Just make sure you know best practices with scripting before using one of those. I checked that out a little bit and have had it write scripts that do extremely inefficient things (like doing single tag writes inside a loop). On a big enough SCADA that could cause chaos.
That's Gold right there!
In the end, I believe, at least for the next few years, it will be developers who know the Fundamentals, and use AI, will replace the ones not using AI, provided that AI is being used in that Industry.
Eh, I'd use that.
I do my coding in vscode with some help from copilot. If you point it at an intructions.md that says to use jython 2.7, you get pretty good results.
But... you then have to copy/paste everything to your actual ignition lib. That's where things get clunky.
I have plans for making some sort of script that will take a local folder and a container + project name, and will restructure and copy the local files to match the format used by ignition... But I haven't had time for that yet.
I ended up building a workflow around this exact problem.
The main issue for me was the same one mentioned above: generic AI agents can be useful, but they tend to write Python 3, invent fake system.* functions, or miss Ignition-specific context.
What worked better was giving the agent two things:
A dev Gateway API it can use to inspect, dry-run, validate, read logs, and check results.
Skill files for the exact task it is working on, like Jython scripts or Perspective pages.
I personally like using Hermes Agent, but the same idea can work with Codex, Claude, OpenAI, etc. The agent itself matters less than the loop: give it the right Ignition context, let it inspect a dev Gateway, make small changes, validate routes/views/scripts, and collect evidence before trusting the output.
With the Jython and Perspective skills I built through a lot of testing, it has gotten me through a large chunk of normal use cases: Jython scripts, Perspective pages, bindings, components, reusable views, Flex Repeaters, and route validation.
I am also working on Ignition 8.3-specific skills now.
If anyone wants to look it up, search âIgnition AI Skills.â For a demo, search YouTube for âBuilding Perspective Pages in Ignition 8.1 with AI Agents.â It shows the broader workflow with an AI agent using an API against a Gateway. The API for 8.1 is also available in the website.
This is incredibly interesting, Kevin. Iâve started going through the Iggy repository, and it has honestly blown my mindânot only because of what it already does, but because of where the same approach could go next.
Iâm working on something closely related: a specialized harness and IDE-like environment for creating and modifying Ignition Perspective projects with AI. Seeing Iggy already operating inside the Ignition ecosystem makes me think the next step could be bringing that intelligence directly into the development workflow.
For example, could an assistant like Iggy eventually be integrated into the Designer, or connected to a dedicated IDE, where it understands the project structure and helps developers create views, configure bindings, work with tags and scripts, preview changes and validate the result?
I already have an early prototype of this approach partially working, and Iâd be very happy to show you what Iâve built if youâre interested. Iâd also really like to dig deeper into the technical decisions behind Iggy, exchange ideas, and contribute or help test anything around this direction.
Iggy was an exchange resource built by sales engineering.
^ This is no longer representative of our AI position.
I can't really say anything more. Come to ICC or watch the keynote/talks after 
So, in roughly a year, weâve gone from âwe canât realistically do anythingâ to âI canât really say anything more.â
Blink twice if IgnitionA is working on AI inside the Designer, or on some kind of development assistant
And if you actually like my prototype, you could invite me to icc⌠or at least sneak me an early preview of whatever has suddenly made IAâs AI position so difficult to talk about. Deal?
A year is an eternity in the AI world 
I wrote that in July 2025. Anybody paying attention knows there was a major step change in ~November 2025 when Opus 4.5 was released, and the models, from all providers, have only gotten more capable since then.
Fair pointâthe models changed fast. But the change from âwe canâtâ to âI canât sayâ is what caught my attention .
So⌠AI inside the Designer or a Perspective copilot? Blink twice 
Also if opus was a change what about fable and gpt sol cause thats another step.
This makes me excited and yet a bit disappointed as my side project may not see the light of day if inductive releases a native designer ai. Still excited though for ICC.
Im exicited for this, im using skills that I built and work pretty great. I can imagine what you guys release will be much better.