I’m pleased to announce that I’ll be seeing you Varsovians tomorrow night at our pre-holiday meet-up. The Warsaw event, here, is on the 19th at WeWork in Warsaw. I’ve already chosen folks to pitch but please come and support your fellow start-uppers.
Special thanks to WeWork Labs in Warsaw for supplying some beer and pizza for the event and, as always, special thanks to Dermot Corr and Ahmad Piraiee for putting these things together. See you soon!
A few months after Nuro href="https://techcrunch.com/2018/08/16/nuro-and-kroger-are-deploying-self-driving-cars-for-grocery-delivery-in-arizona-today/">deployed self-driving cars to deliver groceries in partnership with Kroger, the autonomous delivery startup is deploying its custom delivery bots. Up until now, Nuro was relying on Prius vehicles and safety drivers.
Now, its delivery service in partnership with Kroger will be completely driverless and without a safety driver on board. Nuro has been working on this vehicle, the R1 since 2016.
“Nuro envisions a world without errands, where everything is on-demand and can be delivered affordably,” Nuro President Dave Ferguson said in a press release. “Operating a delivery service using our custom unmanned vehicles is an important first step toward that goal.”
Nuro’s intent is to use its self-driving technology in the last mile for the delivery of local goods and services. That could be things like groceries, dry cleaning, an item you left at a friend’s house or really anything within city limits that can fit inside one of Nuro’s vehicles. Nuro has two compartments that can fit up to six grocery bags each.
Twitter is giving users the ability to easily switch between seeing the latest tweets first and seeing the company’s algorithmically chosen “Top Tweets” when they open the app.
The company began testing this feature a few weeks ago, but they are officially rolling it out globally to all iOS users today with Android and desktop users likely getting access to the feature sometime in January, according to the company.
This is part-resolution and part extended cop-out for Twitter which has spent the better part of the past couple years figuring out how to satisfy a need for growth with vocal, loyal users who want the act of opening the app to continue to mean getting the immediate pulse of the internet. The algorithmic timeline is probably a better business move for Twitter, something that will ensure that more causal users can get a more encapsulated experiences when they open the app rather than a hodgepodge snapshot of their followers’ thoughts.
The company has explicitly said that the “Top Tweets” feature has increased both engagement and conversations on the app.
Twitter’s solution to its algorithmic ails is called the sparkle button and it sits in the top right of your screen allowing users to essentially temporarily disable “Top Tweets” and enjoy a pure reverse chronological Twitter feed. Though the company see a lot of utility in algorithmic feeds, they also acknowledge that recency is critical to the ethos of Twitter and that in certain instances like a sporting event or breaking news situation, there’s a lot of value in seeing what’s new immediately.
You don’t necessarily get to set your preference in stone, though.
In what is likely to be a controversial move, “Top Tweets” is enabled by default and it seems that you will have to re-enable the feature periodically though Twitter says it’s experimenting with how often that is, though it will at least remember your preference for the entirety of your session. So, if you check your mentions and tap on the Home tab you won’t return to “Top Tweets” if you had previously been browsing the reverse chronological feed.
Though the company has said today’s rollout won’t be affecting the volume or frequency of “Top Tweets” postings, it’s clear that this feature launch gives the company a much longer leash to experiment with changing up the core timeline and in general offers the company a proper means to get reverse chronological hold-outs people to gradually adapt the algorithmic feed without feeling quite as forced to do so.
It would have been great if Twitter introduced this months ago, for all the philosophical shifts that the algorithmic timeline signaled the most annoying part of the change was how users were given mixed messages about being able to choose whether they could keep the old timeline.
Today’s move is likely to make most users happy though. Unburying important toggles from the depths of settings is always welcome, and communicating major changes is especially important tp a company like Twitter that doesn’t often change up its core product dramatically.
Bumble has come up with a new way for its dating app and related businesses to generate revenue. The company this week launched filters – a way to sift through potential matches by a set of specific criteria. For example, Bumble Date users can now filter matches by astrological sign or relationship type, among other things, while those on Bumble BFF or Bumble Bizz, could filter matches by interests or industry, respectively.
