While humanoid robots capture headlines, Jake Lusararian sees a higher-stakes arena: industrial infrastructure. On *This Week in Startups* and *This Week in AI*, the Gecko Robotics CEO argued that after 13 years, his company’s purpose-built robots - crawling inside tanks and scanning bridges - prove durable value comes from deterministic data, not speculative demos.
In energy or defense, an AI model’s wrong prediction can cause an explosion. This need for mission-critical, hallucination-free data separates industrial robotics from consumer AI hype. Lusararian’s focus is on preventing physical catastrophes, a requirement that makes flashy humanoids irrelevant for now.
Jake Lusararian, This Week in AI:
- The models are putting a huge spotlight on the importance of valuable data sets that don't hallucinate.
- Especially with things that if they do hallucinate, could cause an explosion and kill people.
The drive to deploy reliable AI into the physical world hits a parallel bottleneck in computing hardware. Chris Lattner of Modular, also on *This Week in AI*, described an ecosystem fragmented by proprietary stacks from Nvidia, Apple, and AMD. This lack of software portability, which he called “duct tape and bailing wire,” stifles the innovation needed for specialized industrial applications.
This pragmatic, hard-tech ethos extends beyond AI models to the operating philosophy itself. On *The a16z Show*, SpaceX and Tesla alumni like Chandler Lujica argued that incumbents in sectors like defense and mining fail due to slow decision velocity and poor software integration. Their playbook: attack the hardest technical problems first and build proprietary operating systems to prevent data silos.
The consensus across these founders is clear. The near-term AI opportunity isn't in mimicking humans, but in building reliable, integrated systems - both robotic and computational - for the physical industries that underpin the economy.
