Wij willen met u aan tafel zitten en in een openhartig gesprek uitvinden welke uitdagingen en vragen er bij u spelen om zo, gezamelijk, tot een beste oplossing te komen. Oftewel, hoe kan de techniek u ondersteunen in plaats van dat u de techniek moet ondersteunen.

Valkyrie Industries off-handedly refers to the current iteration of its VR suit as “Iron Man v. 1.” It’s a fitting reference. There’s a very “first half of the superhero film” vibe to the prototype. There are exposed wires everywhere and large, clunky 3D printed pieces that clip onto various body parts. In a more finalized version, it will probably look like something more akin to a wetsuit. For now, however, the wearable haptic product looks like a bit of steampunk cosplay.

We met with the London-based team at the Brinc accelerator in Hong Kong. I admit to being a bit wary at first mention of a haptic body suit for VR. We’ve seen a number of wearables throughout the years designed specifically to offer a more immersive gaming experience. Among the key places Valkyrie sets itself apart, however, is target market.

Rather than targeting the fairly limited world of VR gaming, however, the startup has its eyes on professional applications. This technology will almost certainly be cost prohibitive for the foreseeable future, making it something of a nonstarter for a majority of home users (the bill of materials for the current version is somewhere in the neighborhood of $ 1.5k). Big companies, on the other hand, would like be far more willing to invest in a technology that could simplify and streamline the training process, particularly for dangerous and otherwise complex positions.

The system utilizes electrical impulses to stimulate muscles, approximating resistance and touch. With the product still very much in the early stages (the three-person company is currently seed funded), we were unable to actually try out the product.

But Valkyrie has already demoed the product for a number of high profile companies and government industries, who are interested in the product for both training purposes and potential teleoperation, giving wearers the ability to control and manipulate objects at a safe distance.


TechCrunch

Five billion dollars. That’s the apparent size of Facebook’s latest fine for violating data privacy. 

While many believe the sum is simply a slap on the wrist for a behemoth like Facebook, it’s still the largest amount the Federal Trade Commission has ever levied on a technology company. 

Facebook is clearly still reeling from Cambridge Analytica, after which trust in the company dropped 51%, searches for “delete Facebook” reached 5-year highs, and Facebook’s stock dropped 20%.

While incumbents like Facebook are struggling with their data, startups in highly-regulated, “Third Wave” industries can take advantage by using a data strategy one would least expect: ethics. Beyond complying with regulations, startups that embrace ethics look out for their customers’ best interests, cultivate long-term trust — and avoid billion dollar fines. 

To weave ethics into the very fabric of their business strategies and tech systems, startups should adopt “agile” data governance systems. Often combining law and technology, these systems will become a key weapon of data-centric Third Wave startups to beat incumbents in their field. 

Established, highly-regulated incumbents often use slow and unsystematic data compliance workflows, operated manually by armies of lawyers and technology personnel. Agile data governance systems, in contrast, simplify both these workflows and the use of cutting-edge privacy tools, allowing resource-poor startups both to protect their customers better and to improve their services.

In fact, 47% of customers are willing to switch to startups that protect their sensitive data better. Yet 80% of customers highly value more convenience and better service. 

By using agile data governance, startups can balance protection and improvement. Ultimately, they gain a strategic advantage by obtaining more data, cultivating more loyalty, and being more resilient to inevitable data mishaps. 

Agile data governance helps startups obtain more data — and create more value 

With agile data governance, startups can address their critical weakness: data scarcity. Customers share more data with startups that make data collection a feature, not a burdensome part of the user experience. Agile data governance systems simplify compliance with this data practice. 

Take Ally Bank, which the Ponemon Institute rated as one of the most privacy-protecting banks. In 2017, Ally’s deposits base grew 16%, while those of incumbents declined 4%.

One key principle to its ethical data strategy: minimizing data collection and use. Ally’s customers obtain services through a personalized website, rarely filling out long surveys. When data is requested, it’s done in small doses on the site — and always results in immediate value, such as viewing transactions. 

This is on purpose. Ally’s Chief Marketing Officer publicly calls the industry-mantra of “more data” dangerous to brands and consumers alike.

A critical tool to minimize data use is to use advanced data privacy tools like differential privacy. A favorite of organizations like Apple, differential privacy limits your data analysts’ access to summaries of data, such as averages. And by injecting noise into those summaries, differential privacy creates provable guarantees of privacy and prevents scenarios where malicious parties can reverse-engineer sensitive data. But because differential privacy uses summaries, instead of completely masking the data, companies can still draw meaning from it and improve their services. 

