Standard Bots says its new Locate capability cuts robot programming time by 90%, and the Glen Cove, N.Y.-based company is betting that natural-language prompting and machine vision can eliminate the waypoint-by-waypoint teaching that dominates deployment cost in small and mid-sized workcells.

The claim sits at the center of three capabilities — Agent, Locate and Inspect — that the company has added to its Standard Bots AI platform. All three target the same bottleneck: the engineering hours required to get a cobot (collaborative robot, a robot designed to work alongside people) from crate to production.

What does each capability do?

• Agent converts natural-language task descriptions into robot programs. An operator describes a machine-tending routine in plain language; the software then asks clarifying questions about the application and builds the program with the operator rather than for them.
• Locate uses the robot's own video feed to identify and track parts. Instead of manually defining coordinates across multiple waypoints, operators tap an object in the feed or describe it in text. Standard Bots positions this for high-mix production (facilities running many part variants), where frequent changeovers make hand-taught waypoint grids expensive to maintain.
• Inspect adds automated vision inspection to robot routines. Built in partnership with Roboflow, it runs custom vision models trained on a manufacturer's own parts to catch defects such as scratches. Manufacturers can load their own inspection models into the platform.

Who said what

The company's argument is that AI-based programming removes the dependency on systems integrators (third-party firms that engineer and commission automation cells), which it frames as a cost and control problem.

"You used to have to hire an integrator; it costs a million dollars. You spend months dialing it in. And when something breaks in the middle of the night, you can't fix it yourself," said Leif Jentoft, head of applied AI at Standard Bots. "It's like a literal black box on your own shop floor."

Jentoft said the platform changes the working model for the operator: "With Standard Bots AI, you don't have to program every position or keep track of where you are in a routine. You simply show your robot what you want it to do, and it will adapt to higher-mix tasks with simpler, more reliable programs."

What does this change for plant teams?

For manufacturing engineers and maintenance staff, the practical shift is ownership. If natural-language and vision-based programming work as claimed, changeovers in high-mix cells stop requiring re-teaching grids of waypoints, and inspection stops requiring a separate vision integration project. The platform also accepts manufacturer-supplied models rather than locking users into Standard Bots' own — a governance point worth noting for shops with existing vision stacks. Standard Bots is additionally developing its own foundation models (large AI models trained broadly and adaptable to many tasks), which suggests the company intends to compete on its models, not just its tooling.

Treat the 90% figure as a claim

The 90% programming-time reduction comes from the vendor, tied specifically to Locate eliminating manual waypoint grids. No independent production data accompanies the launch. Integrator-dependent deployments routinely take months and large budgets, as Jentoft describes, but plants evaluating the platform should benchmark it against their own changeover times and part-variant counts — high-mix machine tending and inspection are where the economics are most likely to hold.

What to watch next: third-party validation of the 90% figure in live production, adoption among tier-one and tier-two suppliers running high-mix cells, and whether Standard Bots' in-house foundation models arrive as shipped products or stay on the roadmap.