Episode summary
AI Demos Are Easy. Real Engineering Is Hard
Episode Summary
AI Demos Are Easy. Real Engineering Is Hard
What does it take to make AI useful in real engineering and construction workflows?
Recorded at ATOMS ACCELERATED, hosted by DeepMechanix during a16z Tech Week San Francisco on October 7, 2026, this panel explores Building Industrial Systems in the Age of AI.
• Where AI is becoming part of everyday engineering work
• Why real-world data matters more than a polished demo
• Building in-house versus buying from startups
• Physics AI, simulation, and the challenge of earning trust
• Overcoming pilot fatigue and demonstrating customer value
• Moving from individual AI tools toward connected, autonomous workflows
The panel
• Kanav Hasija — Founder & CEO, Meltplan
• Michael Finocchiaro — Founder & CEO, ThreadMoat
• Noah Weber — Co-Founder & CTO, Cosmon
• Moderator: Gustavo Navarro — Founder & CEO, Divergent Physics
Thank you to DeepMechanix for bringing the community together, and to everyone who contributed questions.
Explore the engineering and industrial software startup landscape:
https://threadmoat.com
Which engineering workflow would you trust AI to take on next? Share your thoughts in the comments.
#EngineeringAI #IndustrialAI #TechWeek #BetterCallFino
Full Transcript
Transcript pending.
