AI in NYC Show

Sponsor: BePresent - https://www.bepresentapp.com/In this episode of AI in NYC, hosts Rob May, Ryan Eppley, and Anna Kirk sit down with David Justus, VP of Applied AI at Panasonic, to explore one of the most underappreciated frontiers in artificial intelligence: edge AI. David breaks down how Panasonic is deploying AI in internet-constrained environments — from in-flight entertainment systems on transatlantic flights to manufacturing floors and stadium video processing — where sending data to the cloud simply isn't an option.The conversation takes a fascinating geopolitical turn as David contrasts how the US, China, and Japan are each taking radically different approaches to generative AI. While the US doubles down on closed, AGI-focused ecosystems and China pushes open-weight models, Japan is quietly building sovereign, domain-specialized AI — including the recently released Rakuten V3 model that outperforms GPT-4o on Japanese-specific tasks. David argues that the US approach may not be great for edge computing and could be starting to show cracks.David also shares insights from Panasonic's research lab, including new work on using diffusion models for document understanding, and explains why the last six months have been a true inflection point for running meaningful AI on small, resource-constrained devices. Whether you're building products, leading an AI team, or just trying to understand where the industry is heading beyond the data center, this episode is packed with perspective you won't hear anywhere else.David Justus brings a background in applied mathematics and computer science, with experience spanning finance, creative industries, and consulting for companies like Verizon, Nike, and Mayo Clinic before joining Panasonic's global applied AI team two years ago.

In Episode 21 of AI in NYC, Rob and Ryan sit down with Matthew Mirman, founder of Chat.Dev — a cloud hosting platform for coding agents that he describes as 'Heroku for Claude Code.' Matt walks us through how he went from building legal tech AI tools for personal injury lawyers to launching a platform where anyone can spin up a cloud VM, run a coding agent in YOLO mode, and even manage the whole thing via text message.The conversation digs into the pivotal moment last November when Claude Code went from 'okay' to genuinely transformative — enabling people with zero coding experience to build full production applications. Matt shares how showing his early prototype to just 20 friends led to 5 of them using it 5+ hours a day within a week, a product-market fit signal that convinced him to go all in.The episode also covers the latest industry news including Anthropic's Mythos model announcement through Project Glasswing, what it means for the AI safety conversation, and whether competitors like Google or state actors already have comparable capabilities. Plus, the hosts break down Amazon's bombshell $15 billion AI run rate in AWS and Andy Jassy's defense of their massive $200B capex spend — reigniting the debate over whether we're in an AI bubble or just the beginning of an infrastructure supercycle.Whether you're a developer curious about cloud-hosted coding agents, a founder evaluating the AI infrastructure landscape, or just trying to keep up with the breakneck pace of AI news, this episode has something for you. Tune in and don't forget to subscribe.Thank you to our sponsor, BePresent. Check them out: https://www.bepresentapp.com/

In Episode 20 of AI in NYC, we welcome back our original guest — Charlie O'Donnell — to talk about his upcoming book 'Founder Unfriendly: What Investors Won't Tell You About Getting Funded.' Charlie has spent years in the NYC venture ecosystem helping founders navigate the opaque, often misleading world of fundraising, and this book is his attempt to arm the 99% of founders who aren't insiders with the real playbook. Charlie breaks down why the feedback you get from VCs almost never reflects the real reason they passed, how junior associates can inadvertently string you along, and why the fundraising process actually starts way before you ever pitch a deck — possibly as far back as high school. He also shares a fascinating look at how he used AI to organize and structure a 250-page book from a messy list of inside-joke chapter titles. We also discuss the emotional arc of founding a company — including why the day you announce your startup might be the most dangerously misleading day of all. If your network congratulated you but didn't offer a single customer intro, that's a signal worth paying attention to. Whether you're a first-time founder or a repeat entrepreneur, this episode is packed with honest, practical insight you won't hear in a typical VC blog post. Sponsored by BePresent (bepresentapp.com) — the #1 app in its category that uses social media engagement techniques to keep you OFF your phone. Also: join us April 15th at 5:30 PM for our first in-person Cloud Code class for non-technical people in NYC!

In this episode of AI in NYC, Rob and Ryan sit down with Zachary Smith, co-founder of Datum, to explore why the foundational infrastructure of the internet needs a radical overhaul for the AI era. Zach — a lifelong New Yorker who traded a classical music career at Juilliard for the wild world of Linux web hosting in 2001 — brings a rare depth of experience spanning multiple companies, acquisitions, and a front-row seat inside Equinix, one of the largest interconnection companies on Earth.Zach introduces his concept of the 'splinternet' — a world where geopolitics, regulation, and the demands of AI workloads are fragmenting the once-unified internet — and explains why Datum is building an open network cloud to serve the next wave of what he calls 'alt clouds': the roughly one thousand (and growing) new cloud providers that don't fit neatly into the old hyperscaler model. From Databricks to GPU startups to niche SaaS platforms, Zach argues these alt clouds need shared infrastructure primitives they can't afford to build alone.The conversation also gets deeply personal. Zach opens up about the emotional toll of selling his first bootstrapped company after 11 years, the therapy and intentional downtime he needed before starting again, and the unique dynamic of building multiple companies with his identical twin brother Jacob. His mentor Bill Luby's advice — 'this is the best time because you have no past' — becomes a throughline for how Datum approaches building for the long term in a world obsessed with speed.Whether you're a founder, an infrastructure nerd, or just curious about how the physical internet actually works, this episode is packed with insights about what's changing beneath the surface of every AI application you use. Tune in for one of the most thoughtful conversations we've had about the invisible plumbing of the internet age. Check out datum: https://datumdata.ai/ Thank you to our sponsor: BePresent - Download their app here: https://apps.apple.com/us/app/bepresent-screen-time-control/id1644737181

What happens when AI meets your brain data? In Episode 18 of AI in NYC, Rob May and Anna Kirk sit down with Kristen Mathews, cyber/data/privacy partner at Cooley LLP with nearly 30 years of experience, who has carved out a fascinating niche at the intersection of privacy law and neurotechnology. Kristen breaks down what neurotech actually is — from invasive brain implants to consumer wearable headbands — and explains how AI has been the key catalyst turning a century of brain signal data into actionable, decoded information. The conversation dives deep into the different categories of neurotech, including how devices can not only read brain activity but also stimulate it — with real applications like predicting seizures 20 minutes before they happen and suppressing them with electrical pulses. Rob shares his firsthand experience from sitting on the board of a neurotech company, while Kristen paints a vivid picture of the current landscape, including New York City's role as a major hub for the neurotech community. Perhaps the most thought-provoking segment explores the ethical frontier: the difference between decoding 'intended speech' (helping ALS patients communicate) and 'inner speech' (your private thoughts). Where's the line? Can AI tell the difference? Kristen is refreshingly honest about what we don't yet know, while emphasizing that every neurotech application she's seen in practice today is being used for good. This episode is essential listening for anyone interested in AI, privacy, the future of brain-computer interfaces, and why the next big privacy debate may be about your thoughts. Relevant links: www.nytimes.com/2003/06/22/magazine/savant-for-a-day.html https://icaot.org/jose-delgado-a-controversial-trailblazer-inneuromodulation/ Kristen Mathews on Linkedin: https://www.linkedin.com/in/kristen-mathews-6025257?utm_source=share&utm_medium=member_mweb&utm_campaign=share_via&utm_content=profile Download BePresent: https://apps.apple.com/us/app/bepresent-screen-time-control/id1644737181