Anki Closing Down

Anki, maker of Cozmo and Vector, as well as Overdrive etc, is closing.

Their robots rely on an App to function.

Through the years, we've seen many come and go. The one robotics company we've been able to rely on consistently is Parallax.

Anki Shutting Down

We'll see if these Anki robots turn in to expensive bricks. One of the great things about the ActivityBot. You can count on that to stick around.
Founder of Kinvert
https://www.kinvert.com/

Comments

  • Apparently they had 200 million in funding and 100 million revenue 2017.
    Founder of Kinvert
    https://www.kinvert.com/
  • jmgjmg Posts: 13,462
    Apparently they had 200 million in funding and 100 million revenue 2017.

    Hmm $100M in revenue should be enough to keep someone in business, they just need to size their overheads to suit ?
    $200M is a lot to burn through, to then not be able to sustain things.
    Could there be more to this story ?
  • There could.

    There might be a couple guys floating down on golden parachutes.

    If they have about 200 employees, that's a million bucks per employee. Big generalization there, but holy moly.
    Founder of Kinvert
    https://www.kinvert.com/
  • That was abrupt!
  • Indeed it was.

    Will be interesting to see if more light is shed on how they blasted through that much money so quickly.

    One of their devs is still doing commits so there is still some hope the bots won't be bricks. Who knows. Cozmo may be salvaged but Vector will be harder.
    Founder of Kinvert
    https://www.kinvert.com/
  • As an armchair critic, I think their dials were not properly adjusted.

    They likely over-estimated the market potential, underestimated the true design and tooling costs, and also found themselves in a very competitive market with products like the Ozo bot and others (which are usually also funded). Anybody who runs a VC-funded business should first spend a year running one of the following: gas station, convenience mart, a small restaurant, a simple brick-and-mortar store. My thinking is that if somebody can survive in these business models they'll appreciate cost control. For the startup side of things, running a business from a https://www.ycombinator.com/ fund is also a good experience.

    It would be most interesting to talk to people who were in Anki for the entire run in order to understand exactly what happened here.

    Ken Gracey
  • Yeah it will be interesting as more of the story unfolds.

    I don't think they pulled off a lot of cost control if their office was in the heart of San Francisco.

    One of the big eye openers is how these robots could be bricked. This won't happen with the Parallax products we use as they aren't tied to the cloud or an app. But in the case of Anki people could have spent a lot of money on a paper weight.
    Founder of Kinvert
    https://www.kinvert.com/
  • If Anki robots were open source, the community could build a privately owned cloud under the same Internet address. Another +1 to Parallax
    That's not an explosion! That's an unscheduled rapid disassembly!
  • Absolutely.
    Founder of Kinvert
    https://www.kinvert.com/
  • The Vector robot is a self contained robot that does not rely on a cloud server to operate. These units are not like Alexa or Google devices that send everything they here back to the cloud. These units will continue to operate after the cloud is shut down. Just don't try to ask it what the weather is since it has nothing to talk too.

    You can however continue to tell it to do things that don't require an outside connection.

    Mike
  • iseries wrote: »
    The Vector robot is a self contained robot that does not rely on a cloud server to operate. These units are not like Alexa or Google devices that send everything they here back to the cloud. These units will continue to operate after the cloud is shut down. Just don't try to ask it what the weather is since it has nothing to talk too.

    You can however continue to tell it to do things that don't require an outside connection.

    Mike

    Actually, I do think the Vector communicates back to the Anki cloud server since you have to set-up the Vector License Client via WiFi to activate it. Also, I was under the impression that the Vector does send data back to Anki so they can, or could, continue to improve the AI in the Vector; well at least when they were still working on it. Funny, the Anki Privacy Policy is not available at this time. The real question is whether or not the license will still be active when they shut down the Anki servers. If not then this could disable the Vector.
  • When you setup your Vector robot it connects to there server to get a certificate that is only valid with your robot. Once you have that certificate it will except your encrypted connection so that you can control it. Without that certificate you have no way to talk to the robot using SSL. All communication with the robot uses HTTPS.