The new feature is meant to save users time by limiting their selection of potential matches to those who are more relevant to their own interests.
A dating app user may want to filter out those who are only looking for casual situations, while a business user may want to filter matches based on whether they’re looking for a job, mentor, or collaborator, Bumble explains. And on Bumble’s friend-finding platform, Bumble BFF, people may want to filter for people who enjoy the same things they do – like fitness or photography.
“We’ve been working internally and with our users to create just the right mix of filters that allow for deeper, more meaningful connections and we’re very pleased with what we’ve developed,” said Alexandra Williamson, Bumble Chief of Brand, in a statement about the launch. “Whether you’re looking for a new job in media, a new mom friend or a date with a Sagittarius who loves live music, Bumble Filters enable you to tailor your experience in a way that ultimately gives you more control of the kinds of relationships you’re looking to build,” she said.
Filtering matches by specific criteria isn’t anything new to dating apps. Other more traditional dating sites, like Match and OKCupid, have offered ways to filter matches, too. But Bumble’s more direct rival Tinder has focused less on filtering and more on speed of moving through matches. It doesn’t let users specify preferences beyond some basics – like location, distance, gender and age.
Whether or not filtering actually helps in delivering a good match, however, is less clear. But it’s certainly something people want.
Today, many women on dating apps ask men for their height, for instance – so often, in fact, that men began volunteering this information on their profiles, even if the profile doesn’t have a field for height. Often, sober people don’t want to match with people who say they drink regularly. Non-smokers generally want to date the same. And so on. But over-filtering could lead to users missing out – after all, how important is the star sign, really, or whether they have pets? (Allergies notwithstanding, of course.)
On the dating side of Bumble, the new filters include height, exercise, star sign, education, drinking, smoking, pets, relationship type, family plans, religion, and political leaning.
Bumble BFFs can filter for drinking, smoking, exercise and pets, too, as well as type of friendship, relationship status, whether they have kids, or if they’re new to the area.
And Bumble Bizz users can filter by industry, networking relationship type, education, and years of experience.
Bumble hopes filters will be an additional stream of revenue for its business, which it said in September was on track for a revenue run rate to $200 million per year. Bumble now claims 46 million users.
The company says all users will receive two free filters in Bumble Date, Bumble BFF, and Bumble Bizz, but additional filters will have to be purchased through Bumble Boost – the premium upgrade that also allow you to see who liked you, extend your matches, and rematch expired connections. (Boost’s pricing varies based on the time frame – a week, a month, etc. Its weekly plan is $8.99/week, currently).
Bumble’s filters are available on both iOS and Android.
Under the guidance of Shalin Mantri, a former product lead at Uber for its advanced technologies group, Skip is gearing up to make its scooters smarter and improve its businesses unit economics.
“When I think about opportunities to figure out our unit economics,” Mantri, who joined Skip from Uber in November, told TechCrunch. “It’s no secret now — it was probably a dirty secret of the industry, you know, a few months ago — that it’s hard to make money, and some of the biggest challenges to doing that are the cost of charging, the lifetime of the battery, the repair costs, the depreciation of these things being used in a fleet use case and the last is vandalism and theft, which is another big issue.”
What Mantri is referring to is the fact that investors, who have poured millions of dollars into electric scooter startups like Bird and Lime, are now pumping the breaks on funding due to the difficulty of the business. Some scooters reportedly only last about two months, which is not enough time to recoup the cost of purchasing the scooter. Perhaps that’s why Skip reportedly received $100 million in debt earlier this month. Skip, however, declined to comment on the lifespan of its scooters and its debt financing.
So in an effort to improve its unit economics, Skip is unveiling two new scooters. The first is one with a rear-facing camera and swappable battery, and the other is with a retractable lock. In Washington, D.C., Skip recently started testing 200 scooters with rear-facing cameras to look for whether people are riding on the sidewalk, parking compliance and generally safe riding behavior. Today, Skip is rolling out some of these scooters in San Francisco.
The goal of the cameras is to learn more about rider behavior and parking compliance. Skip, however, has yet to determine what it will do with the data collected from the cameras.
