Mark II Update - June 2020

General Discussion
1

Originally published at: https://mycroft.ai/blog/mark-ii-update-june-2020/

Being part of the journey is one of the reasons that angel investors put money into startups and crowdfunding backers support projects. Sometimes that journey is simple and straightforward, but when it comes to developing new technologies it is often a rollercoaster.

Crowdfunding platforms are generally pretty clear about where they stand on this thesis. The early team at Kickstarter even published a page titled “Kickstarter Is Not a Store” where they explain to backers what projects are and are not.

On the spectrum of journeys from “Sean’s trip to the market” at one end and “Frodo’s trip to Mordor” on the other, Mycroft’s journey is definitely closer to Middle Earth than suburban London.

Over the past couple years we’ve tried to communicate this. We’ve had some false starts on the hardware, fundraising challenges and detours into enterprise services to make ends meet, but hopefully we’ve been able to communicate effectively along the way.

After all, backers bought a ticket on our rollercoaster, it wouldn’t be fair if we didn’t communicate that empty stomach feeling you get when things don’t go as planned.

New CEO, New Roadmap

In case you missed it, Mycroft hired a more experienced CEO early this year. Our new CEO shares our original vision, but has much more experience with delivering large complex software projects and complex hardware.

Under our original roadmap and release cycle Mycroft went to alpha on May 20, 2016 and beta on March 7, 2018, but it turns out our leadership team was a bit aggressive in defining what alpha and beta mean.

Among other things the Mycroft stack lacked easily installable packages, automated updates for non-Mycroft enclosures and, importantly, a comprehensive testing framework. The result has been a stack that is feature rich, but buggy, hard to install and difficult to maintain.

So we’re taking our change in leadership as an opportunity to rewind our messaging revising our roadmap to production.

Hardware

Michael has been very clear that he doesn’t see Mycroft as a hardware company. Our smart speakers are reference devices for Mycroft-Core in the same way that the Pixel line of smartphones are a reference device for Android. Our plans are to eventually enable thousands of software developers, automakers, robotics companies, device manufacturers and even drone companies to deploy voice technology in their products.

Many use cases for Mycroft, however, do require a reference speaker. Organizations like hotels and hospitals are eager to explore voice assistants as supplementary technologies for concierge desks and rooms, nursing stations and patient bedsides, but to do so they need a physical device.

So we are working hard to make the Mark II that device. A reference device for backers that can also be used by schools, hospitals and hotels to serve their customers.

So where are we in our quest to ship the Mark II? The good news is that we appear to be past the big drops and loop-de-loops. Today we’ve got more experienced leadership and are moving down the path to delivering on our promises.

Based on our new CEO’s experience, this is where the company is in the development process:

Mark II hardware stages
Mark II hardware stages658×771

We have progressed from using commercial off-the-shelf parts for the proof-of-concept, to developing a PCB for the manufacturable prototype. This decision was made for a variety of reasons, but both performance and cost were significant factors.

We’re not ready to release a timeline for the first DFM Prototype PCBs, but the design is well underway and under active development using a low-risk process of integrating the off-the-shelf components of the PoC into a single PCB.

Mark II daughterboard render

Mark II daughterboard render
Mark II daughterboard render600×536

The hardware production process should move swiftly once we’ve verified the PCB functionality and integrity. In fact, we anticipate that the long-pole in shipping a stable product is going to be the software, which is where the majority of our development team is focused.

Software

Despite more than five years of effort, 4,400 Github stars on our primary repository, contributions from hundreds of software developers and thousands of other contributors our software remains early.

This is understandable given the complexity of the problem we’re trying to solve. After evaluating our software stack, we’ve decided to rewind our messaging to be more in line with industry standards. As a result Mycroft is back in alpha where it will remain until we solve issues related to performance and content.

Updates

We try to balance being transparent and forthcoming with being too spammy. Hopefully we get it right, but if you feel you’re not getting enough information, please take a moment to subscribe to our e-mail list or check in on our blog from time-to-time.