    Mike
  • From the Anki Privacy and Security page:
    https://support.anki.com/hc/en-us/articles/360007560234-Vector-Security-Privacy-FAQs

    "11) How does Vector handle Security?

    Vector includes a number of security best practices, including security and privacy features like visual cues (light on his back) when the robot is streaming data to the cloud. Customer data is encrypted between the robot and cloud, and code running on Vector is signed and authorized by Anki. Our security team extensively tests for and has systems to detect threats and vulnerabilities on our products."

    The link to their actual Privacy Policy is not working.

    I remember the Vector being discussed on a Podcast where the sending of user data being sent back to Anki was in question. Since it has a camera, the real question was whether or not the images were being sent back to Anki. This would be a big issue in the UK especially when kids are involved.
  • I saw another report and a statement from them indicating that all camera data is local only and is not send outside the home. You can view pictures that is takes or you can see what vector sees but the images are not recorded and send anywhere.

    Privacy is there number one agenda with the unit.

    Mike
  • A lot of people in the Anki forums are saying that all speech recognition happens through the cloud.
    Founder of Kinvert
    https://www.kinvert.com/
  • JonM wrote: »
    code running on Vector is signed and authorized by Anki

    So cryptographic lockout is an advertised feature now?

  • ercoerco Posts: 19,441
    WOW, just saw this. Sad to hear about this, Anki always raised the bar. It's definitely worth studying what happened to this once very healthy business. Possibly just like a stock portfolio, this company needed more diversity than just high-tech robots. I will dig deeper, I have a friend who used to work there.

    https://www.vox.com/2019/4/29/18522966/anki-robot-cozmo-staff-layoffs-robotics-toys-boris-sofman
    "When you make a thing, a thing that is new, it is so complicated making it that it is bound to be ugly. But those that make it after you, they don’t have to worry about making it. And they can make it pretty, and so everybody can like it when others make it after you."

    - Pablo Picasso
  • ercoerco Posts: 19,441
    Related, I live two miles from Ozobot's headquarters in Redondo Beach. Now THAT company I could more easily see closing. Their entire product line is miniature line-following robots. I actually interviewed there but just could not get excited over the place. But by all indications they are healthy and doing well; they have plenty of employees here and also a huge staff of programmers in the Czech republic. Anki had much more interesting robots IMO.
    "When you make a thing, a thing that is new, it is so complicated making it that it is bound to be ugly. But those that make it after you, they don’t have to worry about making it. And they can make it pretty, and so everybody can like it when others make it after you."

    - Pablo Picasso
  • Cluso99Cluso99 Posts: 14,911
    edited 2019-05-04 - 21:41:25
    Their privacy policy will mean nothing when the liquidators sell off their database. If you gave them your credit card details, time to get a new credit card!
    My Prop boards: P8XBlade2, RamBlade, CpuBlade, TriBlade
    Prop OS (also see Sphinx, PropDos, PropCmd, Spinix)
    Website: www.clusos.com
    Prop Tools (Index) , Emulators (Index) , ZiCog (Z80)
  • erco wrote: »
    Related, I live two miles from Ozobot's headquarters in Redondo Beach. Now THAT company I could more easily see closing. Their entire product line is miniature line-following robots. I actually interviewed there but just could not get excited over the place. But by all indications they are healthy and doing well; they have plenty of employees here and also a huge staff of programmers in the Czech republic. Anki had much more interesting robots IMO.

    Ozobots as far as I can tell is pure marketing. Teachers talk about it all the time and I ask them what kids are coding, what they're learning, and they just focus on how it's fun and easy to get started.
    Founder of Kinvert
    https://www.kinvert.com/
  • I've been thinking about this thread for days and I came to the conclusion that VC-funded startups can't easily have the best interest of educational customers in mind, or at least their business model isn't totally compatible with educational needs. Since most startups fail (or 9 of 10), an educational institution who buys their products will be left without support in the long term. Further, startups are often one-hit wonders in that they invest tremendous amounts of money in tooling, design, marketing around a narrow product line. This doesn't allow for much product progression, unless they make it big.

