Thank you for publicly sharing and documenting so much information about this epic quest you're on (I'm sure it's a solid PR move, too :).
I wish you all the best and respect your decision to eschew a comparatively cushy and predictable job to take up this risky and challenging endeavor filled with never ending thankless tasks.
Godspeed.
Edit: Sorry to hear it isn't a big boon on the marketing front. All I can say is I recall seeing TinyPilot at least once before, but now it's much more on my radar. If it could do remote power switching I'd be very keen to keep it in my server rack as the sole point of remote administration (sorry, I'm sure you get dumb feature requests and this is probably just another one.. but I couldn't resist).
Remote power switching is what has held me back. Even if it was an additional cost / addon I'd definitely look at picking up a dozen or so. Still I've been following for a while and hope to see some cool improvements over time.
EDIT: Actually being a raspberry pi, It might not be hard to build in that feature. I might just grab one today and see.
Yeah, remote power switching is definitely doable. It would either be wiring directly into the motherboard's or controlling an external power switch. The challenge is making it user-friendly, supporting it, meeting compliance requirements, etc.
I think it's doable from the pi headers inside pretty easy, fortunately these are all homebrew machines i'd use on so compliance issues aren't as big a deal. For work though, i'm right there with you.
Did you move away from Raspberry now or what does the EE guys do?
I don’t have personal experience of outsourcing assembly yet but make sure quality control is flawless before letting that go. I would and will personally fly down to China to set that up.
Will be following to see how that process goes.
Nope, we're still on Raspberry Pi. It's difficult to move to other hardware because it would mean managing a lot more of the OS stack ourselves, and then we'd still have thousands of users still on Raspberry Pi.
I'd like to move to the Pi CM4 because we'd reduce costs and get a lot more control over the hardware, but it's a big up front cost to start over from scratch on a new board.
1. If you're going to have a headshot, get yourself a professional headshot on the About page. You squinting into the sun in a selfie is nice for a Twitter profile, but you're trying to present a real company w/ a real product here, right?
2. It's clear this story is really tied into your identity right now, but nobody cares about this part on a page about TinyPilot:
>> I'm a software developer and worked most of my career at Microsoft and Google, but I quit in 2018 to found a company of my own.
I like the focus on your frustrations and how you solved that. The quitting part people could take negatively ("is this guy a quitter? will I receive product support if I buy this?").
Everything else looks pretty solid for a SaaS homepage.
Thanks for reading and for the constructive feedback!
Yeah, the About page is something I haven't touched much since I first launched the site, and I agree that it feels not quite consistent with the tone of the site or where the company is at this point.
Thanks for sharing your experience Michael! I found out about TinyPilot a while back ago. I made a similar device for a completely different industry and learned that TinyPilot has nearly the same architecture but a completely different use case. Unfortunately, we weren't able to turn a profit with that device either. It seems streaming video is a tough market to crack :(
Our manufacturer in China is a startup as well if you are looking for a connection. He works hard and can give you a good price at smaller scales. I'll send you an email as well but in case anyone else is interested, kevmo314@gmail.com.
Michael, thank you for sharing your story and experience for TinyPlot from inception starting to scale. As a hardware engineer myself and having been in your shoes, founded 10 years ago a startup and launched the product I know what it takes and how hard it is to reach the point you got to, kudos.
This blog mentions that you consider manufacturing in China and drop ship the product. In all likelihood by now you might get PCB/PCBA and/or other parts of it from China, not sure. Outsourcing the whole chain flow is another dimension.
Some post here mentioned it already, but the support/expense you will need to provide that is going to be a longer list. Ongoing quality handling remotely, component shortage, part replacements you did not know about, yield, shipping issue, etc it can be real distractions. That is one of the reasons you need some build/ship volume, probably minimum over several hundred units a month to make the whole flow work efficiently.
One more thought. I am not sure how you will avoid an actual office for a hardware company assuming you want to continue to grow with product improvements, debug and next generation development.
My post is just to give you heads up shifting from in-house build with tight control to outsourcing might not solve all your problems. At least that was my experience and I for one definitely did that outsourcing too early.
Amazing post Michael! I have been following your story for years.
I have also a similar project backing, but not as technical as yours. I use also Raspberries for Smart Kiosks that are set to be in restaurants and increase the demand of the most convenient dishes at a time with great photography.
I see that you think now of Debian packages as an optimal way for installations in your project.
What would you say it is the best way yo keep a Raspberry system "auto-updated"?
PD: My skills are way lower than yours. I use mostly Ruby scripts (SystemD) to contact a backend in Rails.
>What would you say it is the best way yo keep a Raspberry system "auto-updated"?
I think Debian packages are better than what I was doing, but if I were starting from scratch, I'd try to use Yocto[0] or NixOS[1].
Take this with a grain of salt, because this is secondhand from another founder who had good experience with Yocto, but from what she told me it's optimized for the case of pushing out updates to embedded devices. One of the pitfalls of Raspberry Pis is that the microSDs are vulnerable to filesystem corruption, which can leave the device unbootable. I believe Yocto protects against that where there are always two bootable partitions, so you failover to the other partition and can recover.
Yeah, I think taking myself out of the critical path and making the business more location-independent will make it more attractive to a buyer if I decide to sell.
How have you thought about pricing? I would have thought a small increase in price would make a big difference to your bottom line given the difference between your revenues and profits.
Congrats on your amazing success! 0 to 850k revenue with a hardware product in three years is almost unbelievably amazing. I'm totally ignorant about hardware dev. How does one even go about designing and manufacturing a hardware product like TinyPilot? Do you design in some sort of CAD software and then partner with some sort of factory?
Have you considered retail Or other distribution channels?
I've reached out to a couple of resellers, but they weren't interested, though I didn't pursue it very heavily. With supply shortages, I can only produce ~200 devices/month, and I can sell that volume on my own, so the resellers wouldn't help much. Maybe next year when the supply is less constrained.
Do you have a SO, if so what do they think about this experiment
Yep, she helped edit this post! We met when I was a year into bootstrapping, so she kind of knew what she was getting into, and she's been supportive. She worked on the business for the first year when we were quarantined for COVID and doing everything from our house.
Happy to take any feedback or answer questions about this post.