Moderna

Honestly, the biggest hurdle to using AI for sustainability reporting isn't the tech itself—it's the intimidation factor of getting started. Today at the Semiconductor Climate Consortium 's Sustainability Forum, I got to share a behind-the-scenes look at how my team at Google actually uses AI. The goal was simple: make AI feel less like a black box and more like an actionable tool for sustainability professionals. One of the best parts was seeing confidence shift in real-time. We did a poll on AI comfort levels, and the number of people who felt "conversational" with the tech jumped from 48% to 61% by the end of the session. There is still a lot to (un)learn and (re)implement, but shifting that mindset is where it starts. Thank you to SEMI for inviting me, and to everyone who joined the conversation! Bruce Gall Julie Rapoport Kevin Martins Peilun Sun Kenneth Guertin, PMP

Frontiers London is a wrap! Even though I came from Boston, this was the 🦞 capital with 🦄 Peter Steinberger educating our customers (and me) on the future of agents! 🤩 My 3 takeaways: 1️⃣ Every customer I met was raving about Codex. Currently the best coding agent on most benchmarks, and 4x more "token efficient" (that means cheaper) than main competitors. CFOs appreciate! It's OpenAI's new blockbuster. 2️⃣ Transformation and culture change were on everyone's mind. It's increasingly accepted that AI can only deliver its full value when every employee has access. Like computers, or the Internet. this requires organizing the human layer of change, and I'll publish an article about that soon. 3️⃣ The next frontier is how organizations will capture and transform hidden processes and skills into auditable AI systems. I'm not talking about data but implicit knowledge and mental models that currently live in people's minds, not on their hard drives. Huge kudos to Denise Holland Dresser , Peter Doolan , Carrie Lorenzo for organizing this event and all the awesome speakers: Matt Weaver , Anouk Muis , Evan Weiss . What a spectacular lineup! And a special mention for Christina Meng 's team and Champions Network, with which we closed the event yesterday night in a lively conversion. Many of them are now beyond the adoption stage, and working on proficiency and process redesign. The momentum in Europe is nothing short of extraordinary. I can't wait to be back! 🤗🚀

🔥 Most of our expertise lives inside our heads. AI has the potential to elevate us when we externalize it. That simple realization goes against the dominant story, which is about loss: flattened voice, eroded skills, outsourced judgment. The story I want to share is different. If we use AI to structure the editorial process instead of replacing it, the human craft doesn’t disappear. It can even be explored, improved and shared while making our work more engaging and the outcome more ambitious. Over the last few weeks, I tested that idea by building an app called LinkedIn Publisher to help me write long-form articles with more structure, stronger grounding, sharper reviews, and better visual coherence. My goal wasn’t to automate writing but, on the contrary, to be more demanding with my own thinking before it reaches you. What I learned along the way is that writing in itself was never the hardest part of publication. It's everything around it: clarifying what we are trying to say in the first place, deciding how much grounding each idea deserves, crafting an adequate tone and voice, pressure-testing our assumptions, creating visuals that shape the rhythm of reading, and reviewing whether the piece is both intellectually honest and valuable for the reader. To cover that spectrum, LinkedIn Publisher is organized around six stages: Frame, Build, Draft, Review, Illustrate, and Export. Each stage externalizes a part of the editorial process that used to live inside my head. It combines dictation, skills, web search, agents, image generation, and computer use to handle every step from initial inspiration to published article. The videos in the article provide a concrete example, where I show step-by-step how I use this editorial platform to think through an article about the rise of AI skills. What this kind of work illustrates is broader than writing. Lawyers, doctors, and executives also carry mental models that guide their judgment every day. We all do. Yet we rarely make the time to document them and turn them into code. But AI changes the return on investment of doing that work, while providing a cognitive upside that almost no one is talking about. 👇 The article below walks through the full system, the reasoning behind each stage, and the deeper reason I believe AI should challenge our thinking rather than replace it. ✨ If you’re also on that path, I can’t wait to read how you’re thinking about AI to replicate your own mental models and accelerate your growth journey! #AI #Writing #FutureOfWork #Leadership

Last night at PyData Boston-Cambridge, which was sponsored by Cursor Boston for pizza, I demo'd how to use Marimo Pair for agentic data science work. It was super fun! It was also run as a hackthon, and was really cool to see what folks could do with Marimo notebooks when pairing it with Cursor's agent mode. I did bork the live demo recording, so I re-recorded it. Link in comments below :) h/t 👩🏻💻 Jacqueline Valeri , Sebastian Wallkötter, Ph.D. , Benjamin Batorsky , Roger Hunt, et. al. for organizing! #agents #datascience #ai

