Is the "Studio Assistant" Feature as Significant as I Think It Is?

[Craig Anderton]

Disclaimer: I don't know a thing about software and coding. But I am inherently curious, and I wondered if Studio Assistant is an example of a small, trained LLM. Well, I've been digging into how this works, and it's very interesting. I think Studio Pro is basically trying to end Help files, and looking into an assistance mechanism that may not be future-proof, but sure seems to be future-friendly.

Studio Assistant acts like you're interacting with ChatGPT or its ilk, but it's based on a generic, off-the-shelf LLM from Mendix. It's not trained on Studio One, it's just a generic LLM. So how does it "learn" Studio Pro? The answer is...it doesn't. Mendix is embedded within Studio Pro, and basically wraps the off-the-shelf LLM with Mendix-specific guardrails and tools. So, the LLM isn't trained on Studio Pro, or becomes an "expert" on how Studio Pro works. Instead, it's an agent that does information retrieval. Because it's inside Studio Pro, it knows what document you're looking at, what you're trying to do, and the context of what you want to know. It can retrieve documentation, examples, metadata - basically, everything that exists in the Studio One environment. This is why it doesn't have to "learn" anything, it just finds relevant stuff that already exists.

This seems like a really smart approach. It's the same type of architecture used by Microsoft Copilot, Google Gemini Agents, OpenAI GPTs, etc., applied specifically to our world. Mendix doesn't have to be retrained when the next update or version of Studio Pro comes out, nor does it need a huge proprietary dataset. There are very low hallucination risks, and no legal/IP complications because it's interacting in a strictly regulated environment with guardrails. Software geeks who know more about this stuff than I ever will can check out this link about Mendix AI Assistance. And please, correct me if I'm wrong and missing or misinterpreting something! Although I'm concerned that this will become a "Studio Pro subscribers get X amount of free tokens and then you have to pay," to be fair, not everything can be free.

As Spock used to say, "Fascinating, captain." Or at least I think so

 

[SwitchBack]

In a time where no-one read manuals anymore it makes sense to give them to a tool that reads them (and whatever else you feed it with) and then sits there to answer your questions. To stay with Star Trek: "Computer, what is the closest distance to a restroom?"

 

[BobF]

I like better than help/manual b/c I can ask how to questions instead of keyword searches.

 

[TimC340]

Going into the Mendix documentation, Studio Pro has a different meaning in the Mendix environment; "Mendix Studio Pro is a tool for creating, viewing, and editing your Mendix applications". This could get very confusing - the only saving grace is that their Studio Pro is v11, whereas our Studio Pro is v8!! https://docs.mendix.com/refguide/studio-pro-overview/

As far as I can discover, Mendix Studio Pro has nothing whatsoever to do with Fender Studio Pro or its AI Assistant. Mendix's 'Maia' LLM is intended to help companies develop Java coding for enterprise software architecture.

 

[SwitchBack]

Would be even better if the Assistant could "Show me how to ..." rather than "Tell me how to ...", using animated gifs like some of the regulars post here. AI could generate tons of those to preload into the (offline) Assistant, right?

 

[Craig Anderton]

As far as I can discover, Mendix Studio Pro has nothing whatsoever to do with Fender Studio Pro or its AI Assistant.

As I said, And please, correct me if I'm wrong and missing or misinterpreting something! I guess this is just further proof I should never trust what AI searches unearth .

So now I wonder what Studio Pro is using? Or is it time to be mega-embarrassed and ask for the thread to be deleted?

 

[Gary Shepherd]

Isn't this something that is not available to licence holders, just the Pro + crowd?

 

[tpittman]

Studio Assistant acts like you're interacting with ChatGPT or its ilk, but it's based on a generic, off-the-shelf LLM from Mendix. It's not trained on Studio One, it's just a generic LLM. So how does it "learn" Studio Pro? The answer is...it doesn't. Mendix is embedded within Studio Pro, and basically wraps the off-the-shelf LLM with Mendix-specific guardrails and tools. So, the LLM isn't trained on Studio Pro, or becomes an "expert" on how Studio Pro works. Instead, it's an agent that does information retrieval. Because it's inside Studio Pro, it knows what document you're looking at, what you're trying to do, and the context of what you want to know. It can retrieve documentation, examples, metadata - basically, everything that exists in the Studio One environment. This is why it doesn't have to "learn" anything, it just finds relevant stuff that already exists.

