Horizontal puts AI agents between factory systems and daily disruptions

A stopped production line creates work well beyond fixing a machine. Orders need new schedules, material availability needs checking, and someone has to decide what happens to the unfinished batch. Horizontal is building AI agents to coordinate that work across the software a factory already uses.

The startup launched on October 21 with a modular approach to factory operations. Rather than asking manufacturers to replace their planning, maintenance and quality systems, it puts agents above them to carry out a process that spans several tools. Customers can begin with one operational problem and add more agents over time.

A demonstration on its website illustrates the idea with an overheating drive. The displayed workflow reads a fault and the running order, checks maintenance history, opens a work order and holds recent output for inspection. It then identifies orders at risk of missing their deadlines and proposes moving or resequencing them.

That example ends with an approval step for the shift lead. It is a product illustration, not evidence of a measured customer outcome, but it makes the intended division of labor clear: software gathers the connected facts and prepares a response, while a person can approve or adjust the plan.

Watch Horizontal's launch demonstration.

The launch describes agents for production replanning, materials, quality investigations, batch release and shipments. These jobs depend on context that can sit in different systems. A maintenance record alone doesn't reveal which customer order will be late; a production schedule alone doesn't explain whether a batch is ready for release.

Horizontal's proposed value is in connecting those decisions to actions. Its website describes operating on top of enterprise resource planning, manufacturing execution and quality tools, with deployment options in the cloud or inside a customer's network. The company also describes logging agent actions and requiring approval for critical steps.

Founders Niklas Marxer and Jakob Hermeler are mechanical engineers who met through robotics at Stanford, according to their launch announcement. The company is in Y Combinator's Fall 2026 batch. Their product focuses on the coordination around manufacturing, where an operational exception often requires several teams to reconcile the same facts before anything changes.

The broader ambition is autonomous factory operations, but that is a direction for the company rather than a demonstrated claim that an entire plant can currently run without people. The modular rollout is a more concrete proposition: put an agent on a bounded workflow, observe its behavior and expand from there.

The announcement doesn't establish how much of an individual manufacturer's process can be automated without custom integration work. That will depend on the systems involved and the decisions the customer is willing to delegate. Horizontal's starting point is deliberately narrower than full autonomy: help a plant respond to the next disruption without making its staff manually carry every update from one system to another.

‍