Native Instruments to be acquired by inMusic

[Gamma-UT]

Does every f*cking thing have to turn into an argument over AI now? Can you move this sh*t to the AI forum?

There’s probably more mileage in a jamcat forum where jamcat can pretend to be the font of knowledge in everything.

[FigBug]

You're just being silly. There is a huge gap between "the AI revolution is it renders technical debt irrelevant." and "they’re dangerous and don’t work!"

LLMs work great for some stuff, ok for other stuff and not at all in other areas. They aren't magic, you can't just point Claude Code at a 20 year old code base and say "upgrade this plugin" and have a working plugin for all modern operating systems pop out the other end. It's still a lot of work.

Modernizing the UI is the area a LLM will be least useful, and the one users want the most.

[pough]

inMusic is the company that rubbed its taint all over BFD3. It's good news for afficionados of that stank, I guess. I personally will be perusing at least some of the inevitable "alternatives to..." threads.

What ever happened to 2getheraudio?

Some day some enterprising AI agent will do a proper survey and examination of all the acquisitions of plugin companies over the years and the resulting effects. I wonder what it will hallucinate.

[Amberience]

Keep telling yourself that.

As someone who uses agentic code tools every day to help me with audio plugin development, aye. Sure. I will keep telling myself that.

Oh, I see. You use tools you don’t believe in to complete your work. Got it.

Listen up everyone, Amberience is warning you to stay far away from the tools he takes advantage of, because they’re dangerous and don’t work!

I never said that, you total weapon.

[jamcat]

You're just being silly. There is a huge gap between "the AI revolution is it renders technical debt irrelevant." and "they’re dangerous and don’t work!"

LLMs work great for some stuff, ok for other stuff and not at all in other areas. They aren't magic, you can't just point Claude Code at a 20 year old code base and say "upgrade this plugin" and have a working plugin for all modern operating systems pop out the other end. It's still a lot of work.

Modernizing the UI is the area a LLM will be least useful, and the one users want the most.

Of course you have to be very hands-on when working with AI. A software development background is assumed here.

What you absolutely can do is point Claude Code at a 20 year old code base and have it tell you what the code does, how it works, what each class function does, who inherits what, and how it all fits together.

Technical debt is when code is so old that no one remembers what it does, or has become so complex that no one knows quite how it works anymore. Old deprecated instructions may be deeply embedded And need to be untangled. Old defunct architectures may need to be ported to shiny new ones.

But what about that can’t AI do quite easily? It’s not going to be as simple as a one-prompt pass, but it’s quite manageable and will be infinitely less time consuming and painful than if you tried to do it yourself by hand.

[TechHaus]

Jamcat take a timeout brutha.

[jamcat]

Keep telling yourself that.

As someone who uses agentic code tools every day to help me with audio plugin development, aye. Sure. I will keep telling myself that.

Oh, I see. You use tools you don’t believe in to complete your work. Got it.

Listen up everyone, Amberience is warning you to stay far away from the tools he takes advantage of, because they’re dangerous and don’t work!

I never said that, you total weapon.

You can’t have it both ways. But I understand your psychology here. You’re using AI to develop some plugins, so you want to downplay its abilities. It’s a cope.

Please do tell me, though, what is your programming background? Do you have a computer science degree? Have you ever worked for an audio software company?

[Amberience]

Keep telling yourself that.

As someone who uses agentic code tools every day to help me with audio plugin development, aye. Sure. I will keep telling myself that.

Oh, I see. You use tools you don’t believe in to complete your work. Got it.

Listen up everyone, Amberience is warning you to stay far away from the tools he takes advantage of, because they’re dangerous and don’t work!

I never said that, you total weapon.

You can’t have it both ways. But I understand your psychology here. You’re using AI to develop some plugins, so you want to downplay its abilities. It’s a cope.

Please do tell me, though, what is your programming background? Do you have a computer science degree? Have you ever worked for an audio software company?

Yes. I worked for FXpansion from 2008 until the ROLI acquisition in 2016, then worked for ROLI from 2016 to 2020, when inMusic acquired BFD and I followed with the software. I left inMusic 11 months ago to pursue freelance software development. I've been a product owner, a 3D designer, a software developer, an audio engineer, an editor, and a QA manager.

I have an arts degree. Programming was something I learned on the job.

[SLiC]

touché

[jamcat]

Ok, thanks. So what I don’t get then, is with your background and personal, hands-on experience with current AI coding agents, how you can justify this statement:

LLMs are good at pattern completion. They are not magical replacements for system understanding.

If anything AI risks creating even more debt because people can now generate thousands of lines of plausible looking code faster than ever, often without understanding the architecture underneath it.

This sounds like what people were saying a few years ago with fledgling LLM models, not what people with extensive hands-on experience with the current generation of AI coding agents are saying. That’s more ChatGPT 3.0 than Codex 5.5 talk.

