Most of the work in building a Data Vault isn't the code. It's the design. Reading metadata, profiling samples, identifying business keys, capturing definitions nobody updates later. We filmed a walkthrough of the IRiS Assistant doing exactly that work. One source table, start to finish: profiled, modelled, definitions captured, production-ready metadata pushed to the repository, in a single conversation. If you've ever sat through the design phase of a Data Vault build, this is worth eleven minutes. The IRiS Assistant is live in beta: https://ignition-data.com/iris-assistant

We said watch this space. Here it is. The IRiS Assistant is live. It connects to your source data, profiles it, and walks you through the integration model design one source at a time. Each table lands where it belongs, extending the model where new territory is needed and integrating cleanly where structures already exist. Which relationships need history and what terms need definitions. All captured as you go, all landed in your repository as a versioned, auditable record. The design phase used to be the expensive part. The part that needed weeks of senior architect time before a line of code got written. Not anymore. Learn more about the IRiS Assistant here: https://ignition-data.com/iris-assistant

There's been a story doing the rounds this week that an unnamed company blew half a billion dollars in AI tokens in a single month, Uber went through it's AI coding budget in the four months of 2026, Microsoft revoked access to Claude Code for its engineers and both Amazon and Meta have closed down employee driven AI token usage. Which begs the question, what happens now? Well that's what's being explored in our latest Podcast episode. You can have a listen here - https://youtu.be/xPpnEe2TAjk or where ever you get your podcasts from!

Six months on from the CES 2026 Atlas debut, the real story is what happens to your data stack when intelligence moves off the server and into the physical world. This week's edition of www.datapro.news is a check-in on how robotics is directly reshaping Enterprise Data and AI Management. We cover: 🎡 The Physical AI Flywheel — why circular, real-time pipelines are replacing batch ETL for good ✅ Demonstration quality scoring — mutual information estimation as the new data quality frontier 🤖 Synthetic data pipelines becoming core infrastructure, not a research investment Edge-cloud MLOps — what Jetson Thor actually changes for your deployment architecture. At $5.71/hr all-in versus $28/hr for a warehouse worker, the economic case has already been made. The data infrastructure question is no longer theoretical. Which of these shifts is closest to landing in your organisation right now — pipeline architecture, data quality, or MLOps at the edge? This week's edition is in your inbox. 👇

This week on our Data Radio Show Podcast, we take a look at the AI Inflection Point. Exploring the significant shift in the Artificial Intelligence landscape, moving away from basic model selection towards a focus on operational reliability and data infrastructure. You can find the Episode on all Podcast platforms now - just search for The Data Radio Show.