“I think we’ll experiment with both approaches — which is providing sort of real-time feedback and saying, ‘hey, we think this isn’t parked correctly — please move it to the side” or while you’re riding on the sidewalk, sound a little bell and that’s your signal to try to get off the sidewalk,” Mantri told TechCrunch.
Another experiment is remotely slowing the vehicle if someone’s riding on the sidewalk. But Mantri says, if he “were to pontificate, it will be better for us to do something post-trip — just to ensure we’re not distracting or taking away from the base simplicity of experience and creating a positive tone with riders.”
The cameras are not always on, and only take snapshot photos during a rider’s trip. Riders will eventually have the option to opt-out of having the camera on during their ride. There is not currently an opt-out feature, but Skip says the cameras it will be testing in SF and D.C. will not take photos of riders in a way that could identify them.
“There are interesting moments where you may want to turn on the camera and wake it up,” he said. “I think we’re going to find these interesting use cases of stuff we’ll want to do but the caveat will be privacy of course — is really important.”
Meanwhile, theft and vandalism are rampant in the electric scooter space. This is where the locks can come in. Skip has been testing these locks in San Francisco and plans to deploy more in January.
“It’s been a real problem,” Mantri said in response to a question about theft and vandalism. “I don’t have numbers at my fingertips to quantify that for you but it’s a lot. It’s getting a lot of our attention. Our hypothesis is that certainly locking and having a locking mechanism will help with this.”
Skip doesn’t expect this to be a catch-all solution, Mantri said, but the company thinks locking will surely help with parking compliance. Down the road, Skip will deploy a handful of scooters with a front-facing camera. The cameras are part of Skip’s larger vision to become an innovator in the electric scooter space.
“I think in scooters specifically, there’s a lot of room for product innovation,” he said. “And this is full stack innovation. This is hardware, sensor technology and software. The scooters need to be smarter and they need to be safer. I look at what is currently on the streets and it’s a version zero dot one of what that thing needs to be.”
And Mantri sees some of this innovation being informed from his work on autonomous vehicles at Uber. Many of the technologies Uber built in autonomous cars, Mantri said, are relevant for solving certain problems with scooters. For example, autonomous vehicle technology requires knowing where vehicles are, if they’re navigating properly and other types of system intelligence.
“That sort of system intelligence is not something that off-the-shelf scooters have,” Mantri said. “It’s just hardware but that’s where the operating system sort of comes in, and people don’t really understand that or see that as much yet. It’s such a critical layer because that’s the first line of defense that even a human doesn’t have to get involved in.”
Skip also wants to bring more intelligence to repositioning the fleets in places that have high demand, and build a smarter task platform for the people who charge the scooters.
“We have the ability to work on things that bridge us to that future, but the point is, it’s not technology for technology’s sake,” Mantri said. “It’s about the unit economics and doing what we can to make the business work.”
At this point, Skip’s scooters are not totally custom but Skip plans to eventually get to the point where it designs its scooters from the ground up. And as it experiments with different scooter types — ones with swappable batteries, locks, cameras, etc. — those learnings will contribute to Skip’s hardware product roadmap.
When you have as much data under management as Box does, you have the key ingredient for artificial intelligence and machine learning, which feeds on copious amounts of data. Box is giving developers access to this data, while letting them choose the AI and machine learning algorithms they want to use. Today, the company announced the general availability of the Box Skills SDK, originally announced at BoxWorks a year ago.
Jeetu Patel, Box’s chief product officer and chief strategy officer, says Beta customers have been focusing on use cases specific to each company. They have been pulling information from different classes of content that matter most to them to bring an element of automation to their content management. “If there’s a way to bring a level of automation with machine learning, rather than doing it manually, that would meaningfully change the way that business processes can function,” Patel told TechCrunch.
Among the use cases Box has been seeing with the 300 Beta testers, is using artificial intelligence to recognize the contents of a photo for the purpose of auto tagging, thereby eliminating the need for humans to do that tagging. Another example is in contract management where the terms are pulled automatically from the contract, saving the legal team from having to do this.
