I think these playbooks will definitely be a game-changer in building multilingual and culturally aware models that promote equitable and inclusive AI deployments. Thanks Jacki O'Neill and our team at Microsoft Research Africa Lab, Nairobi.

Learn how to create and publish professional children's storybooks on Amazon KDP using AI tools. From story generation to illustrations, discover the complete workflow for indie authors. https://lnkd.in/eWsvFaGT

Three seconds. That's what a book gets to land one thing that sticks in your mind. On a shelf, on a retailer page, in an AI recommendation. Most book positioning tries to say everything, so nothing lands. Those getting this right have found the single truth about a book that holds across every channel - Amazon, BookTok, a press release, an AI assistant. And the message resonates with, entertains, challenges customers. But the number of channels just keep growing, each one with its own discovery mechanism and messaging patterns - not to mention AI recommendations. On Tuesday, I'm running a workshop on how to get to a strategy that will last across channels and messages, with PublisHer at Bologna Children’s Book Fair: Designing for Discovery - How Data-Led Positioning Shapes the Way Readers Find Books . Tuesday 14 April, 12pm CET - PublisHer Lounge. See you there 🖖

AI is already part of library workflows. In our most-attended webinar, we looked at how libraries are using it today, from drafting content to building reports and answering routine questions. Here’s a recap of what came up. Read more → https://hubs.la/Q048Tn0N0

Your AI does not need intent to cause harm. It only needs to act. We are entering a world where outcomes matter more than intentions, and where responsibility cannot rest on what a system was meant to do. My latest Substack post explores what accountability looks like when intent is no longer required. This is a core thread in my next book, Machinehood. Not whether systems think, but how we govern what they do. The shift is already happening. 📖 Read: https://lnkd.in/g8cEi2YC

I've stopped using AI tools one at a time. Claude for thinking. Perplexity for anything that needs the live web. NotebookLM for loading in full ebooks and pulling exact details, so Claude isn't hallucinating half a chapter when I'm working through an idea. The tool-vs-tool arguments are mostly noise at this point. What's actually changed my workflow is getting them to talk to each other.

Writing metadata for every video in your library? That time is better spent elsewhere. Zype AI now generates titles, descriptions, keywords, genres — all from your video’s transcript, guided by your brand voice. Enable it in minutes under AI Tools in your Zype dashboard. Learn more in our latest announcement → https://hubs.ly/Q04cz95B0

The designers who'll struggle most in 2026? The ones that are still learning in isolation. This is what learning looks like in my cohort. Real applications you can build from Week 1. We fed JUST ONE stakeholder meeting into Claude and walked out with a full clone of a product manager, live, in front of the group. It wasn't a vague persona. We got a detailed guide that breaks down: → Their exact phrases. "I don't think that's possible" means "Soft no — don't push further on this path." → What to avoid. The specific triggers that shut down a conversation before you've even made your case. → The one-line summary. How this person actually thinks, compressed into a sentence you can reference before every review. All from a transcript that was already captured weeks ago. This is what my cohort looks like. It's not fluffy slides about AI. We're building real tools you use in your next meeting. The May cohort is open now, check link in the comments.

Made it back home to Toronto! After hearing a ton about the developments in AI this past week, I felt in over my head as a salesperson who didn’t study computer science or learn how to code growing up. Sam Bhagwat’s book Principles of Building AI Agents was extremely helpful to understand how agents work and how to think about building them. My biggest takeaway: Start thinking in workflows for each component of my day to day. Try to break the workflows down into a step by step sequence that can build on previous context. If anyone is selling AI and doesn’t come from a technical background, would highly recommend this book!

AI/ML folks don't typically read these two books. They are - technically - not AI/ML books. But imagine - just imagine - that you have depth of knowledge in both the fields here. Strong math background, and a strong systems background. You will reach a certain level of knowledge that very very few people have. You can talk of service level optimization (load balancing, p99 latency) and design loss functions (total variation for smoothness etc). If you like math and systems thinking, join our upcoming webinars: topmate.io/ajay_shenoy