In Conversation With

Ashe Magalhaes

Conversations about frontier tech, AI, Aesthetics, and Culture in SF

  1. 5d ago

    Rogue AI Agents & What Should Stay Human | Ben Hylak

    An AI agent can follow your instructions and still do the wrong thing. So how do we know when our agents are going off track—and what should we choose not to automate? In conversation with Ben Hylak, cofounder of Raindrop and former Apple designer, in San Francisco's North Beach. We get into rogue AI agents, agent observability, AI safety, privacy, and why more capable models can create more ways for things to go wrong. And then... the parts of life we don't want to automate. Friendship, faith, love, doing hard things, and having the freedom to build. I'm interested in the people behind the tools, and what we're building all of this for. We also talk about Apple's design culture, rage bait on the tech timeline, AI coding, and what it means to feel cared for by an agent. Dive in: 00:00 Meet Ben Hylak 00:55 From a Pennsylvania farm to Apple 03:10 Building Raindrop and detecting agent failures 06:05 Rage bait and the social cost of tech marketing 11:46 Rogue agents, alignment, and defining failure 18:46 Why AGI would not end mistakes 21:39 What the AI safety conversation leaves out 25:03 Who owns the AI agent interface? 28:38 AI coding, code review, and two-person teams 32:27 Privacy needs more than permissions 35:41 Feeling cared for by an AI agent 37:57 What should stay human? 41:39 Faith, doubt, and discipline 44:48 Love, friendship, and reasons to build 47:46 Growing up with the freedom to make things 50:19 Raindrop, hiring, and closing thoughts Very curious what you think: what would you choose not to automate? Ben: https://x.com/benhylak Raindrop: https://www.raindrop.ai/ Follow In Conversation With for more conversations on frontier tech & the people building it. Newsletter: https://ashe.ai/newsletter Find me on X: https://x.com/ashebytes

  2. Aug 4

    Kimi K3 and Chinese AI, American AI Researcher Debates, and X Influence with Signüll

    In this episode, I sit down with Signull—one of the most recognizable anonymous voices in technology—for a conversation that spans artificial intelligence, open-weight models, entrepreneurship, venture capital, spirituality, and the future of human intelligence. We begin with the story behind Signal’s anonymous Shakespeare persona and how an online pseudonym unexpectedly became one of the most influential voices in the AI ecosystem. From there, we explore what it means to build at the frontier of technology, why incentives shape nearly every public conversation, and how founders, researchers, investors, and policymakers influence the direction of artificial intelligence. We also dive into the rapid rise of open-weight AI models, the implications of Kimi K3, competition between American AI labs and China, whether intelligence becomes commoditized, and what lasting advantages companies like OpenAI and Anthropic can realistically maintain as model performance converges. The conversation then shifts toward creativity, company building, and entrepreneurship: finding leverage, building products people genuinely love, balancing writing with investing, and why uncertainty is an unavoidable part of creating something meaningful. Finally, we venture into spirituality and philosophy—simulation theory, synchronicity, psychedelics, migraines, the Bhagavad Gita, dharma, and the relationship between technology and the search for meaning. Whether you’re an AI researcher, founder, engineer, investor, or simply fascinated by the future of technology, this conversation offers a thoughtful look at where artificial intelligence is heading—and what it means for the people building it. Topics discussed: • Artificial intelligence and large language models • Open-weight AI and Kimi K3 • OpenAI, Anthropic, and the future of frontier models • China and the global AI race • Entrepreneurship and startup building • Venture capital and investing • Incentives, trust, and online discourse • Anonymous internet identities • Creativity and technology • Simulation theory, spirituality, and philosophy • The Bhagavad Gita and finding meaning while building If you enjoy conversations at the intersection of computer science, AI, startups, philosophy, and spirituality, subscribe and follow for future episodes.