You can say that things between me and Claude Code are getting serious 😏. Can't believe I've used 18x more tokens than War and Peace (13.2 M?!). Though, compared to others, I think these are rookie numbers. I'm going to have to pump these numbers up! #ClaudeCode #BlackSwanCausalLabs #RWE #AI #LLM

Tableau Virtuoso just got a significant upgrade. It's still the highest-rated Tableau custom GPT agent. More than 5,000 convos. 4.3 out of 5 stars. And it just got better. At #TC26 , I served as a Tableau Doctor, supporting attendees with their questions. Many were new to Tableau. While I typically provided help directly, I also introduced this tool as a way for learners to continue exploring (friction-free) on their own after our sessions. New users often ask about the latest features, so it is important they have access to up-to-date information rather than outdated guidance. So the weekend after #TC26 , I rebuilt it from the ground up. The updated version now retrieves live documentation from the official Tableau Help site for each question. I also revised the routing logic, knowledge index, and instructions to align with this approach. If the help documentation does not provide an answer, the tool continues searching through GitHub, the Community, the Knowledge Base, and web search, in that order. Only after these steps does it attempt to generate an answer. This process covers all Tableau products and features, including Cloud, Desktop, Prep, Server, Blueprint, Developer APIs, Mobile, and Public. The information remains current and comes from official sources each time. #DataFam #AIAssistedLearning #TableauConference

One year of writing on LinkedIn… an honest reflection✍️ About a year ago, I started posting regularly, almost every week, sharing my thoughts on Leadership, Quality, and AI in Pharma, Biotech, and MedTech. These are the topics I am very passionate about. With over 20 years in Quality leadership, I had certain views and experiences, but I also wanted to learn, explore ideas, and hear from others. That part has been even more valuable than I expected. I truly appreciate every interaction, comments, messages, even a simple emoji 😊 I did not receive much criticism, which is fine either way, but I have definitely learned a lot through out this process. What I didn’t expect was where this would lead. Through LinkedIn, I connected with the great team at Daikon AI, started supporting them as a Quality Advisor, and learned much more about AI along the way. That also led to something I never thought I would do, joining video podcasts. I’ll be honest, I was never very comfortable on camera. I enjoy real conversations, but recording myself was not something I naturally leaned into. Still, I pushed myself, and it led to more conversations, including another podcast with Hemish Ilangaratne where I shared more of my journey, leadership values, and challenges with global organizations. Looking back, all of this started with something simple, just writing consistently. I don’t claim to be perfect or an authority. I write based on my experiences, my daily work, conversations I have, and what I observe. Some weeks ideas come easily, some weeks they don’t, like this one, where I just use this platform to reflect. And yes, I write my posts myself and sometimes use tools like ChatGPT to help with grammar and clarity, but the thoughts are always mine. I appreciate your support more than you know. Please keep sharing your thoughts, challenging ideas, and engaging in the conversation. Let’s continue learning together around Leadership, Quality, and AI. Thank you all 🙏 #Leadership #QualityLeadership #AI #ContinuousLearning #Gratitude #LinkedIn

17 hours before my DEFCON talk, I got a call from a legal team: “You can’t go on stage. You’re in violation of responsible disclosure.” Months of research now dead in the water. But I kept calm: “I’m not aiming to expose your company and want to get this right. I thought I disclosed this properly.” He was expecting a fight, but got the opposite. The tone shifted immediately. He then asked: “We're not seeing your disclosure. What's the HackerOne case ID?” That's when it clicked. There was never a HackerOne case. I disclosed everything over email. Once that was made clear, the tension dropped, and they supported me going on stage. Lesson: At some point, someone will come at you—hard. Right or wrong doesn’t matter in that moment. Handle pressure well, and you don’t just save the situation… You control the outcome.

With 15 days to go until #TC26 , I’m taking a moment to celebrate the last Tableau Conference I attended in person: #TC24 . This collage is from Day 0, which was full of the best kind of community moments: meetups with Annabelle Rincon, Data + Tacos, and so many familiar faces. I almost didn’t make it that year because my back was in so much pain. Thankfully, with my TENS device running basically 24/7 for the entire conference, I was able to get through it and be there with this amazing community. This year will feel a little different. I’ll be attending as a Salesforce employee instead of as a customer and community leader, but I’m just as excited to see everyone. Registration link is in the comments, and it looks like you can still save $300. Hope to see you there! #DataFam #Data26