    Boot-strapped businesses, on the other hand, have a short hand-to-mouth pipeline and don't have the luxury of extensive credit or cash. While it takes them longer to achieve product lines but if they succeed I think there's going to be more stable. This long cycle allows a company to truly understand customer needs, too.

    I'd like to know what others think. Can a startup reliably serve educational customers for the long-term?

    Ken Gracey
  • Cluso99Cluso99 Posts: 14,911
    edited 2019-05-05 - 03:26:36
    @Ken, etc

    When we (my business partners) spoke to VCs in the US back in the 90s, it was 1:100 that became successful. Of course, successful was for the VCs to make their killing from floating on the stock exchange.

    As for the market they serve, it depends on the product(s) they have. If they have something no-one else has, and it solves your requirements, then go for it. If it is similar to other products, then IMHO it's a steer clear product.

    I'd rather go for a smaller company that has been around for years. They have more than likely understood their market, look after their customers, and will maintain and improve their products. Parallax is one that fits this model nicely :smiley:
    My Prop boards: P8XBlade2, RamBlade, CpuBlade, TriBlade
    Prop OS (also see Sphinx, PropDos, PropCmd, Spinix)
    Website: www.clusos.com
    Prop Tools (Index) , Emulators (Index) , ZiCog (Z80)
  • Ken, as someone who has considerable startup experience, and who is currently keeping the lights on at a company while the VC tries to sell off the IP, I think your assessment of the situation is right. I think that pressure is even worse these days, due to all the companies going public you read about. It's "Go Big or Go Home" which is not what the edu market needs.
    San Mateo, CA
  • Since were talking about startups and company failures which I have been apart of several times now.

    Unfortunately this product requires backend servers where as my previous failures where just products that would continue to work long after the company has ceased operation.

    I just invested through Kickstart, and Anki was also through Kickstarter, in RVR. The Sphero company setout to build this product and was asking for 150,000 in pledges to build it. Well they go 1 million dollars and just short of 3500 people to fund it. Now they have to build it and at the same time make a profit.

    I like Kickstarters as it builds a product that I like at a reasonable price. But I don't believe this is any kind of business model since once the people have there product, the demand was the initial offering.

    So how long before Sphero has the same problem?

    Like 3DR and Anki they decided to do retail. This seems to be common issue as well.

    The RVR product however requires no backend servers so it will continue to work so I'm good with that.

    Mike
  • Yeah they'll be sending us a RVR I think in October. Interested to see how it goes with that robot.

    It's tough because on the one hand with VC funding they can quickly get a good product up. Cozmo is way more engaging with student than anything I've seen and it has a pretty solid SDK in Python.

    But some parts seem rushed. For example they have an algorithm to dock with cubes. Well in our experience Cozmo misses say 25% of the time in Code Lab and Python.

    So when it was time for a robotics competition I had to rewrite all that. Took a long time. Now his miss rate was 5% or so.

    Another issue is I think VC backed companies will generally take a more black box approach. S3 has a hacking port. Cozmo? Vector? No way.

    Oh, by the way Parallax and @Ken Gracey , they had a lot of great engineers looking for work. Hint hint.

    The Anki community absolutely loved Michelle Sintov who is likely looking for work right now.
    Founder of Kinvert
    https://www.kinvert.com/
  • I can't speak to what the current hotness is in education, but I did grow up in my father's physics lab. He did some research -- his lab used nuclear magnetic resonance to detect lead from gasoline in the air at major intersections in New Orleans in the 1970's -- but mostly he was an educator, teaching undergraduates at Southern University in New Orleans, or SUNO, an historically black university. Those who met me at the UPEW might find this a bit odd since I'm about as black as a bucket of titanium oxide. Dad was literally the token white guy on the faculty. And I learned early the usefulness of improvised tech. We made a jig to split the wax paper strips for the "acceleration due to gravity" experiment so that more students could do the experiment. And as a high schooler I built a better spark generator for it from an automobile spark coil, and instructions from The Physics Teacher magazine. It was his lab's HP2100 minicomputer that introduced me to computers when a TRS-80 cost as much as a running used car.

