Dr Bart Jaworski

Dr Bart JaworskiOct 14, 2026

Everyone's an AI expert until you ask how LLMs work. Stanford has the cleanest answer I've found. Check out this FREE LLM course: Their full LLM course is 9 lectures, from Transformers to Agentic systems. If you're trying to separate AI hype from architecture decisions, start here: Lecture 7 (Agentic LLMs) and Lecture 8 (LLM Evaluation) are where most enterprise teams need depth. Full list: Lecture 1: Transformers https://lnkd.in/gTfXfZJC Lecture 2: Transformer-Based Models https://lnkd.in/gKKysY3x Lecture 3: LLMs https://lnkd.in/grEVzSpj Lecture 4: Training https://lnkd.in/gH9dcShh Lecture 5: Tuning https://lnkd.in/gcm-Adf4 Lecture 6: Reasoning https://lnkd.in/gd8ZC4Pj Lecture 7: Agentic LLMs https://lnkd.in/g4AzfJ-J Lecture 8: LLM Evaluation https://lnkd.in/ghrrNYai Lecture 9: Recap & Trends https://lnkd.in/gMp6_y77 Which lecture would change how your team builds? Please save this post and follow me :) Credit to Basia Kubicka for the original post on this :)

Dr Bart JaworskiOct 15, 2026

Were you ever accused of ignoring the product's quality? Here's how to reverse this perception: Before I start, let's face it: Often, the Product Manager will primarily serve the company and the business, not the users, as it is often portrayed. The PM can either accept it or fight for the principles and inevitably lose their job. Given that the former is far more likely, the business's focus on short-term results may put you at odds with your dev team, which pursues tech excellence, complete releases, and will often stand up for the users. As much as you'd like to side with them, your duty makes you push them outside those principles. However, in order to maintain their morale, the high quality of whatever is going out, and protect your mutual relationship, you need to face this no-win scenario. Here's how to do it: 1) Address the concerns before they arise - let them know you are between a rock and a hard place. It's not who you cares about quality, but the circumstances that push you to that stance. 2) Be transparent about your reasoning - perhaps there is a valid business/strategic reason to do what you have to do. Fill the team in; they are smart fellows, and they'll understand. 3) Have a plan for improvement regardless - maybe you have no time to make improvements and fix bugs, but at least record what is needed. That shows you at least care. 4) Show empathy At the very least, acknowledge that the situation is not ideal, and no one wants to work in it, and release garbage. You are, after all, not doing this to mess with your team; you are going against your better judgment yourself in this instance! 5) Set conditions that will allow you to re-examine the quality Perhaps if an important OKR is met, things can slow down for a second, and improvements/bugfixes can come into the forefront again. 6) Look for solutions on retro You are not alone, and at the end of each sprint, there is a meeting dedicated to finding solutions to problems together. You are not in this mess alone, and perhaps together you can think of a sensible way forward. 7) Sneak in at least some minor bug fixes and improvements - You can declare a sprint goal focused on what you are forced to do, but all of the team usually can't work on one thing at the same time anyway. Perhaps some quality work can be delivered under the radar? Were you ever disappointed with something you had to release? Let me know in the comments. Also, please follow me and save this post - it helps out a lot! #productmanagement #productmanager #agile

Dr Bart JaworskiOct 14, 2026

Do you know when to start as an AI PM? Who is this role in 2026, anyway? Here's all you should know: AI PM used to mean a person who's mostly an engineer who knows how to build machine learning AIs and how to maintain, improve, and evaluate. In 2026, this meaning is basically obsolete. Nowadays, an AI PM is a person who can successfully and profitably, introduce AI features using the models available. He also uses his creativity, productivity, and output with AI tools and techniques. More and more, this basically means that an AI PM spends a lot of time in Claude, being able to provide data and generate artifacts that allow the PM to excel over his non-AI peers. The gap between those two groups grows every month. The good news? Crossing it doesn’t require an engineering background or a single line of code. This Friday, I’ll show you exactly how to cross it. Simply join my free webinar: AI for Product Managers, Friday, October 16, 9:00 AM CEST In one hour, I’ll take one example end to end: We'll create a PRD with the context of the company and product. This will then lead to creating a clickable prototype there and then. Zero coding required. You know me; I’ve never had ‘developer’ on my resume. You’ll also see which PM responsibilities have disappeared, which have been automated, and which have just become more important. I want to set you on a path to excel as an AI PM after this webinar, not blindly follow trends to your detriment. Can’t make 9 AM? Register anyway; everyone who signs up gets the recording. But only live attendees can bring their own product or blocker to the Q&A. Register here: https://lnkd.in/dHUEKd7Z See you on Friday? Also, please follow me and save this post.It helps out a lot! #productmanager #productmanagement #aiagents