As a potential customer, I feel bad you feel that way.
It may seem overpriced when compared to collecting all of the pieces and putting them together for yourself, but the person trying to do this as a business has to make a big investment in up front purchasing as well as labor to get them all together and shipped out. It quickly reaches a point where doing it for an “undercutting” level price doesn’t justify the risk and effort.
An arm soc is probably $100. So break even is maybe $200 per day getting the thing working. Thus should buy the thing, unless you particularly enjoy the construction process more than whatever else you could spend the weekend doing.
The posted profit over time figures suggests the above back of envelope is rather optimistic, so there's a non-zero risk that after the weekend the thing wouldn't work.
(and thus I've learned that the UK distributor is in Germany with possible brexit/VAT hazards, but hopefully it'll turn up at some point)
I understand the feedback about the price. When I created the original TinyPilot,[0] I wanted a low-priced alternative to the $600 KVM over IP options I was seeing, and now I sell my devices for $400-500/unit.
The thing that's tough about selling hardware is that it seems easy to produce a few units cheaply, but creating a sustainable business for it, the costs add up. You have to fulfill orders, provide support when there are delivery issues, provide support when the customer has trouble understanding how to connect to a device on their local network. And there's certification, stockpiling components so you don't run out of stock, managing inventory from many different vendors.
TinyPilot became profitable in 2023, but for the first 2.5 years, we were operating at a loss. [1]
Also, small correction: no KVM over IP vendor is really selling open hardware. The part that's open is the software. I haven't seen any vendors publish their hardware schematics. It's expensive to create PCBs, and it's not like open-source software where you can hope to get useful external contributions.
[0] https://tinypilotkvm.com/blog/build-a-kvm-over-ip-under-100
[1] https://mtlynch.io/retrospectives/2023/10/#tinypilothttpstin...
But of course github stars and positive reviews won't pay our bills. So it totally makes sense to charge what you are charging for your product. You deserve it, and you're clearly putting in hard and honest work. If and when I need a KVM I will 100% be reaching for a TinyPilot.
But I personally don't care about support, the enclosure can be 3d printed or some cheap plastic box, if sold as DIY kit it would save assembly costs and avoid need for certification and so on. I think there is a small market for homelabers like myself who would pay maybe $100-150 for a simpler product.
And congrats on being profitable.
We used to sell both the pre-made units and a $190 DIY kit. Surprisingly, when I tried removing the DIY kits, sales jumped by 50%. And it wasn't just higher revenue because people were buying the more expensive product - order volume increased too.[0]
My theory is that when we asked customers if they wanted to pay more for a pre-made version, some of them stopped to think about which version they wanted and then just never returned. So, when we simplified to just one product, it eliminated friction in the buying process.
With the Pi 4 shortage the past few years, using our limited Pi 4 supply for DIY kits instead of our pre-made product would be a terrible move. We're expecting the shortage to be over in the next few months, but even if there was a way for me to sell an extra 50 units per month at a $50 profit, selling a second product line and managing it end-to-end is so complicated that it wouldn't be worth the extra $2.5k/month. And that $2.5k/month is before chargebacks, hiring people to assemble and fulfill the kits, etc.
We do still work to serve DIY customers and make sure our software stays compatible with users who just use off-the-shelf parts. And we still sell a power splitter,[1] as it's the one thing you can't really buy off the shelf from other vendors. We're able to do that because they cost about $10/unit to make, so we can make a year's worth in a week and just ship them to our warehouse. But doing that with a whole kit would be way more costly.
[0] https://mtlynch.io/retrospectives/2021/12/#reducing-to-a-sin...
[1] https://tinypilotkvm.com/product/tinypilot-power-connector
Is it possible to order current revision of TinyPilot for someone in a remote location and connect to it on arrival through some managed service - without going through all the hoops of helping end user with setting up some form of reverse shell (e.g. either meshcentral or openvpn/wireguard for webinterface)?
