As some of you may be aware, Michael Lewis, CEO of Mycroft AI Inc, issued an email to the backers of the Mycroft home assistant device mid-February of this year:
tl;dr: Mycroft was on track for a successful launch of a privacy-respecting home assistant device but ultimately failed, despite delivering a limited number of devices to the earliest backers. This was due to a variety of issues, including some technical issues requiring hardware redesign, insufficient foresight of high injection molding costs, the global pandemic, etc… but most of all, they were caught in litigation by patent trolls.
So, what now?
A product like this may be in line with the ethos of Purism’s mission. After all, Mycroft was very transparent about their unwavering vision of privacy-respecting software. (I know, I know, the existing design leverages a Raspberry Pi 4 and that’s not FOSH, but that goes down the rabbit hole of a different conversation altogether.) Although Mycroft’s CEO has clearly indicated that the company is ceasing most operations, he did state in his email:
“For all these reasons, I am continuing to explore ways to continue Mycroft’s mission.”
I’m curious what the possibilities could be if Purism reaches out to Michael Lewis regarding a partnership with Mycroft and development of a “Librem Home” product. We could collectively create “skills” (small programs that tell a home assistant how to interpret/process certain language cues/questions/etc) in the existing Mycroft repositories which would add value to current Librem One capabilities. The R&D is already largely completed but there would need to be substantial investment with which to manufacture inventory.
I actually have a decent amount of experience in the stack they were using and I think we could pretty easily get it working on the phone itself. I noticed the keyboard doesn’t have speech recognition. Happy to help with that effort if anyone is interested. That is what I am doing professionally now(working on speech tools).
It was sad news, but I was also one of their early backers. I was pretty happy to have gotten one of those first devices. It is pretty cool.
One could run the full solution on the phone, or have a backend hook-up, especially nice if installable via .deb or flatpak. Would be handy to use via app interface, keyboard dictation, TTS reader (right click menu option, Firefox Reader View, Foliate eBook reader)
A key enabler for especially the former would be if there was a way in the PureOS store to move/install to microSD card, as 32 GB was way to small even in 2017, but especially now.
Nonprofit OpenVoiceOS team in partnership with Neon AI are continuing to work on two parallel, but complimentary forks of MyCroft. One of the key engineers from Mycroft is also now working again on Rhasspy, this time now in partnership with Nabu Casa (HomeAssistant), they plan on releasing a MyCroft II successor.
I haven’t even gotten OVOS working on my Mark II yet, only NeonOS. Haven’t had much time to fiddle with the Librem 5 yet either. Storage space is a constant battle.
The MyCroft Mark II uses a 4GB RBP4 with an ARM Cortex-A72 @1.8 GHz, while the Librem 5 has 3GB RAM using an ARM cortex-A53 @1.5 GHz, so ideally it should still work.
I’ve built a pi-croft and it does indeed work offline. I am curious how this phone CPU performs. I guess I will have to do some testing. Would be cool to have the feature on the keyboard.
JCS, thanks for reaching out to me. There may be some interest by Purism. Do you have Michale Lewis’s contact info? If so, you can send it to me via email at Rex.Lee@Puri.sm
Thanks for add @francois-techene and myself to the conversation. I will be traveling to Africa this week on business so please keep Francois in the loop. Thanks Rex
My prior conversation with Todd indicated that this partnership is still of great interest to Purism, but this likely won’t be kicked off for at least a couple months. I don’t want to make any promises I can’t keep, but I’m looking forward to getting conversations rolling between Mycroft+Purism leadership to see the possibilities of getting a privacy-respecting home assistant to market. I suspect the conversation will primarily involve a mutual understanding of the current state of R&D, anticipated capital investment for inventory, expectations of the burden on the support team, etc.