    Thirty plus years later I can say that Parallax is definitely the company my father and I would have wished into existence in those days. Their presence is a constant and their commitment to service is unlike that of any other company I have ever dealt with. I never worry whether they will be around next year. They innovate, sometimes dramatically, but always within an envelope that does not create an existential threat. They never burden their customers. A couple of years ago I received qty 10 Human Interface Boards, p/n 40001, for an order of ten quickstarts p/n 40000. When I called them Parallax shipped the quickstarts next day air and did not ask for the HIB's back. I am still using a few of those HIB's. Really nice product. No RMA's or any of that noise. Just total customer service and, as always in my case, satisfaction.

    The problem with startups is that somebody is looking to turn over their investment. No matter how good they are, no matter how wonderful their product, it's about the $$$$$. By contrast, I have never gotten that impression about Parallax. They are about the product, the customer, the user experience. I believe Parallax follows the old canard that they don't do those things to make money, they make money as necessary so they can do those things. It shows in everything they do from their presence on these forums to how they handle a screwed up order, or making the P2-ES's even though there was never any money in that for them but they knew it was a chance to jumpstart a community of P2 users.

    I just took a robot off a shelf that I built in 2002. Haven't touched it in many years but at work we have a use for the electronics now. It is of course built around a Basic Stamp 2, a product that is still around and supported so I know I will be able to get it back online. That is a depressingly infrequent thing to find in industry. Even the PLC business, which used to be legendary for its support of 20 year old hardware, is slipping. But if you need a BS1 and the means to program it, Parallax is there for us. Nobody else does that.
  • Hey the lead in gas research was featured in the new Cosmos show. Fascinating story of Physicist vs. energy industry.

    Ozobots are great. That was a great idea. My lab has a STEM fair at TOCTWD and they had a fleet of those again this year.

    Anki overdrive was great.
    Sorry to hear they are closing. I was just looking at expanding our set before xmas and decided not to.
    I feel like they could have made improvement to the software that would make me want to by in more.
    I think there's something there.
    My son played with it for a while and then it just sat in the closet.
    Maybe hard to compete with Fortnite….
    Prop Info and Apps: http://www.rayslogic.com/
  • I'm not so sure that it is just a startup thing, but more of having some agility in the concept so when things go south, the company can refocus and try another path. With regards to Anki, it is still unknown exactly why they failed and is still not mentioned on their site. I've heard the employees have been given walking papers and the main location is a ghost town, but Anki is being a bit quite as to why they are shutting down.

    There have been seemly cool products such as Jimbo and Kuri that have met their doom recently as well as Baxter by Rethink Robotics so I think the market these are trying to hit is just a tough one. I'm still keeping my eye on Misty Robotics which is marketing it's robot as educational, but is in the $2,400 USD price range for pre-order and requires something like a Raspberry Pi to run it.

    When looking a success story such as iRobot, they started with the Genghis Robot 1990 which was designed to find a way to make robots intelligent using a "subsumption architecture". iRobot even had the iRobot PackBot® which was used to search the World Trade Center area after the 9/11 attack. It wasn't until 2002 that they came up with the first Roomba thus launching them as a major player in the consumer space. They do have the iCreate 2 which is an educational platform version of the Roomba.

    Then you have the whole Lego Mindstorm thing which was first conceived in the '80s and inspired by the book Mindstorms: Children, Computers, and Powerful Ideas by MIT professor Seymour Papert and his Logo programming language. Although the Mindstorm Brick has evolved, the premise is still that it is an educational tool that is easy to use. After the recent NorCal Robotic event at Sierra College, the Mindstorm is still quite popular.

    In both the iRobot and the Mindstorm cases, each where designed to solve a problem and not just to be a cute little desktop companion. It will be interesting to see how companion bots do evolve over time though.

    NVIDIA has had some pretty good traction with their Jetson line of Boards, and with the new $99 Nano, I can see this as a major player in the Robotic Educational space given its power and flexibility and AI capability.

Sign In or Register to comment.