For founders building in the age of AI, the focus often falls on the model itself: the algorithm, the data, the intelligence. But what if the model is just a sliver of the actual solution? Peter Caster, a leader at Applied Intuition, just dropped a truth bomb for anyone tackling physical AI: the core model is only one percent of what you're actually building.

His company, Applied Intuition, just launched Dana, an agentic platform that’s the culmination of a decade of work. Dana tackles the messy, often ignored 99% of the problem that makes physical AI excruciatingly hard, especially in safety-critical applications. Think robots, autonomous vehicles, or any machine where human lives are on the line. Caster is clear: “The model is is not enough. That's like 1% of the full solution.”

Key Takeaways

  • Applied Intuition's Dana platform is designed to make complex physical AI development as accessible as building an iPhone app.
  • Peter Caster states that the core AI model accounts for only 1% of the solution in safety-critical physical AI systems.
  • Dana orchestrates the remaining 99% of the physical AI technology stack, including toolchains and data science, via an agentic, plain-English interface.
  • The platform aims to lower the barrier to entry so that any engineer—and ultimately anyone—can develop robot solutions, according to Kensho S.

The 99% of AI No One Talks About: Safety and Orchestration

Most founders get excited about the intelligence – the shiny new large language model that can process data or generate actions. But for physical AI, particularly in areas like autonomous systems, the stakes are far higher. As Caster puts it, you're dealing with “things where lives are literally on the line based on on what is being developed.” This isn't just about making something smart; it's about making something safe, reliable, and deployable in the real world.

This is where the 99% comes in. It’s the intricate web of complex development toolchains, data science workflows, testing, validation, and production deployment that surrounds the core model. These elements are non-negotiable for safety-critical systems. Applied Intuition’s new Dana platform acts as the conductor, simplifying these workflows “seamlessly orchestrated from an agentic interface where you're able to to write things in plain English and receive answers in plain English,” Caster explains.

Kensho S., his colleague, underlines the mission: “The thing that Peter's really emphasizing and thing that we we really want to do is just lower the bar. So anybody really out there, starting with engineers but ultimately really anybody can develop you know, robots.” The goal is to democratize robot development, making it as straightforward as coding an iPhone app. This doesn't just reduce friction; it “supercharge[s] our customers,” allowing them to innovate faster and build a wider variety of solutions.

What to Do With This

If you're building in physical AI, or any complex, safety-critical system, map out the entire 100% of your solution. Don't stop at the core intelligence or model. Identify every toolchain, testing protocol, data pipeline, and deployment step. Then, instead of building all 99% yourself, actively seek agentic platforms or specialized tools that can abstract away that complexity, allowing your team to focus on the unique 1% that truly differentiates your product.