Two Nines Isn't Enough: Why Workflow Reliability Is Robotics' Real Dexterity Test

A robot that succeeds 99% of the time sounds ready for the factory floor. The math says otherwise, and Nicolas Sauvage, founder and president of TDK Ventures, argues the gap between demo performance and deployable reliability is where the industry's next battle will be fought.

The Compounding Math of Reliability

The core problem is arithmetic. Consider a workflow that requires 100 physical actions in a sequence. At a 99% success rate for each action, the probability that all 100 succeed on the first attempt is just 36.6%. Increase the success rate to 99.9%, and the probability of completing the workflow rises to 90.5%. At 99.99%, it reaches 99.0%.

"A robot with a 99% success rate is not necessarily 99% automated."

That is the uncomfortable takeaway. Small reliability gaps become much bigger when repeated across an entire workflow, which means a robot that looks almost flawless in a demonstration can still be far from ready for autonomous deployment. The difference between two nines and four nines may ultimately decide whether a robot is ready to work alone.

Reliability Is a System Property

Useful dexterity isn't simply the ability to complete an individual action once. It is the ability to sense, grasp, move, adjust and recover reliably enough to complete an entire workflow without constant human assistance. Recovery, in particular, is a critical component of robotic dexterity — and it only improves through exposure.

Deployment in real-world environments is essential for improving robotic systems. Each deployment cycle feeds learning that makes the next one more reliable. That, in turn, reframes how progress should be measured: the success rate of complete workflows is a better indicator of robotic readiness than individual action success rates.

Form Follows the Job

Sauvage's portfolio illustrates the principle that the right physical form for a robot depends on the specific job it needs to perform.

• Agility Robotics: built Digit, a bipedal robot that connects to specialized tools rather than relying on a human-like hand.
• ANYbotics: deploys quadruped robots that have conducted over 33,000 inspections across 450 inspection points at industrial sites.
• Starship Technologies: uses small wheeled delivery robots, an approach behind more than 10 million reported autonomous deliveries.

Three very different machines, one shared logic: match the body to the work, then drive reliability up through deployment. Robotic dexterity is key to commercial success, he notes — but dexterity measured by workflows completed, not fingers counted.

TDK Ventures, the venture arm of TDK Corp, manages $500 million across four funds, backing startups across AI, energy, industrial and robotics, mobility, and advanced materials. The industry challenge ahead, as Sauvage frames it, is turning robotic capabilities into reliable systems customers can trust.