As we stated earlier this year, the expectation for Mark II delivery stands at 2021. When we can give you a firm timeline, we will. If we can make it happen sooner, we will. Progress is dependent on staffing and distractions like patent trolls, which in turn affect, and are affected by, fundraising. But for now we can commit to keep you updated on our actual progress, and setbacks, on the path to fulfilling this Kickstarter.

Building new technology is quite a ride. Thank you for joining us on the journey.

5 Likes
2

Joshua, thanks for the update. Can you elaborate a little bit more on the hardware part, please?

Looking at the PCB drawings I conclude that the SBC will be Raspberry Pi - 3 or 4?
From the way the RPI connector is aligned on the PCB and the LED pixel ring on the backside I guess there will be a new hardware enclosure?
The daughterboard shown is obviously for audio output only (DAC and Class-D amplifier ICs). What about microphone input, will there be another custom PCB or will the final design still use the Respeaker 4mic USB board?

Thanks in advance.

3

Everything needed will be on board ( DAC, mics, LEDs, buttons ). It is indeed a Pi4 daughter board.

You are correct that we’re refining the enclosure. It will continue to have the same overall configuration ( stereo speakers, screen, mic, physical microphone toggle ), but will likely have a different shape to accommodate the daughterboard.

We haven’t made any decisions on the specific shape, but are leaning toward a cube configuration. There are so many tradeoffs between form factor, manufacturability, features and performance that the form needs to be driven by the function.

4

Thanks for the update, that was long waited, but let me be a bit critical here.

I’m a bit concerned about the departure of Steve and Ake lately, it looks to me that we’re leaving behind the tech professionals who could boost Mark II launch. It seems to me Mycroft Inc is leaving behind engineers and just keeping CEOs and community managers, so I don’t see who’s really working in deliver a real product.

When I saw Michael Lewis taking the CEO’s hat, my first impression after reading his profile was become afraid of the Mycroft’s future, as it seems Michael is very good at selling whatever it touches to someone else. And an irrational fear arose: will be Mycroft AI sell to a big player who still doesn’t have any assistant.

I know some of you have spent many years of your life in this project, it has been long, hard, and very difficult, and perhaps if someone can ease your burden with some millions and forget it all… well, everyone is human and can get burnt. I’m an investor myself, and I invested in Mycroft because I believe in an open source AI, not to gain revenue.

I guess if your next movement is to introduce several new engineers instead of even more leaving the ship (if still remains any), will make change my mind and air out my fears.

As for now, I’m probably wrong and there are still a few engineers working on it, I hope so. But I don’t really know who is really working on the Mark II at the moment. In fact, I’ve lost sight of any “pure” tech reference of Mycroft Inc, and it feels as is for now just @gez-mycroft is doing what he can, but to be honest, I saw more improvements lately on the community development field than in the core, and I saw community Mark II assemblies done just by one member. Sure, that community 3D printed Mark II is not a ready-to-sell product, but it has been done by just one person in a handful of weeks, so I don’t really get why Mycroft Inc, with years of delay and several electronics and software engineers, didn’t advance in that field, just delaying and changing of mind of every aspect every few months.

Don’t get me wrong, I’m with you since the beginning of this journey, and I will probably stay with you if nothing changes because I share your vision, just becoming concerned because it seems to me that talented people are leaving and with them, a great hole that nobody is filling it, and Michael’s experience can be awesome, but I don’t see him developing or assembling anything.

3 Likes
5

@malevolent - Ake @forslund was with us for 5 years and spent the whole time working in his teeny tiny attic all by himself. From what I understand he wanted to go to work with other people every day and I don’t blame him. He’s now working on new challenges, but he’s continued to be involved as a volunteer and I hope that will continue.

I’m not going to comment on any of the other staffing changes other than to say that our team has always had dedicated engineers who are passionate, capable and hard working. I wish them well.

Michael is a much better engineering leader than I’ll ever be. Hell, the whole company now is managing projects via Jira and I can barely work a single ticket into my workflow. He’s lead large teams building really cutting edge stuff and he’s already stepped up our engineering game in significant ways. As an example, the Voigt-Kampff framework was much needed and I’d have never dedicated the resources to it.

