IgnitionClaw

Since 2022 Large Language Models have found their way into most aspects of our daily lives. Corso Systems is no exception. From a tongue in cheek post about using ChatGPT as the New Ignition Intern to speeding up Ignition development and beyond, we have found a lot of use cases for LLMs.

To make LLMs even more useful you need to use AI Agents. Agents give you the tools you need to build and manage complex workflows to interact with an LLM, and help you move beyond a 1:1 relationship asking chat questions into a workflow where you can use LLMs to help improve overall business effectiveness.

Agents also give you a different way to interact with your LLM of choice. Rather than simply using a prompt interface you can integrate with the agent using API integrations, and tools like Slack, Teams, and Whatsapp. This opens up a lot of possibilities for integrating LLMs into your operational workflows in new and interesting ways.

Using some of the open source agents available today we have found some really amazing enhancements combining the power of LLMs with Ignition and manufacturing operations, as well as some limitations we decided to tackle ourselves!

n8n, OpenClaw, and Hermes Agent

Popular open source options we have worked with include n8n, OpenClaw, and Hermes Agent. n8n is great for someone new to agents as it provides all of the building blocks you need to integrate complex systems with tools like Ignition. We have found it does require a fair amount of setup and configuration compared to some other agents, although once you are up and running it gives you everything you need.

In late 2025 OpenClaw took the world by storm. It scratched an itch giving you an agent with some personality, easy integrations to just about any technology you might want to use with AI, and opened the world’s eyes to the power of AI agents. One of the interesting and also scary things about OpenClaw is you can set it up to do pretty much anything you desire. This is a double-edged sword. People have used it as a personal assistant, a stock trading bot, and a coding tool. Some folks have also found it might decide to wipe out your filesystem if you don’t have proper security precautions in place.

While security precautions are a concern with all AI tools, OpenClaw has a stigma of being somewhat “dangerous” although out of the box it is a relatively safe offering.

In early 2026 another tool called Hermes Agent was released. Hermes Agent gives you all of the capabilities of OpenClaw, including a “self-learning” algorithm designed to simplify complex workflows into skills it will automatically build as you use it. This streamlines the overall LLM experience by taking multiple prompts you might need to run to accomplish a task and building a skill to handle them automatically.

With our overall focus on security and eliminating risk this feature has become a powerful tool when working with locally hosted LLMs. While local models are not at the forefront of what is possible with AI, you can run them on your own hardware without exposing your systems directly to the internet.

Because local models aren’t “frontier” AI tools, we have found leveraging agents fills a lot of the gaps, and makes it possible to combine local LLMs with Ignition to provide some really powerful benefits.

Iggy AI Demo at ICC 2025

One of the limitations we have found with LLMs in the manufacturing world is they aren’t necessarily designed to be more than a chat interface. You prompt the LLM, it works on the problem and returns the results. Depending on the LLM this could be used to generate presentations, images, or simply act as a replacement for a search engine.

At ICC 2025 Inductive Automation released their Iggy AI Demo along with some interesting use cases. This included generating Perspective Views based on screenshots, building Perspective Forms from a prompt, and generating artifacts like alarm status screens and trends from prompts, displaying them directly in the chat interface.

This moved well beyond “simply a chat interface” and opened up the possibilities of building something much more interesting for Ignition.

This project was built using the WebDev module along with n8n as the backend agent. It was hooked up to the Gemini LLM from Google, which requires connecting your Ignition system to the internet to get everything to work.

Enter IgnitionClaw

Taking inspiration from the Iggy demo, and working with Hermes Agent, we found ourselves going down a rabbit hole of Ignition and LLM integration. Using Local LLM models to provide a solid foundation for actual project deployments we ended up doing a fair amount of integration work to get everything functioning.

After looking at what it would take to get everything we wanted put into place using Hermes Agent a quick calculation led to the decision we could put a similar amount of effort into building our own AI Agent, directly in Ignition as a Module.

Taking advantage of the brand awareness of OpenClaw, we decided IgnitionClaw would be a fun name for our agent and we were off to the races!

We have been hard at work for the past few months building out IgnitionClaw and are excited to open it up for testing!

IgnitionClaw began as simply a chat interface, building in some of the artifact generation tools from the Iggy Demo, and has grown in scope since then. First we expanded on our AI for Heat Treating examples, and have found it makes a great interface for people who don’t necessarily need to look at operator screens but want to get access to information about their processes.

We also built out a skills framework so IgnitionClaw can take advantage of one of the most powerful features of AI agents, simplifying the overall workflow to answer complex questions.

We built in full chat history out of the box similar to any LLM interface you might be familiar with, and have it working really well with local LLM models.

What’s Next?

We’re shooting for a full release in September 2026, and will continue to upgrade IgnitionClaw’s capabilities well into the future. We’re currently figuring out how to build in some of the self learning capabilities from Hermes Agent to make IgnitionClaw a top tier tool in the manufacturing world!

If you are interested in IgnitionClaw and how it might be able to help improve your manufacturing operations please reach out using the contact button below, we’d love to hear from you!