Customers could do that on their end, but they'd have to buy the device from us, configure it for remote access,[0] and then forward the device to the final destination.
Technical details here: https://www.kickstarter.com/projects/mdevaev/pikvm-v4/posts/...
If you REALLY want to try it and you have PiKVM, you can contact us at Discord (https://discord.gg/bpmXfz5) and we will give you access to the beta version.
usb splitter $9 - https://www.amazon.com/dp/B08C5FWQND
usb power blocker $7 - https://www.amazon.com/dp/B094FYL9QT
hdmi-csi2 adapter $32 - https://www.amazon.com/dp/B09GY9M9BX
atx power control looks less off-the-shelf
They don't seem to directly support the upstream project, and I believe they may have even forked some of the software from PiKVM and TinyPilot both and may or may not have provided proper attribution/recognition... so take that as you may.
This project is a full-time job for an entire development and support team, and it's a way to keep the software open. For this money you get excellent hardware, software without any restrictions and a huge number of features. We also have a production-grade support: if you find a bug, it will be fixed quickly, and if the device fails due to a defect, we will quickly replace it.
In fact, we have developed and support all the key components of the KVM stack for Linux - the video server, kernel patches, and so on, which is also used by our competitors, like TinyPilot. But unlike them, we don't impose any licensing restrictions - for example, buying TinyPilot you find yourself bound by a subscription, and if you stop paying, your OS will no longer receive updates. With PiKVM, you don't need to do this - the device is yours forever, with updates until the end of time.
In addition, the production of large quantities of iron is quite expensive. Each device must be assembled, tested, packaged, and so on. When you do some small project, you can do it yourself, but in a large batch it inevitably has to be delegated to the factory, where each operation costs a certain amount of money.
In summary, that's where the price comes from: software, support, development, people.
This is true and is a fundamental difference between PiKVM and TinyPilot.
TinyPilot charges a yearly subscription for customer support and software updates because those things have ongoing costs.
I don't think small businesses can realistically promise customers free support and updates for life. It may work in the short term, but at a certain point, it's not sustainable for a vendor to spend their limited support and dev resources on customers who purchased hardware five years ago and will never pay the vendor another dime.
Another thing that made it possible to do this is the automation of all routine processes, such as testing and package builds. In PiKVM, the first component that was written was a build system to solve this issue once and for all. In this vein, I made a huge contribution to the "pre-launch" before flying. BTW, to be honest, I was somewhat surprised that you used Ansible instead of packages, it seems that this caused you a giant overhead for support.
Of course, in the future we plan to provide some paid services, but this does not concern the regular OS and access to updates at all, they should remain free.
Why didn't you tell me three years ago?!? : )
Yeah, Ansible was a huge mistake.[0] I knew it wasn't the right tool for distribution, but starting out, I just stuck with the tools I knew. We finally purged it in our last release and moved everything to standard Debian packages.
I'm glad to hear that PiKVM's stragegy has been working. I know our two projects are often pitted against each other, but I think there's ton of space for both of us to win back market share from the huge enterprise players and their consistently weak KVM offerings.
[0] https://mtlynch.io/solo-developer-year-5/#ansible-and-git-ar...
You didn't ask :P
> Yeah, Ansible was a huge mistake.
I have some experience operating large clusters and distributing software, and these damn things like Ansible have never led us to anything good. So either packages or the whole chroot (docker, whatever).
In fact, I'm a fan of the good old time-tested solutions. You know, Makefiles, packages, maximum integration with native OS tools. A lot can be done with simple tools, and although the start will be difficult, the support will be simple.
> there's ton of space for both of us to win back market share from the huge enterprise players
Ofc. In addition, users clearly benefit from competition.
In the past I've considered making and selling a cheaper version. Aside from lack of time (or motivation), one thing which discouraged me is that I would feel bad "undercutting" someone selling open hardware.