Where this can get really powerful though is that the Skill can drive a more complex automated workflow inside of Box. If, for example, the Skill is driving the creation of automated metadata, that can in turn drive a workflow, Patel said.
Box is providing the means to ingest Box data into a given AI or machine learning algorithm, but instead of trying to create those on its own, it’s been relying on partners who have more specific expertise such as IBM Watson, Microsoft Azure, Google Cloud Platform and Amazon Web Services. In fact, Box says it is working with dozens of AI and machine learning partners.
For customers who aren’t comfortable doing any of this on their own, Box is also providing a consulting service, where it can come into a customer and help work through a set of requirements and choose the best algorithm for the job.
Today was supposed to be a historic day with four rocket launches by four different companies. But that’s not going to happen. As of publication, three of the four rocket launches are canceled. The flights were pushed until tomorrow, setting up another and more significant historical event.
If all the rockets currently scheduled launch as planned, there could be five launches within a 24 hour period.
Originally, today, SpaceX, Blue Origin, Arianespace, and ULA were supposed to launch their respective rockets. Rocket Tuesday, people were calling it. And for good reason as, if successful, it would have been a historic event. But then Arianespace canceled their flight due to high-altitude winds. Blue Origin scrubbed its launch because of the rocket’s ground infrastructure. And seven minutes before SpaceX’s Falcon 9 was set to launch, an onboard computer triggered an abort, causing the rocket to stand down for the day.
ULA’s Delta IV Heavy launch is still on the books for later today, and if successful, could mark the start of a fantastic day of rocket launches.
Hayneedle, the home furnishings retailer that joined Walmart by way of an acquisition by Jet back in 2016, is now serving as the testing grounds for Walmart’s own in-house visual search technology. Built in collaboration between Hayneedle and Walmart Labs, the new technology has been slowly scaling up on Hayneedle over the course of the year.
Hayneedle had been offering visual search on its mobile app since 2017, but it was doing so through a partnership with third-party vendor Slyce.
To use the feature, customers take an image of a piece of furniture they like by snapping a photo in someone’s home or in a retail showroom, for example, or even taking a photo of a page in a magazine. The app then uses that image to find similar products from Hayneedle.com – much like how visual search works on other sites, like Pinterest or Google Shopping.
Walmart says it sees the potential in visual search because of how it can better guide customers to products, even when they don’t know what they’re looking for by name.
“We’re continually focused on improving the customer experience,” says Shelley Huff, Vice President of Home, Walmart U.S. eCommerce and President of Hayneedle. “And I think in the home industry that’s particularly challenging just with the breadth of offering of the products, and the fact that the majority of customers don’t really know what they’re looking for in terms of home furnishings. It requires, in many cases, a large level of inspiration,” she continues.
“If we look at how we can enable that shopping experience…and give them more confidence in their purchases and their ability to find products, visual search place and incredible role,” Huff says.
Visual search can also help the retailer with downstream effects, like reducing returns and boosting overall ratings for products, leading to more customer satisfaction with their purchases, she notes.
Walmart had been working on its own visual search technology before Hayneedle, but was having trouble launching it at Walmart’s scale. That’s where Hayneedle comes in.
After its acquisition by Jet, Hayneedle discovered there were a lot of technologies – including those in the realms of machine learning and A.I. – that it could take advantage of from Walmart and Walmart Labs. It could also launch these more quickly and easily because it’s a smaller organization from an engineering perspective.
In the case of visual search, Hayneedle partnered with a team at Walmart Labs, which worked on the backend machine learning pieces for the new platform. On its side, Hayneedle provided the site’s product catalog of over a million SKUs, and trained the system. It then integrated the visual search technology with its own site.
The relationship proved to be mutually beneficial, explains Hayneedle’s Head of Engineering, Benjamin Dekarske.
“We don’t have the dedicated core technologists that Walmart has with Labs, and Walmart gains from Hayneedle a platform [for visual search] that they don’t have to scale to Walmart’s scale – they can try it out at Hayneedle’s scale, then learn from and grow from there,” he says.