  3. Jul 8

    On the Frontier of Ocean Robotics

    Will O'Brien, founder of Ulysses, is building underwater robots and autonomous drones to do the important work of the ocean, for industry, researchers, and the U.S. Navy. In this episode we go deep (literally) on ocean tech, underwater autonomy, illegal fishing, mine detection, UFOs in the ocean, faith, growing up Irish, and why he thinks reading is overrated in San Francisco.We cover the frontier of subsea robotics: why only ~2,000 underwater drones are built each year vs. 50 million aerial drones, the unsolved problem of underwater communications (acoustics vs. optical), decoding whale and dolphin song with AI, and the USS Nimitz "Tic Tac" UFO case. Plus: spearfishing invasive lionfish in the Cayman Islands, the Irish language, trad music, Catholicism, and building a company culture that takes the work seriously but not itself.If you're into ocean technology, robotics startups, marine science, defense tech, or founder stories, this one's for you.🌊 Ulysses is hiring across engineering and more.⏱️ CHAPTERS0:00 Cold ocean dips & morning swims01:05 What is Ulysses? Underwater robots for industry, researchers & the Navy03:38 Growing up by the sea in Ireland05:37 Speed-running friendships from Twitter/X07:44 The Irish language10:58: Will's aesthetic inspiration13:22 The state of underwater autonomy (2,000 vs. 50 million drones)15:08 Decoding whale & dolphin acoustics with AI17:23 Hot take: reading is overrated in SF19:34 Faith journey: Catholic → Buddhism in Nepal → back to Catholicism23:14 UFOs, disclosure & non-human intelligence27:01 Illegal fishing, mine detection & the Strait of Hormuz29:38 The frontier: autonomous subsea robotic manipulation31:07 Scuba diving & spearfishing lionfish (closest call underwater)34:37 Ulysses is hiring🔗 LINKSUlysses: https://www.theoceancompany.com/Follow Will on X: https://x.com/Willob📲 CONNECTSubscribe for more founder interviews on ocean tech, robotics & AI: https://www.youtube.com/@ashe_aiInstagram: https://instagram.com/ashebytesX / Twitter: https://x.com/ashebytes#OceanTech #UnderwaterRobots #Robotics #StartupFounder #DefenseTech #MarineTech #AI #Ulysses #UFO #Ireland #DeepTech #Autonomy #Podcast #SanFrancisco

    On the Frontier of Ocean Robotics
  4. Jun 25

    From Math to Mysticism: On AI's Acceleration of Spirituality

    What happens to meaning when AI accelerates faster than our nervous systems can handle? In this episode of In Conversation With, I sit down with Nicole, astrologer, writer, and rationalist, to trace the path from math to mysticism: why engineers, physicists, and tech builders are turning to astrology, tarot, psychedelics, and the occult during a moment of profound uncertainty.We get into synchronicity and Jung's scarab, the spiritual vacuum left by the Enlightenment, plant medicine and ego death, Rudolf Steiner's warning about Ahriman and the cold mechanistic intellect of technology, non-duality and the singularity, manifestation vs. the shadow, quantum physics and the observer effect, and what really happens when people have deep spiritual experiences with LLMs like Claude and ChatGPT, including local models and "parasitic AI."If your nervous system feels shot to hell and you're searching for something bigger than yourself to hold onto, this one's for you.🔮 Connect with Nicole: X https://x.com/digitlartifacts & Substack https://substack.com/@digitalartifacts🔮 DM Nicole on X to learn more about her chart readings📩 My weekly newsletter: https://ashe.ai/newsletter🐺 Follow me: X https://x.com/ashebytes • Instagram https://instagram.com/ashebytesRecorded during Edge City, Healdsburg @JoinEdgeCity ⏱️ Chapters00:00 Intro: synchronicity & the spiritual vacuum AI is opening up01:21 Why the Enlightenment left us without shared meaning04:23 Where to start when your nervous system is shot to hell07:27 Plant medicine: the real upside and the real risks09:16 Astrology as a mirror, not a fate sentence11:33 Spiritual emergency & overwhelming synchronicity13:42 Rudolf Steiner predicted AI in 1919 (Ahriman & Lucifer)17:46 The 60s, decentralized truth & the loss of a mainstream18:53 Erik Davis, the occult resurgence & conspiracy going mainstream22:11 Reality glitches: Jung's golden scarab story29:12 Manifestation, the shadow & hyperstition31:54 Quantum physics & the observer effect35:40 Ego death, psychedelics & dying before you die39:25 Escape velocity, longevity & SF's relationship with death42:35 Richard Tarnas: the myth of progress vs the myth of the fall44:44 Approaching reality as a mystery, not a machine46:54 How frontier labs constrained spiritual & creative use48:53 Local models, "parasitic AI" & seed phrases between models50:47 Edge City & why in-real-life community matters52:28 The chart reading & astrology as intimate mirror55:17 Esalen, relationships & you can't heal someone for them1:00:09 Tarot, the hero's journey & the tower before the star1:02:11 Religion, Catholicism & the drive toward self-betterment1:03:16 Gender-war content, psyops & emotional manipulation#AIandSpirituality #Astrology #Synchronicity #Psychedelics #TechAndMeaning #Tarot #Consciousness #NonDuality #Manifestation #SanFrancisco

    From Math to Mysticism: On AI's Acceleration of Spirituality

About

Conversations about frontier tech, AI, Aesthetics, and Culture in SF

You Might Also Like