Viam Panel Discussion Highlights Key Challenges in Scaling Physical AI Deployments
Key Takeaways
- ▸Rapid iteration capability is the critical differentiator between successful physical AI deployments and those that stall, requiring software engineering practices adapted to hardware development
- ▸Industry leaders advocate for human-machine collaboration over full automation, recognizing that real-world complexity necessitates keeping humans in the loop for judgment and decision-making
- ▸Hardware constraints, particularly in mechanical components like robotic hands, are becoming the primary bottleneck as AI software capabilities advance faster than physical systems can implement them
Summary
On February 19, 2025, robotics platform company Viam hosted a panel discussion at its headquarters bringing together industry leaders to discuss the realities of deploying physical AI systems at scale. The panel featured Simone Kalmakis, VP of Engineering at Viam, Nicole Maffeo, Cofounder of Gambit Robotics, and Kirin Sinha, Founder and CEO of Illumix, with moderation by Laura Rippy, Managing Partner at Alumni Ventures. Over one hundred attendees gathered to hear insights on practical challenges facing physical AI implementation in production environments.
The discussion emphasized that rapid iteration is crucial for success in physical AI, with panelists noting that unlike pure software development, physical AI requires constant experimentation across hardware, sensors, and algorithms. Maffeo highlighted how Viam's platform enables systematic debugging and quick iteration when prototypes fail. The panel stressed a human-augmentation approach rather than full automation, with the philosophy of "human with machine, not human versus machine" emerging as a key theme. This collaborative model acknowledges that real-world environments remain messy and unpredictable, making human judgment essential for complex decision-making.
The panelists also addressed the gap between AI software capabilities and hardware readiness, with Sinha predicting a period where the industry waits for mechanical components to catch up to software advances. Infrastructure challenges including edge computing requirements, latency concerns, and connectivity costs were identified as critical deployment considerations. The discussion offered a measured perspective on humanoid robots, suggesting they may not achieve large-scale deployment in the immediate future, while emphasizing that specialized physical AI applications present significant near-term opportunities.
- Edge computing and local processing are requirements rather than optimizations for many physical AI applications, addressing latency, connectivity, and privacy challenges in real-world deployments