This was also the first time the Walmart Labs team worked with one of Walmart’s acquisitions, in terms of collaborating on the development of new technology together, rather than porting over technology from one place to another.
While visual search had already been in use at Hayneedle via Slyce, Hayneedle’s test of Walmart’s own visual search technology began this March.
It has since been running A/B tests between the two technologies to compare performance and results between the two, says Dekarske. The company continues to have an agreement with Slyce, but it has been shifting more of Hayneedle’s visual search results to its internally-sourced effort over time.
Walmart won’t say if it’s preparing to end its contract with Slyce, as a decision hasn’t been formally made.
Visual search is not the only technology Hayneedle is testing that’s meant to one day scale up to Walmart.com after testing on a smaller site. However, the others in the works are not necessarily customer-facing, and include efforts around customer profiling, segmentation, and those focused on improving operations.
This isn’t the only area where Walmart has visual search in use. The retailer just snapped up another e-commerce site, Art.com, that offers its own visual search technology.
It’s now preparing to look at Art.com’s implementation, the company says, to further refine its existing efforts.
Hayneedle, the home furnishings retailer that joined Walmart by way of an acquisition by Jet back in 2016, is now serving as the testing grounds for Walmart’s own in-house visual search technology. Built in collaboration between Hayneedle and Walmart Labs, the new technology has been slowly scaling up on Hayneedle over the course of the year.
Hayneedle had been offering visual search on its mobile app since 2017, but it was doing so through a partnership with third-party vendor Slyce.
To use the feature, customers take an image of a piece of furniture they like by snapping a photo in someone’s home or in a retail showroom, for example, or even taking a photo of a page in a magazine. The app then uses that image to find similar products from Hayneedle.com – much like how visual search works on other sites, like Pinterest or Google Shopping.
Walmart says it sees the potential in visual search because of how it can better guide customers to products, even when they don’t know what they’re looking for by name.
“We’re continually focused on improving the customer experience,” says Shelley Huff, Vice President of Home, Walmart U.S. eCommerce and President of Hayneedle. “And I think in the home industry that’s particularly challenging just with the breadth of offering of the products, and the fact that the majority of customers don’t really know what they’re looking for in terms of home furnishings. It requires, in many cases, a large level of inspiration,” she continues.
“If we look at how we can enable that shopping experience…and give them more confidence in their purchases and their ability to find products, visual search place and incredible role,” Huff says.
Visual search can also help the retailer with downstream effects, like reducing returns and boosting overall ratings for products, leading to more customer satisfaction with their purchases, she notes.
Walmart had been working on its own visual search technology before Hayneedle, but was having trouble launching it at Walmart’s scale. That’s where Hayneedle comes in.
After its acquisition by Jet, Hayneedle discovered there were a lot of technologies – including those in the realms of machine learning and A.I. – that it could take advantage of from Walmart and Walmart Labs. It could also launch these more quickly and easily because it’s a smaller organization from an engineering perspective.
In the case of visual search, Hayneedle partnered with a team at Walmart Labs, which worked on the backend machine learning pieces for the new platform. On its side, Hayneedle provided the site’s product catalog of over a million SKUs, and trained the system. It then integrated the visual search technology with its own site.
The relationship proved to be mutually beneficial, explains Hayneedle’s Head of Engineering, Benjamin Dekarske.
“We don’t have the dedicated core technologists that Walmart has with Labs, and Walmart gains from Hayneedle a platform [for visual search] that they don’t have to scale to Walmart’s scale – they can try it out at Hayneedle’s scale, then learn from and grow from there,” he says.
This was also the first time the Walmart Labs team worked with one of Walmart’s acquisitions, in terms of collaborating on the development of new technology together, rather than porting over technology from one place to another.
While visual search had already been in use at Hayneedle via Slyce, Hayneedle’s test of Walmart’s own visual search technology began this March.
It has since been running A/B tests between the two technologies to compare performance and results between the two, says Dekarske. The company continues to have an agreement with Slyce, but it has been shifting more of Hayneedle’s visual search results to its internally-sourced effort over time.
