Welcome to our Discover Roundup, where we highlight communities doing creative and inspiring things with Discourse.
This month, we’re focusing on a particular kind of community: knowledge bases built on ongoing conversation. These communities prove that when you give people the right structure and tools, shared knowledge compounds.
This is a topic that’s particularly close to my heart. I’m a long-time community manager specializing in technical support forums for Autodesk CAD software, with nearly 30 years of experience serving the French-speaking community. Over the years, I’ve created and managed multiple forums, using platforms ranging from the oldest to the most modern—my latest site runs on Discourse. I won’t go into the details of this site, which combines a forum and a blog in a fairly classic way. Instead, I want to focus on a very specific point: the role of artificial intelligence in assisting software users, particularly in the Autodesk CAD field (as an example, but can be applied to any software).
I’m not referring to AI as it’s commonly used internally—like on platforms such as Meta—to streamline forum management or enhance user experience. My focus is on a practical application I developed using Google AI Studio and Gemini: a troubleshooting assistant for AutoCAD users, accessible via a simple web page.
This leads me to a pressing question that keeps me up at night: Could tools like this—and others tailored for specific needs—spell the end for traditional forums like Discourse? The initial trials of my assistant have been (very) successful, and I’m not just talking about the public AI tools that have gained popularity over the past few years. I’m referring to custom-built solutions like the one I’ve created. Might users of technical support forums, who have long relied on platforms like Discourse, increasingly turn to these specialized AI assistants instead?
The question remains unanswered for now, but it raises a critical issue: Could AI eventually make the community-driven support spaces I’ve helped build for decades obsolete?
I don’t think it will be one or the other. There are benefits to both and I’m optimistic that companies will continue to see the benefits of community-driven support forums even as they bring on AI agents to support their users.
Of course they were. You are trying to solve today’s problems with a model trained on previous knowledge. Which is a whole new dimension and so it can be a real productivity boost if done well.
The problem is that the model provides you exactly the knowledge (and nothing else) of what was written on the web, including the forums – and especially the forums – until TODAY. Whatever we learn tomorrow and whichever products we design and consume in the future – there will be no knowledge about them without forums. There will be no millions of Q&A topics about it. And the AI will have nothing to base the results on.
Thank you for your comments. Knowledge must reside somewhere, but what is available on forums and websites in general is not sufficient. Artificial intelligence is not just a search engine; that is what distinguishes it from search engines. In this case, the assistant I was talking about is an expert system that is certainly based on existing knowledge, but also on its own methodology that neither search engines nor forums possess. It is this difference that led me to hypothesize that perhaps people seeking to solve IT problems, and in this case software problems, might not be satisfied with knowledge bases such as forums and might turn to something that combines these knowledge bases with an expert system methodology.
Still the whole LLMs are based purely on pre-existing information and their magical way on how to access it. So e.g. you create (with the help of AI) a completely revolutionary product. Let’s say a new type of phone – so much different as iPhone was to everything before.
How will your assistant be of any help? There will be no experience with the product, and no one will ever write anything down in the forums. Not just about your product – about the whole category, that never existed before. The LLM would not know anything about it whatsoever.
Or let’s say Elon is successful and people get used living on Mars. And it would run Discourse-less (can you imagine 
?). Where will LLM get any answers about anything practical on Mars?
And there you go. It’s a combination. Without the knowledge base (forum) you just have an expert system but without the content.
So let’s say ChatGPT is able to help me with Discourse today. But how come? Well, because of Meta. Without it, LLM would just hallucinate.
Certainly, my initial hypothesis in my first message was not that Discourse forums would disappear. Certainly not. It was that users looking for a solution to a software problem would turn instead to artificial intelligence systems, which they are already doing, and in the future even more so to expert systems, which are not yet fully developed, which use data found everywhere on the web, including Discourse forums. I was expressing a specific concern, but I think that ultimately you share my view that artificial intelligence and, in this case, the use of expert systems will lead to the impoverishment of forums and websites. I think we can agree on that?