With tools like differential privacy, organizations move beyond governance patterns where data analysts either gain unrestricted access to sensitive data (think: Uber’s controversial “god view”) or face multiple barriers to data access. Instead, startups can use differential privacy to share and pool data safely, helping them overcome data scarcity. The most agile data governance systems allow startups to use differential privacy without code and the large engineering teams that only incumbents can afford.

Ultimately, better data means better predictions — and happier customers.

Agile data governance cultivates customer loyalty

According to Deloitte, 80% of consumers are more loyal to companies they believe protect their data. Yet far fewer leaders at established, incumbent companies — the respondents of the same survey — believed this to be true. Customers care more about their data than the leaders at incumbent companies think. 

This knowledge gap is an opportunity for startups. 

Furthermore, big enterprise companies — themselves customers of many startups — say data compliance risks prevent them from working with startups. And rightly so. Over 80% of data incidents are actually caused by errors from insiders, like third party vendors who mishandle sensitive data by sharing it with inappropriate parties. Yet over 68% of companies do not have good systems to prevent these types of errors. In fact, Facebook’s Cambridge Analytica firestorm — and resulting $ 5 billion fine — was sparked by third party inappropriately sharing personal data with a political consulting firm without user consent. 

As a result, many companies — both startups and incumbents — are holding a ticking time bomb of customer attrition. 

Agile data governance defuses these risks by simplifying the ethical data practices of understanding, controlling, and monitoring data at all times. With such practices, startups can prevent and correct the mishandling of sensitive data quickly.

Cognoa is a good example of a Third Wave healthcare startup adopting these three practices at a rapid pace. First, it understands where all of its sensitive health data lies by connecting all of its databases. Second, Cognoa can control all connected data sources at once from one point by using a single access-and-control layer, as opposed to relying on data silos. When this happens, employees and third parties can only access and share the sensitive data sources they’re supposed to. Finally, data queries are always monitored, allowing Cognoa to produce audit reports frequently and catch problems before they escalate out of control. 

With tools that simplify these three practices, even low-resourced startups can make sure sensitive data is tightly controlled at all times to prevent data incidents. Because key workflows are simplified, these same startups can maintain the speed of their data analytics by sharing data safely with the right parties. With better and safer data sharing across functions, startups can develop the insight necessary to cultivate a loyal fan base for the long-term.

Agile data governance can help startups survive inevitable data incidents

In 2018, Panera mistakenly shared 37 million customer records on its website and took 8 months to respond. Panera’s data incident is a taste of what’s to come: Gartner predicts that 50% of business ethics violations will stem from data incidents like these. In the era of “Big Data,” billion dollar incumbents without agile data governance will likely continue to violate data ethics. 

Given the inevitability of such incidents, startups that adopt agile data governance will likely be the most resilient companies of the future. 

Case in point: Harvard Business Review reports that the stock prices of companies without strong data governance practices drop 150% more than companies that do adopt strong practices. Despite this difference, only 10% of Fortune 500 companies actually employ the data transparency principle identified in the report. Practices include clearly disclosing data practices and giving users control over their privacy settings. 

Sure, data incidents are becoming more common. But that doesn’t mean startups don’t suffer from them. In fact, up to 60% of startups fold after a cyber attack. 

Startups can learn from WebMD, which Deloitte named as one standout in applying data transparency. With a readable privacy policy, customers know how data will be used, helping customers feel comfortable about sharing their data. More informed about the company’s practices, customers are surprised less by incidents. Surprises, BCG found, can reduce consumer spending by one-third. On a self-service platform on WebMD’s site, customers can control their privacy settings and how to share their data, further cultivating trust. 

Self-service tools like WebMD’s are part of agile data governance. These tools allow startups to simplify manual processes, like responding to customer requests to control their data. Instead, startups can focus on safely delivering value to their customers. 

Get ahead of the curve

For so long, the public seemed to care less about their data. 

That’s changing. Senior executives at major companies have been publicly interrogated for not taking data governance seriously. Some, like Facebook and Apple, are even claiming to lead with privacy. Ultimately, data privacy risks significantly rise in Third Wave industries where errors can alter access to key basic needs, such as healthcare, housing, and transportation.

