Watershed

If we can measure it, we can manage it. Today, Watershed introduced the first open framework for measuring corporate AI emissions. Putting real data behind what's previously been a black box is a critical step toward responsible AI usage. Framework below, plus helpful recommendations like: 👕 choose the right model for the task >> no need for Fable if you're just looking for light edits on an email to your dad (am i the only one who does this? 😂 ) 🌍 grid regions matter >> when you can, choose regions with cleaner electric grids 📏 keep your prompts concise >> reduce tokens 💰 , save power

This is exactly why we started the Watershed AI Fellowship. Katie's work on turning messy environmental data into clean, auditable outputs is the kind of practical application that makes AI genuinely useful for sustainability teams. Excited to see these capstone projects come to life over the next few weeks.

Today, I had the privilege of joining smart people on a fun panel hosted by Dalus Capital: "AI: Threat, Opportunity, or Both?!" (Spouse, acting like he learned about Claude more than a month ago: "Both, obviously!") Great conversation, great questions, and it prompted some thoughts: 🧨 the core tension: AI can accelerate climate insights & action—something we see firsthand in the companies we work with at Watershed. At the same time, it's causing unprecedented grid & water stress, notably here in Latin America. How do we actually ensure the benefits outweigh the costs? We need good data. 📊 ⚡ the opportunity: What if AI energy demands happen at the right pace, in the right places, to actually accelerate clean power deployment? To borrow from the inimitable Ayana Elizabeth Johnson: could we actually get this right? We need good actors. 🙇 🧲 the need: Clear guidance on how to mitigate AI resource consumption. For me, it starts with 2 questions: 1) Is AI the best tool for the job? Just because it's new and cool doesn't mean we need it for, say, simple math. 2) Are you using the right model? You don't need to use the most powerful AI model to draft talking points for your next panel. Start lightweight; add model power if you really need it.* *Do we need the capybara-doing-ballet video? Sometimes, it's absolutely essential. But maybe not every time. 🩷

Plenty gets in the way of true decarbonization. Top blockers for sustainability teams include 1) how much time they have to spend wrangling / cleaning / struggling through data, and 2) lack of clear data on where & how to decarbonise. Today, we announced Watershed agents to take care of all things data cleaning, analysis, and insights so companies can make time for higher-impact work. Last week, I had the chance to interview the dream team of Emma Bayliss-Chan from Royal Mail , Christian Boothby from Smiths Group plc , and Nadia Carroll from Specialized Bicycle Components to hear how they're putting agents to use to free up space for hard decarbonization work. Hear from them here: https://lnkd.in/eNm5A2X4 Last: to reach net zero, real incentives are crucial. While the long-term cost of climate action is clear (there's not much economic growth in a world ravaged by climate change!) and some short-term actions are zero-cost, there are short- and medium-term costs to critical actions that drive us towards a world that limits warming. The right policies help corporates justify (and benefit from!) taking action. Environmental policy and [overly complex] disclosure have been conflated to the point where people have an allergic reaction to the former. If you care about an abundant world, stay engaged. We've all been reminded recently that a transitioning, electrified world is more stable, anyways (see: the price of oil; interest in EVs this month). Thank you to Meredith Hein Mallory Pickett Matt Plays Andrew Lakata Meghan Keaney Anderson Alexa Schirtzinger Hannah Edgerly Avi Itskovich and Yubing Zhang for making this launch possible!

This is the kind of use case that gets me excited about AI again. The world has this new technological superpower in AI, and it's being applied to everything from space discovery to spam emails. The seemingly endless array of uses from the banal to the extraordinary, underscores that AI is only as good as what we point it at. We should pay attention to what AI makes easier, but we should pay greater attention to what it makes possible. This morning, Watershed introduced new sustainability AI agents for data cleaning and analysis. Anyone who has spent a nanosecond in sustainability knows that it is hamstrung by data. In fact, it's one of the hardest, messiest data problems to exist in business. The agents announced today will automate the data cleaning, correction, and analysis work that has consumed sustainability teams for years. That alone is worthwhile for those that have been stuck on a data management treadmill when they'd rather be driving down corporate emissions. But these agents also go further to surface insights hidden in the data so sustainability teams can make greater progress faster. One early customer told us these agents conducted analysis that would have taken three analysts weeks to uncover, and did so with full data lineage and explainability. This is where breakthroughs can start to happen. And I start to get excited about AI again. Congrats to everyone at Watershed who worked on this release and the scores of alpha customers who pushed these agents and sharpened our view on what they could do. Thank you to Yubing Zhang for her continued vision on how to advance sustainability through Watershed AI, and special congrats to Anna Cerf who led the charge on this release for product marketing and the rest of the marketing team who helped bring it to life. Links to learn more in the comments.

Big day today as we launch Watershed agents -- with the goal of enabling the superheroes on sustainability teams to spend less time stuck in data and more time decarbonizing. I love this video of Emma Bayliss-Chan at Royal Mail , Christian Boothby at Smiths Group plc and Nadia Carroll at Specialized talking about their work. More details at the product page in the comments.

All good things come to an end, and I’m moving on from #Watershed . Watershed is a fantastic company that has been great for me, as I’ve immersed myself in a mission that is both important and I’m passionate about. The experience of working in such a fast-moving, category-creating company with so many real experts devoted to solving a meaningful problem has been awesome. Watershed is a leader in so many ways, not least sustainability AI. Thank you to everyone who helped me learn. Thank you to the customers and prospective customers I worked with: some of the most impactful organisations in the world. Please keep doing what you are doing to change climate change and drive the transition to clean, secure energy. Thanks Charles Bergevin , Taylor Francis , Robert Hayden for your trust and leadership. And to everyone else at Watershed, too many to mention. Best wishes to you all. #changeclimatechange