A majority of our team continues to be technical. I’m an aerospace engineer, Michael’s a computer scientist and engineer, @ChrisVeilleux Chris Veilleux is a computer scientist, Ken Smith is a very experienced programer and Kris Gesling is…apparently…a social worker ( I’ll admit that today is the first time I’ve ever looked at @gez-mycroft’s credentials ). Explains how he is so patient the the community. Seriously though, Gez is a pretty talented developer in his own right and contributes a lot of code and expertise to the stack. Finally, there is Derick Schweppe who is an industrial designer by trade, but has stepped up to manage all of our branding, promotion, design for manufacturing, prototyping, acoustics and user experience work. He probably does more, but that is just off the top of my head. Without @derick-mycroft’s willingness to step up and become a SolidWorks guru we’d never have delivered the Mark I let alone the Mark II.

The only non-technical people on staff are Johnny DeMaddelna who basically runs all of the operations ( contracts, business development, government paperwork, shareholder paperwork, legal, etc. ), Emily St. Martin who aspires to be an attorney for some reason, but for now keeps Michael operating efficiently and Kris Adair who is part time, a co-founder and runs both the accounting effort and all of our social media and marketing.

The reality is that we’re late because we fucked up. The original vendor for the Mark II hardware couldn’t deliver on schedule and our original software development effort didn’t result in a sable software stack that could be demonstrated reliably, much less sold. That led to a crisis of confidence in the investor who signed on to provide $5M in funding ( which would have delivered the Mark II in early 2019 ).

If our software game had been on point by November 2018 and our hardware vendor had been able to deliver a working PCB we’d have delivered the Mark II 18 months ago, grown our year-over-year sales in 2019 and would be shipping both retail and wholesale globally.

Ultimately the responsibility for all of this is mine. I didn’t implement a “tight” fundraising strategy for our Series Seed which left me distracted by constant fundraising rather than focused on creating a product that didn’t suck. I closed a Series A anyway, but we weren’t as ready for our partner as the technology team claimed and it blew up in the company’s face.

So I replaced myself with someone more capable and am working to support my replacement.

I do want to address the “Michael builds and sells companies” perception though. I’m not sure what propelled Michael to start Stellar and Cryptic, but the way he explains it he was motivated by the problem he was solving and the people he was working with, not by the payoff.

He’s been crystal clear why he got involved with Mycroft. He was a backer of our Kickstarter and ended up as an investor because he believes that voice technology should be available without the ubiquitous surveillance that infests the technologies that Big Tech is shipping. He doesn’t need the money. Not only is he the lowest paid employee on staff, his foundation has provided more than $650,000 in funding over the past 6 months. Other than taking his responsibilities to our investors seriously, I’ve never gotten the impression that he’s motivated by a future payday. He was happily ensconced in his family life, sponsoring interesting projects at the university, contributing to his local maker space and teaching art to kindergarteners at his kid’s school when I talked him into taking the reigns at Mycroft.

And he is absolutely moving the ball forward with both software and hardware development. Michael’s been instrumental in implementing a software development process that will ( one day soon ) result in a stack that doesn’t suck and can be supported long term. Our current hardware vendor comes from Michael’s personal network and is making rapid progress toward a PCB that we can use on the Mark II ( and the Mark III when we’re ready ).

Honestly I’m probably the most useless person at Mycroft right now. With no product on the market and fundraising passed to Michael I’m in a bit of a holding pattern. I can spend time landing new enterprise business or getting retail distribution, but it won’t matter until we have Mark II units in boxes and ready to ship. I could help with software development I guess, but it takes so long to spin up on the process and I’m distracted so often that it is very difficult to contribute meaningfully. I’m also extremely burned out after 5 years of startup. It’s been years of long hours, stress, uncertainty, self doubt and taking big financial risks that may one day sink my family.

And the patent suit hasn’t helped. It is sucking up a huge amount of everyone’s time. It’s also very expensive. We’ve now got four separate law firms working on the case. Being principled and fighting bogus claims is our responsibility and we take it seriously. We’ll fight to the bitter end and beyond ( if the trolls think they’ll ever get a penny they are out of their freaking minds ), but if anything sinks the company I expect it will be the patent litigation. We’ll probably spend more on the bogus patent suit than we pre-sold during crowdfunding.

