Can GPT-6 build Ignition Perspective pages? An AI data center demo

Yes. I used GPT-6 Astra in Codex to build a working Ignition Perspective demo on a local Ignition 8.3.9 Gateway. The page shows an AI data center with six GPU racks, two cooling pumps, and a UPS. All process values are simulated.

Thermal, Power, and Cooling change the readings over the same equipment scene. Thermal shows rack inlet temperatures, Power shows electrical demand, and Cooling shows loop conditions and pump feedback. Clicking equipment opens its detail page. Navigation goes one level deep, with a route back to the facility.

The GPU rack page shows utilization, power, inlet and outlet temperatures, and fan speed. Changing the workload target makes utilization move toward the requested value. Power and temperature respond through the simulation, and the trend shows the response over time.

The pump page shows speed, flow, pressure, and motor power, alongside start, stop, and fault-reset controls. In our duty-pump failure scenario, flow falls and the zones warm up. Starting the standby pump restores flow, then temperatures recover gradually. Acknowledging an alarm does not repair the pump fault.

The UPS page shows input and output power, battery state of charge, and estimated IT runtime. During the utility-loss scenario, the battery supports the six racks. Cooling still depends on utility power in this simplified model, so keeping the racks powered does not keep the room cool.

We started in GPT-6 Pro with the demo idea and asked for a build plan. Before touching the Gateway, we chose the equipment, navigation depth, readings, and fault and recovery behavior. That produced a written plan and a tag catalog.

We brought those files into Codex with the Perspective and Jython skills. The skills supplied Ignition-specific instructions for components, tag bindings, script context, and Jython 2.7.4. GPT Image created the equipment artwork. Codex placed native Perspective readings, buttons, charts, and alarms over it.

This distinction matters when repeating the build. The artwork provides the equipment scene; the Perspective components provide the data, navigation, and controls.

With API-key access to the local Gateway, Codex imported the project and created 241 tags and three UDT definitions. UDTs are reusable tag templates for equipment types. The Jython simulation updates the process values through a shared Gateway timer, so separate browser sessions see the same process.

The trends start with five minutes of simulated history already loaded. That gives the presenter something to inspect immediately. Those readings are synthetic history, not measurements collected from physical equipment.

Ignition's built-in API handled project and tag import/export, Gateway logs, and timer diagnostics. Our 8.3 API enhancement supplied two additional actions used here: runtime-profile-v2 and tag-read-v1.

Those two actions ran through our custom llmImport Web Dev endpoint. They confirmed the scripting runtime and read live tag values, timestamps, and quality. The table separates the two providers.

We took browser screenshots and corrected overlaps, unwanted scrollbars, and hard-to-read labels. We also fixed the Power layer, which initially repeated temperature readings. That was the kind of mistake a polished screenshot alone could hide.

The final review covered 42 layout states and 41 browser behavior checks. We exercised navigation, workload changes, pump recovery, alarms, utility loss, and missing telemetry, then checked tag readbacks and Gateway logs. For example, starting the standby pump had to restore simulated flow.

This build was tested on Ignition 8.3.9. Adapting it to another 8.3 release or to 8.1 requires version-appropriate scripts, API access, and testing. This case study does not establish compatibility with every Ignition patch or a production control system.

The most useful input was a clear description of how the demo should respond. It gave us something concrete to test after the pages appeared.

I maintain the Ignition AI Skills resources used in this build. The Ignition 8.3 collection includes the Perspective Builder, Jython 2.7.4 Script Builder, and API enhancements. Use the package that matches your Gateway version, and verify each change in a development environment.