Episode: Talk4 Podcast EP165
Guest: Bryant Cruse — Founder & CEO, New Sapience | Former U.S. Navy Aviator | Space-Systems Engineer (Hubble operations)
Watch on YouTube: https://youtu.be/CXX53BN0FFg
Listen on Spotify: https://open.spotify.com/episode/0BSa8MeoFDshkGqTJ7jaPU?si=8b9fb8fe4cb547b8
What if AI actually knew things?
In EP165 of the Talk4 Podcast, I sit down with Bryant Cruse — a Naval aviator turned space-systems engineer who helped pioneer AI within Hubble Space Telescope mission operations, and is now building what he calls artificial intellect at New Sapience. This isn’t just another chatbot story. Bryant’s approach starts with knowledge, world models, and determinism — aiming for AI that won’t hallucinate, runs on-device, respects privacy, and is designed to help humans live better lives.
If you’ve ever wondered, “What actually makes intelligence intelligent?”, this episode is your map.
⸻
Who is Bryant Cruse?
• Naval Aviator: Flew the C-130; learned cockpit discipline, systems thinking, and human-machine trust under pressure.
• Space-Systems Engineer (Hubble): Brought early AI tools to mission operations; saw the limits of humans reading tables of numbers in real time.
• Founder & CEO, New Sapience: Building a knowledge-centric, deterministic cognitive core — an “artificial intellect” designed to understand the world instead of predicting the next word.
⸻
Watch / Listen
• YouTube (full video): https://youtu.be/CXX53BN0FFg
• Spotify (audio): https://open.spotify.com/episode/0BSa8MeoFDshkGqTJ7jaPU?si=8b9fb8fe4cb547b8
⸻
What we cover (at a glance)
• Why knowledge ≠ tokens: LLMs are brilliant at language; Bryant argues intelligence requires a structured internal model of reality.
• “Atoms of thought”: How New Sapience decomposes concepts into primitives and builds them back up into a usable world model.
• Deterministic by design: Why his cognitive core can’t hallucinate in the LLM sense — and why that matters for trust, safety, and reliability.
• On-device companions: Practical apps that understand context (“Take metformin after lunch”), without forms, menus, or data centres.
• Human-centred guardrails: Privacy, agency, and AI that helps you be more human — not a replacement, and not a simulated person.
Key ideas & takeaways
1) Intelligence needs something to think about
LLMs process language patterns. Bryant’s view: intelligence + knowledge are two sides of the same coin. Without a robust knowledge structure (a world model), “smart” outputs can still be wrong or incoherent in practice.
2) Models are tools — judged by utility
From celestial navigation tables to inertial nav systems, different models serve different jobs. New Sapience applies that philosophy to AI: build the right model for the task, then reason on top of it.
3) Determinism builds trust
If your system is deterministic and grounded in an explicit model, you can verify behaviour, audit reasoning, and avoid hallucinations as a class of failure. That’s essential for high-stakes domains.
4) On-device ≠ toy
New Sapience’s goal is a compact cognitive core that lives on your phone. That means privacy by default, lower energy use, and apps that feel like talking to an informed assistant — not clicking through a call-centre menu tree.
5) Education still matters — massively
Even if AI can answer questions, your internal model still shapes your life. Bryant’s stance: AI should mentor, uplift, and help you assemble your own mental “puzzle” — not replace learning.
⸻
Memorable moments
• From cockpits to control rooms: Why a pilot’s “scan” and a mission controller’s dashboard demand clear, human-centred interfaces — not walls of numbers.
• First principles over hype: The honest post-mortem of past AI waves (expert systems) and how they informed a knowledge-first architecture.
• Humanity as the end goal: New Sapience avoids building simulated lovers/friends; instead, its companions are designed to nudge you into real life — better habits, better decisions, better relationships.
⸻
Why this matters now
We’re in a moment where AI can sound right while being quietly wrong. Bryant’s work tackles that directly: build knowledge, reason deterministically, stay on-device, and optimise for humans. Whether you’re shipping products, leading teams, or just trying to make better decisions, this lens will sharpen how you evaluate and use AI.
⸻
Links & resources
• New Sapience — About: https://newsapience.com/about/
• Invest (crowdfunding portal): https://investnow.newsapience.com/
• Bryant on LinkedIn: https://www.linkedin.com/in/bryantcruse/
• Bryant on X: https://x.com/bryantcruse
⸻
Listen to the full conversation
• YouTube: https://youtu.be/CXX53BN0FFg
• Spotify: https://open.spotify.com/episode/0BSa8MeoFDshkGqTJ7jaPU?si=8b9fb8fe4cb547b8
If the episode hit you like it did me, share it with a friend, drop a rating on Spotify/Apple, and tag me with your biggest takeaway. Let’s keep pushing for AI that’s useful, safe, and genuinely human-centred.
⸻
About Talk4
Talk4 is where extreme performance, aviation, leadership, and technology collide — with practical takeaways you can use today. Hosted by Louis Skupien. New episodes weekly.
Follow:
IG: @louisskupien | @talk4podcast
X: @louisskupien
LinkedIn: Louis Skupien
Website: louisskupien.com