That said there are a lot of good things going on. Our partners at Texas A&M are about to deliver the final deliverable on our NASA SBIR phase I grant which will free us up to pursue a phase II award. Our upcoming fundraising effort is looking very promising. Michael’s becoming a solid pitch man and his passion for our community shines through. We’re targeting enough money to operate for the next 2 years with a significantly larger staff and I think it will happen.

There is huge demand for what we are doing. I suspect if I had 10,000 Mark II devices to sell I could do it in a month. Between individuals who want a private, open, customizable experience and companies that want a voice stack, but eschew Big Tech there is probably $3B in pent up demand. We just need to get a quality product out there for them to purchase.

Keep the faith. Startup is a marathon, not a sprint. We’ll keep after it until we get there.

9 Likes
6

Wow, thank you for your long and crystal clear answer, @J_Montgomery_Mycroft.

I have no doubt Michael’s expertise will improve and boost how Mycroft Inc manages all the production, besides he must be tough when negotiating, so it’s great to have it onboard. As I said, I felt an irrational fear, because when we met each other in Barcelona, I asked the very same question to you, and you stated you are committed to the vision of an open source AI as I have. So I guess Michael’s vision is the same as yours.

On the other hand, I didn’t even know the Voigt-Kampff framework has been released because of Michael, so that’s why regular updates, for insignificant they can seem, are important to us.

On my first post, I was obviously speaking just about my own fears, but surely many other community members, backers and tiny investors shared my own concerns about this awesome entrepreneur. You already aired out my fears, I hope this will be left as clear as for any other who could think about that subject.

I didn’t know about Ken Smith, I did take a look on his LinkedIn profile, looks like he’s the latest incorporation to the team, I hope he can be engaged by this project as others did before him!

What @gez-mycroft is doing is incredible, it was presented as @KathyReid successor, but, with all my love to Kathy, her padawan has surpassed the master in many fields: not only deals with all the community members, and created a comprehensible documentation, but it also links community efforts with core improvements (let’s rock lingua_franca a bit harder!) and finally he codes and improves some annoying skills (NPRGrrr skill, for instance)

The other members are as old as the project itself, but they always remained in the dark, so they are “hidden” to community members. And it’s a pity. Yes, we agree they must work on Mycroft, not creating bonds with community, but mycroft rely heavily on its community, and would be great some “daily” feedback and responses only a developer can offer. Those were done by Ake and Steve back in their days, but now we -the community- can feel somehow orphaned (i.e.: if we need an urgent PR, or we noticed skills translations were quite outdated, Ake answered in a heartbeat and arranged it quite fast, now this will fall in Gez’ shoulders, who probably will need to escalate to whom has to do it, slowing the community feedback)

I can see on the career page you’re looking for two senior engineers and another manager, so I can expect the core team will be expanded soon, although you’re limiting to USA residents, which can let out very talented and committed developers around the world.

That’s what most of us want to hear about the project. I won’t ask for an ETA, but I hope this stack will be ready sometime later this year, and be ready on 2021 when the M2 will be launched at last.

I keep the faith, the message has been received crystal clear for me. In fact, I preferred to start myself this discussion, and I really appreciate your sincere and fast answer.

Keep doing the good job! :wink:

3 Likes
7

I am really busy with other things lately, but reading back here and there, but have to compliment @J_Montgomery_Mycroft for that blunt but very honest post.

Link this thread on Kickstarter and people will appreciate it.

5 Likes
8

I’ve long been a fan of publishing all of our updates across all of our channels ( forums, Kickstarter, Indiegogo, Facebook, Crunchbase, AngelList, etc. ), but apparently their is a group of backers on Kickstarter/Indiegogo who consider this spam and get hostile about it.

There is also a group that gets angry no matter what. They basically view Kickstarter as a store and have gotten worked up because the “product” they bought wasn’t shipped on time. Lots of hostility and hate.

So we try to manage that channel ( and Indiegogo ) with a little more sensitivity to that audience.

