Can plain English replace a 1991 factory controller?

Industrial control software is not where most people expect a plain-English AI prompt to end up. Neuron Industries wants to put it there. The company announced a $10 million seed round on October 29 to develop an AI-native controller and the software around it, according to its funding announcement.

The proposed system pairs Cortex, a physical controller, with Synapse, a development environment. Neuron says engineers can describe machine behavior in ordinary language, generate readable control code, inspect and edit that code, then test it in simulation before deployment. This is a deeper bet than adding a chatbot to an existing factory tool: the company is trying to rebuild both the hardware and its programming workflow.

The most concrete example in the announcement is a retrofit at CureWood Engineered Building Products. Neuron says it replaced a Siemens S5 controller dating to 1991 on a finger-jointing line with 189 hardwired input and output points. The company reports about three hours of human and agent engineering and eight hours of downtime. Those are company-reported figures from one project, not an independent benchmark.

Neuron says Cortex is planned for release in the first quarter of 2027. Factory controls have to behave predictably around real equipment, so readable code and simulation are useful design choices, not proof of safety by themselves. The compelling question is whether the workflow can make modernization faster without making verification harder. Industrial operators will need repeatable results, clear failure handling, and careful commissioning before a striking retrofit becomes a dependable product.