredible opportunity to be a part of it; couldn't be more excited to build alongside this group Fabien Kintzig Alexandre Frot Francesca Morichelli Guillaume Fontaine Ava Gamon Angelica Pacini Chrisas AGBOSSOU Davide De Caterina Francesca Maria A. Giorgia Verduci Mathieu AZZOPARDI Paul Morelon Thibault Rannou Lyza Midouche John Dutamby Edouard Bilan Abdellah Khouya There is soooo much to do — week 3 (and beyond), let's go 🚀

This image upscaler is something else: 1. Trained for & on AI images (standard upscalers miss the unique textures of AI and magnify artifacts) 2. True 105 MP that outperforms any 4K upscaler 3. Covers everything, from social campaigns to massive printed billboards 4. From AI generation to print in a single workflow via Canva I’m glad I got to write the tutorial for this awesome release by the Leonardo.Ai team. Find more examples and the how-to here: https://lnkd.in/dvgj4mji

The data stack just changed again! The latest challenge is figuring out make it work for both humans and agents. In this episode of DataFramed, I chatted with Shireesh Thota , CVP of Databases at Microsoft , about his take on the new stack, and the implications for data teams. There's some great advice on how to stitch together all your tools into a coherent, ahem, fabric, and I was pleasantly caught off guard by a Microsoft exec espousing the importance of open source! Full episode link in the comments.

There are only 3 types of companies in Silicon Valley right now. My main takeaway from the Human[x] conference, after dozens of talks and browsing hundreds of booths, is that everyone is either 1. Creating AI agents. 2. Creating tools to help you create agents, or 3. Creating tools to help you manage agents. That's it. I guess 2026 is officially the year of the agent.