Serg Masís

Sometimes the smallest projects are the most rewarding. I’ve spent six years working in AI and machine learning for agriculture, yet I never felt as connected to farming as I did after watching my own garden fail. A week away at a conference was enough. Temperatures topped 95°F, there was no rain, and every plant paid the price. So I started over. The new setup is fairly simple: • Plants arranged according to their light and water needs • Shade cloth to reduce heat stress • Soil moisture and temperature sensors • Smart drip & spray irrigation and misting controlled remotely • Two small random forest models to decide when to irrigate and when to cool the canopy The irrigation model predicts whether soil moisture will cross a critical threshold and continue to dip before the next rain. The misting model activates only when temperatures are expected to climb and humidity is low enough that evaporative cooling is worthwhile. I have yet to make the shade cloth retractable with motors, but that’s my next project :) It’s hardly cutting-edge AI, but that’s the point. Sometimes a simple random forest is exactly the right tool. I’ve included a few photos and a timelapse of the dashboard. The best outcome is simple: I can travel without worrying that my garden will dry out, while using only the water it actually needs.

My first day at the Ai4 - Artificial Intelligence Conferences was magical! I met some great AI people (and robots!). And today I’m looking forward to meeting some more. I’ll be speaking today at a panel discussion titled “Deploying AI Agents for the Real World” at 10:30 PDT for anyone that is at the conference (Veronese Ballroom 2504) and would like to learn more about the challenges and benefits we gain from agentic AI. I’ll bring my perspectives from Agriculture. It should be very fun! For good luck I couldn’t help but take a picture with the VERY REAL godparents of AI #NotADeepFake 😂 PS: yes, I have a 3D printed cast since I broke my wrist a few weeks ago.

Sometimes the best AI conference takeaway isn’t technical at all… At Ai4 conference, Fei-Fei Li said something during a panel with Andrew Ng and Geoffrey Hinton that left a lasting impression. She was concerned about the impact of AI on children’s education because learning requires motivation, and simplifying everything can diminish the motivation to learn. As the father of a 2-year old and a life-long learner, that resonated with me. Motivation doesn’t just make learning possible. It’s what we are wired to do. It drives almost everything humans choose to excel at. That was just the start of a packed second day. I then, joined SImba Khadder, Waqas Makhdum, and Rahul Zende (moderated expertly by Ram Kumar N. ) to discuss “Agents in the Real World”. There wasn’t much disagreement, which made it an enjoyable exercise in “yes, and.” I added my perspective from agriculture to an already thoughtful conversation. <Photo credit: @stevehall> Lunch was with The Velocity Room , where I met Brad Tompkins and Richard Piasentin, some great colleagues, and caught my good friend Cal Al-Dhubaib speaking. Throughout the day, I attended talks on AI-powered drug discovery, agent-human collaboration, self-healing data pipelines, and plenty more. Then dinner organized by the kind folks at CDO Magazine (Gerri Caveness , Anthony Losanno, Sheila Garcia) had veteran data leaders from every industry imaginable and all over the country. I loved hearing the career stories around my table before “Hot Take” time with every table turned dinner into a much broader and livelier discussion. I’ve been wanting to write more about Ai4 - Artificial Intelligence Conferences for weeks now. It seems like attending conferences is easier than finding time to write about them afterward. Or maybe it’s just one more skill I need to master!