DeepTech Unleashed Winning at Physical AI: A Founder’s Playbook with Igor Marinelli
Mar 18, 2026
Igor Marinelli, founder and CEO of Tractian and a mechanical engineer, recounts building a Physical AI firm that monitors industrial assets. He discusses starting from customer pain, iterating fast with focused products, and why owning rugged sensors matters. Topics include data scarcity as a moat, field deployments and human ties, empirical failure testing, rapid decision-making, and building a sales and fundraising playbook.
AI Snips
Chapters
Transcript
Episode notes
Let Sales Drive Product Development
- Make product follow sales: map customer pain first and build the minimum that moves the needle in sales.
- Prioritize proximity and velocity with customers—visit them and iterate fast to create retention and expansion.
Attack One Pain Like A Doctor
- Start extremely focused on one pain point and be a 'doctor' who prescribes the right fix rather than a 'nurse' who gives what customers ask for.
- Say no often and concentrate on the single problem that delivers tangible impact.
Physical AI Is Manufacturing Scarce Real-World Data
- Physical AI is defined by manufacturing proprietary, scarce real-world data via deployed hardware, creating a durable moat.
- Improved connectivity and cheaper hardware have made large-scale physical data manufacturing feasible and strategically valuable.