As Spock used to say, "Fascinating, captain." Or at least I think so

I love all the great liners Spock had during Star Trek!! One of my Favorites is (as he was dying) is: Captain: The Needs Of The
Many Outweigh The Needs Of The Few. It's just an easy and simple statement but yet reflexes in my mind how things should be in general. I know kidda got off base but I loved Spock / Leonard Nimoy and all the Fascinating things Spock had to say. Very relevant at any point in time. Wonder what Spock would think of Fender Pro? Wait for it I got it "Fascinating" HaHaHaHa lol
Live long and prosper my friends As you can see Im very easily amused lol

 

[Crossinger]

My assumption: they have trained a RAG (Retrieval Augmented Generation). They feed their documentation in small chunks into a database and give each chunk an ID (well, it’s more like that: the ID might consist of 768 or 1024 numbers, it’s called a “vector”). When you ask a question, then the best matches from the database are taken and silently put along your question. All the AI needs to do is to form an eloquent response based on your question and the found chunks.

I have written such a RAG system myself for internal support purposes in my company. So it’s a quite common approach.

 

[FinalGod]

It's quite nice, but so far I don't see anything particularly spectacular about it.

If it mainly answers questions based on the Studio Pro documentation, I could do essentially the same thing by giving an AI the manual and asking it questions.

It would become much more interesting to me if the Assistant could actually interact with the DAW. For example, I could ask it to create an example track or bus with a suggested insert chain, configure the plugins, and let me A/B its suggestion against my existing setup. If I like it, I can keep the track or copy the inserts over.

Of course, for that it would need to know which plugins are available and how to control them. It could ask which suitable compressor I want to try, for example. Or Fender could provide an MCP interface, which would open up this kind of interaction to external AI agents as well.

It would get even more interesting if the AI could analyze the actual music and make suggestions based on what it hears, rather than only retrieving information from the documentation.

So maybe this is the beginning of something much more interesting, but in its current form I find it a little underwhelming.

 

[Jonny Lipsham]

Isn't this something that is not available to licence holders, just the Pro + crowd?

Currently, that is the case. But that may change in the future. I don't actually know whether it will or not. We'll just have to see.

 

[Mako]

I have not heard about this... oh I see its on the subscription.

You can do a lot with computer use now with Claude desktop and I assume with OpenAI. With computer use the LLM drives Studio Pro directly and will do what you want, you can watch it if you want to learn.

For me I would prefer to see Studio Pro expose an MCP server then users could drive it with their LLM of choice (bring your own tokens). Fender or others could come up with skills files to help drive some of the specific info you might want. I can tell you Claude is pretty decent with Studio Pro right out of the box with no skills added.

I am a coder and use MCPs all the time so that factors in, but they are not hard to use.

 

[TimC340]

As I said, And please, correct me if I'm wrong and missing or misinterpreting something! I guess this is just further proof I should never trust what AI searches unearth .

So now I wonder what Studio Pro is using? Or is it time to be mega-embarrassed and ask for the thread to be deleted?

This is what I've found out so far: instead of building a bespoke model from scratch, Fender uses an 'AI Agent' architecture. This framework takes a commercial natural language model (probably Chat GPT or Claude) and routes user queries through a Retrieval-Augmented Generation (RAG) pipeline (intended to prevent AI 'hallucinations'). Rather than relying on a general knowledge base, Fender's system acts as a curated tool by strictly limiting it to a specific dataset including the official Studio Pro manual, the software’s internal support documentation and a curated database of official tutorials and troubleshooting knowledge bases - probably not including a certain Big Book (unless you're getting royalties, Craig!).

 

[Craig Anderton]

This framework takes a commercial natural language model (probably Chat GPT or Claude) and routes user queries through a Retrieval-Augmented Generation (RAG) pipeline

Cool, thanks for doing the extra research. Actually, I think that pretty much describes what Mendix does, so people can check out the link in the original post to get more of a feel for how this kind of process works.