Walmart won’t say if it’s preparing to end its contract with Slyce, as a decision hasn’t been formally made.
Visual search is not the only technology Hayneedle is testing that’s meant to one day scale up to Walmart.com after testing on a smaller site. However, the others in the works are not necessarily customer-facing, and include efforts around customer profiling, segmentation, and those focused on improving operations.
This isn’t the only area where Walmart has visual search in use. The retailer just snapped up another e-commerce site, Art.com, that offers its own visual search technology.
It’s now preparing to look at Art.com’s implementation, the company says, to further refine its existing efforts.
As Uber continues to expand the footprint of JUMP bikes, it’s aiming to make the fleet of bikes smarter and easier to unlock, ride and charge. At Uber’s Pier 70 offices in San Francisco, JUMP Head of Product Nick Foley showed off the new bikes, which are aesthetically similar but should improve unit economics thanks to self-diagnostic capabilities and swappable batteries. These bikes, Foley said, have been years in the making.
The 4G-enabled bikes also ditch the clunky rear keypad for a sleek front screen, swap the U-lock for a retractable cable lock, and now have a phone mount for turn-by-turn navigation. JUMP is gearing up to widely deploy these bikes in January, but there are already some on the streets of San Francisco.
To the rider, the most obvious difference will be the interface on the front where you scan a QR code or tap your phone or RFID card to unlock the bike.
“We did a lot of research about how people were interacting with the bike and streamlined the whole interaction flow to be a lot simpler,” Foley said.
Riders will also notice the absence of a steel U-lock bar in exchange for the retractable cable. Riders may also notice the swappable batteries, which will help with vehicle availability, but this feature is especially useful to Uber and its unit economics for JUMP bikes.
“That is a major improvement to system utilization, the operating system, fleet uptime and all of the most critical metrics about how businesses are performing with running a shared fleet,” Foley said. “Swappable batteries mean you don’t have to take vehicles back to wherever you charge a bike or scooter, and that’s good for the business.”
Also, this custom-engineered battery sends diagnostic information to the central unit, which lives near the bike’s handlebars, and then sends that to JUMP’s cloud. That central unit can also monitor the motor system, locking system and drive system, and then send that information to the cloud. From there, the bike can self-diagnose many issues and ultimately reduce the amount of time these bikes spend in warehouses awaiting repairs.
“It’s by building this whole platform to be highly intelligent, that we’re able to more efficiently operate these vehicles,” Foley said. “And some things that previously are users might have had to report to us, we can now self-diagnose. And we can actually update the drive parameters — like how the bike responds to pedaling — in real time. It can also do things, like, if you a move a bike between geographies where the speed limit is 20 miles per hour to 15 miles per hour, it’ll automatically pull down new settings from the cloud and operate in a way that the city says it’s supposed to operate.”
Since JUMP first launched in San Francisco earlier this year, people have ridden millions of miles on the bikes, which have since expanded to dozens of cities. Version one of these bikes were relatively durable, but Foley said these new ones should last for several years — give or take a few component changes.
“We saw more aggressive abuse than expected in certain areas,” Foley said. “But one of the major things has been putting bikes in and out of vans. That’s been a really tough moment for bicycles because they’re getting thrown all over the place, so swappable batteries are really going to extend the lifespan.”
It’s worth noting that as JUMP deploys version two of its bikes, it’s not going to sunset the hardware from version one. Instead, JUMP plans to simply upgrade the hardware with the new lock module and new system intelligence.
Meanwhile, Uber also operates an electric scooter service under the JUMP brand. In October, Uber deployed JUMP scooters in Santa Monica, along with bikes, in Santa Monica, Calif.
“The most exciting thing about all the pieces of technology that exists is that they’re all fundamentally valuable, whether on a bicycle or a scooter,” Foley said. “And so we’re building scooter hardware that leverages a lot of these pieces of technology.”
Foley said there’s more to come next year, but said the motor system, tracking and user interface are universal pieces of technology that JUMP can bring to its scooters. And as Uber gets closer to its Q1 2019 initial public offering, improving its unit economics across all of its businesses is going to be key.