If you ask the team they’ll tell you that I’ve brought up the “lets just be blunt” communications strategy dozens of times and been ( subtly and with great sensitivity to my feelings ) overruled.

Since I trust the team we’ve got working with the community, crowdfunding backers, and social media, I’ve deferred to their judgement on the matter.

I don’t even get carte blanche to share my opinion that the patent trolls are blood sucking, inbred, anti-social, wastes of space. They whined to the judge that I was in some way harassing them ( I’m not ) and got my original post sensored. Have a peek at the way back machine or the press coverage to see what was removed.

I do have nearly complete freedom with investor updates and I try to make them comprehensive.

5 Likes
9

I completely appreciate full transparency, or being blunt. I know it is difficult, but ignore the loud few on Kickstarter and the like please :slight_smile:

2 Likes
10

I generally try to discount the shriller of the Kickstarter backers, but they are still customers. Companies ignore their customers at their peril. The shrillest customers are often helpful indicators of what the company is doing wrong ( in this case - being VERY late with the product ).

2 Likes
11

Congratulations on a fabulous piece of spin!

I have bitten my tongue many times when reading updates on the project, primarily because the goals of an independent non-network based voice system are so important for the future. Critically important.

And yet…

Every time I have tried to make the deliverables (on a Pi) work, they fail. Now, I’m no software wizard, but I do have a modicum of experience and a reasonable degree of determination. I also own a hardware/software company that designs, manufactures and sells a technically complex product, so have a reasonably good understanding of most, if not quite all, the issues.

Frankly, I would like to hear directly from the new CEO - a loooooong overdue change BTW -and get a straight version of what the future holds. Even if it is “you won’t be getting a MkII” then we can move on with a different approach - perhaps Jasper is the platform that we can (should?) put our efforts into.

Your credibility, and authority, vanishes every time I hear something like “The April update was a bit late so decided to reset for June.” - “The June update was a bit late so decided to reset for October”. I don’t know if you understand how risible a throwaway comment like that is. And a glib throwaway of two years late. Thirty years ago I moved on development managers who reported progress in such a way. With todays systems, processes and team management techniques, the days of lame excuses like that, late in the day, are long gone.

Yes, projects are hard. And this may be leading edge technology, but NOT bleeding edge. At least not two years later. And the last 10% is 90% of the struggle. But we don’t appear to be even close to the stage of refining and polishing, probably not in 2020.

I would ask you to step completely out of the way and hope that in doing so, competent managers can step up and deliver on the laudable goals of the project. I applaud your efforts in starting - and sticking with (it’s easy to give up) - the project. But it’s time to let the juvenile fly the nest without parents.

Cheers

12

Hi Graham, I wanted to clarify that it’s often me responding on Kickstarter and I posted that we were resetting for June in terms of the updates. I wanted to let people know when to expect further communication rather than just ghost the community. The way the platform is setup you can’t really have multiple users so it all comes from Joshua but they are updates from the company.

We try to be as open and honest with where we are at. If we were not going to deliver the Mark II, we would have said this a long time ago. There is absolutely zero benefit to us lying about this. As we’ve said, the Mark II won’t be delivered in 2020 but we’ll be keeping everyone updated on the progress and provide a clearer timeline for delivery when we can.

13

See, @gez-mycroft with such diplomacy. Putting his social work skills to use. He’s awesome, right?

I do want to clarify that, as Harry Truman said, “the buck stops here”.

Ultimately I’m responsible for the late delivery and anything else the company has got wrong including late updates and any inconsistent messaging.

Graham has some valid criticisms and I acknowledge his feelings on the matter. Like many backers he is expressing frustration. We’re listening. We’ll continue to work to do better with communication and progress.

14

Steve and Ake are both still involved, though the manner and depth of that has changed.

15

One can be involved by volunteering or getting paid, the former means they pop up when they want or have time to, just as me or you, while the later means they are there every day -and, in the case of Mycroft- of the week at any time.

Anyway I didn’t say they weren’t, or at least, that wasn’t what I wanted to express. I meant they left Mycroft and nobody replaced their vacant (later I knew Ken Smith is probably replacing Ake because he started 2 months ago and being a developer, and Mycroft is hiring two engineers more).