Has anyone tried selling digital products using a custom operating system?

[Regis]

Hello. I've been working on my own microkernel for a while now about 2 years, actually and I've reached a point where the OS is stable enough to run some small tools I've developed for it. Are there any people here who’ve tried selling digital products directly related to their OSDev project? I’m not necessarily talking about selling the OS itself, but related stuff: drivers, ported libraries, homemade debugging tools, that sort of thing.

The main issue holding me back is distribution. On a custom OS, you don’t have Gumroad or an equivalent that installs properly you have to handle everything yourself when it comes to payment and delivery. Has anyone else faced this problem before?

[dustinmacdonald]

I think you'd have the most luck selling educational products to teach others how to do what you've done, but so much of that sort of thing is available free and open source that it would be a tough sell. You do see YouTubers monetizing video collections of how they wrote an OS, and similar.

Who do you see as the consumer, purchasing a driver (for example), for an OS that only you use? Understanding the value proposition would be critical before you began thinking about that.

[savoita]

If you want to monetize anything about any kind of software product, you have to understand what problem you are solving, and then sell your thing as the solution (actually that's.. like any product, I guess). For example, one of the BSD OSes (can't remember which one) is used by Netflix specifically because of its (maybe at the time, not sure about now) superior network stack to Linux. It being free probably helped, but they do pay people to work on the drivers they require for the OS to function as they require.

If your OS is FOSS, then you'd probably be looking at SLAs (Support License Agreements) or just sponsorships (very hard to get).

But, down to brass tacks, you have to do something better than the current offerings. And you likely need to pre-include all or almost all of the nuts and bolts.

[bowdrill]

But, down to brass tacks, you have to do something better than the current offerings. And you likely need to pre-include all or almost all of the nuts and bolts.

The cost for this increases faster than inflation over time, hence the few power players. To the OP, I'm working on system programs now and you might find some interest in a thread I'll start in this board soon, since it focuses on those who don't have the resources to compete with incumbent operating systems.

A small player could write UEFI or otherwise bootable applications that serve as potboilers for larger projects.

Speaking for myself:

• I'm willing to shut down some systems to "swap cartridges."
• I'm open minded about cost/benefit analyses. For instance, I don't need or expect a huge driver selection. But if someone gives me a system that can't talk to printers, what do I get for that?

[AndrewAPrice]

I could see this working for like a kiosk system. But, if you ever disappear or discontinue support, whatever consultant they bring in might have a hard time maintaining it.

[JackScott]

Back in the depths of time, one of the forum members here, 01000101, had some business success writing a deep packet analysis toolkit based on a custom kernel: https://wiki.osdev.org/User:01000101. The website for it has since disappeared so it was definitely 'some' success rather than 'lots' of success.

[IanSeyler]

I am working on this. It's based around my exokernel being used to run small sandboxed instances, either on-prem, or my cloud-hosted solution. It's currently in "alpha" at https://baremetal.returninfinity.com

For a small example I just published a Discord bot written in C that is able to run in a 4MiB microVM - https://github.com/IanSeyler/BareMetal-Discord-Bot