When you talk about “without understanding the architecture underneath it,” are you talking about the AI agent or the person using it?

You seem to be addressing something I wasn’t even talking about. You seem to be going after a “vibe-coder” strawman, when I’m talking about a contemporary custodial development team leveraging AI to unpack, navigate, document, and manage large legacy code bases. You cannot possibly believe in all honesty that AI can’t be of massive assistance in that task.

So then what was the point of attacking what was an obvious, basic statement of fact?

[VOODOO U]

You want to talk about integration that makes most sense? I've said it before and I'll say it again: NI DAW.
They have Maschine now add audio tracks. Tracks can be MIDI or Audio. On the audio side the tracks have the option to run as Traktor's giving unlimited tracks for DJing. Integrated at the core of Maschine is a modular environment called Reaktor (think Grid for Bitwig, M4L for Live).
Sampling is of course the Kontakt environment.
Pro version comes with additional instruments (Massive etc), mastering plugs (Izotope) and The Mouth.
Full integration with Akai hardware and possibly Numark for various control options.
I don't want to hear about any bullshit politics that makes this "impossible". Make it happen or go home because A.I. is here and we can make what we want for our needs individually.
Thank you and may God bless.

[Amberience]

Ok, thanks. So what I don’t get then, is with your background and experience with current AI coding agents, how you can justify this statement:

LLMs are good at pattern completion. They are not magical replacements for system understanding.

If anything AI risks creating even more debt because people can now generate thousands of lines of plausible looking code faster than ever, often without understanding the architecture underneath it.

This sounds like what people were saying a few years ago with fledgling LLM models, not what people with extensive hands-on experience with the current generation of AI coding agents are saying. That’s more ChatGPT 3.0 than Codex 5.5 talk.

When you talk about “without understanding the architecture underneath it,” are you talking about the AI agent or the person using it?

Also, you’re addressing something I wasn’t even talking about. You seem to be going after a “vibe-coder” strawman, when I’m talking about a contemporary custodial development team leveraging AI to unpack, navigate, document, and update large legacy code bases. You cannot possibly believe in all honesty that AI can’t be of massive assistance in that task.

So then what was the point of attacking what was an obvious, basic statement of fact?

Your claim was that technical debt is a thing of the past. I disagreed. You got the hump. I don't quite know what else to tell you. I stand completely by what I said. Nothing to do with vibe-coding, and everything to do with just having the experience and the knowledge and the foresight, to make good architectural decisions. An LLM simply does not replace that.

Of course I think AI is useful, which is why I use it to augment my development approach. I never said otherwise. But I don't trust it blindly. That isn't a position I've come to through prejudice or ignorance. It is one I've come to through trial and error, and experimentation.

If you use AI agentic workflows poorly, and don't have a solid understand of the system you're building, then you risk increasing your technical debt. 100% true statement.

[VitaminD]

I also wonder if we'll end up with a replacement for NI Access. Perhaps even an all-in-one across all the InMusic brands? Buy an MPC get NI and brainworx products bundled. Buy an NI Komplete and get a voucher to buy an MPC expansion.. or something along those lines.

It seems odd (to me) to have one company own so many large pro audio brands.

1. AKAI Professional
2. Moog Music
3. Rane
4. Alesis
5. Numark
6. Denon DJ
7. Headrush
8. M-Audio
9. Alto Professional
10. AIR Music Tech
11. BFD
12. ION
    
    now add
13. Native Instruments
14. Izotope
15. Brainworx
16. Plugin Alliance

Beyond MPC and Maschine there is also Numark and NI Traktor will compete with each other

[VOODOO U]

How about Big f**king Moog?

[jamcat]

Your claim was that technical debt is a thing of the past. I disagreed. You got the hump. I don't quite know what else to tell you. I stand completely by what I said. Nothing to do with vibe-coding, and everything to do with just having the experience and the knowledge and the foresight, to make good architectural decisions. An LLM simply does not replace that.

Of course I think AI is useful, which is why I use it to augment my development approach. I never said otherwise. But I don't trust it blindly. That isn't a position I've come to through prejudice or ignorance. It is one I've come to through trial and error, and experimentation.

If you use AI agentic workflows poorly, and don't have a solid understand of the system you're building, then you risk increasing your technical debt. 100% true statement.

I think we are approaching this from two very different sets of assumptions. Mine is that a competent person or team is directing it. Yours is than an incompetent person is blindly setting it loose.

My assertion is that AI renders the major hurdles of technical debt largely moot, particularly when a new unfamiliar team takes over. The major hurdle is reading and digesting someone else’s code, and becoming familiar with it enough to take ownership. It’s a gargantuan task that discourages many from even starting. And I know for a fact that AI assistance lightens that burden substantially. And I think you know it too.