While many incumbents have well-resourced legal and compliance departments, agile data governance goes beyond the “risk mitigation” missions of those functions. Agile governance means that time-consuming and error-prone workflows are streamlined so that companies serve their customers more quickly and safely.

Case in point: even after being advised by an army of lawyers, Zuckerberg’s 30,000-word Senate testimony about Cambridge Analytica included “ethics” only once, and it excluded “data governance” completely.

And even if companies do have legal departments, most don’t make their commitment to governance clear. Less than 15% of consumers say they know which companies protect their data the best. Startups can take advantage of this knowledge gap by adopting agile data governance and educate their customers about how to protect themselves in the risky world of the Third Wave.

Some incumbents may always be safe. But those in highly-regulated Third Wave industries, such as automotive, healthcare, and telecom should be worried; customers trust these incumbents the least. Startups that adopt agile data governance, however, will be trusted the most, and the time to act is now. 


TechCrunch

The Google News tab is getting a makeover. Google announced this week, by way of a tweet, a significant redesign of the Google.com News tab on the desktop, which will organize articles in a card-style layout, while also better emphasizing publisher names. The end result makes Google News more aesthetically pleasing, but it comes at the expense of information density.

To be clear, the changes here are focused on the News tab of Google.com — not the dedicated Google News product at news.google.com. You land on the News tab when you search for a term on Google.com, and then click over to “News” to see the latest coverage instead of Google’s list of search results.

As the preview of the redesign shows, news articles are currently organized in a compact list of links, allowing you to see several headlines around a single topic with just a glance. This design, admittedly, is a bit old-school — but it works.

Within the stack of links, the headline is blue, the publisher is green, and the articles are labeled as “In-depth” or “Opinion,” when relevant. There are small photo thumbnails by the lead story, with other publishers’ links underneath appearing as only text.

Screen Shot 2019 07 12 at 11.16.03 AM

 

The updated design is more readable as articles are spaced out and placed in cards, similar to the main Google News product. There’s more white space and longer previews of each story, as well.

But the change means you’re seeing far fewer results on the screen before you have to scroll down.

 

The updated News tab makes it more obvious where the news is coming from, because publishers’ names are given more prominence. They also get their logo next to the headline, so it’s easier to identify your favorite news outlets with a glance. This is reminiscent of the recent mobile redesign for Google Search, which also put increased attention on the publishers by featuring them at the top of a link alongside their logo.

In addition to providing you with a set of News search results, the redesigned tab includes a new carousel labeled “People also searched for” that points you to other relevant news based on your search query.

Not everyone is thrilled about the update, given it makes it more difficult to quickly scan a number of headlines at once. And because there are fewer publishers’ articles on the first screen, traffic to those “below the fold” will likely drop.

Google says the changes will roll out over the next couple of weeks.


TechCrunch

The machines have proven their superiority in one-on-one games like chess and go, and even poker — but in complex multiplayer versions of the card game humans have retained their edge… until now. An evolution of the last AI agent to flummox poker pros individually is now decisively beating them in championship-style 6-person game.

As documented in a paper published in the journal Science today, the CMU/Facebook collaboration they call Pluribus reliably beats five professional poker players in the same game, or one pro pitted against five independent copies of itself. It’s a major leap forward in capability for the machines, and amazingly is also far more efficient than previous agents as well.

One-on-one poker is a weird game, and not a simple one, but the zero-sum nature of it (whatever you lose, the other player gets) makes it susceptible to certain strategies in which computer able to calculate out far enough can put itself at an advantage. But add four more players into the mix and things get real complex, real fast.

With six players, the possibilities for hands, bets, and possible outcomes are so numerous that it is effectively impossible to account for all of them, especially in a minute or less. It’d be like trying to exhaustively document every grain of sand on a beach between waves.

Yet over 10,000 hands played with champions, Pluribus managed to win money at a steady rate, exposing no weaknesses or habits that its opponents could take advantage of. What’s the secret? Consistent randomness.

Even computers have regrets

Pluribus was trained, like many game-playing AI agents these days, not by studying how humans play but by playing against itself. At the beginning this is probably like watching kids, or for that matter me, play poker — constant mistakes, but at least the AI and the kids learn from them.

The training program used something called Monte Carlo counterfactual regret minimization. Sounds like when you have whiskey for breakfast after losing your shirt at the casino, and in a way it is — machine learning style.

Regret minimization just means that when the system would finish a hand (against itself, remember), it would then play that hand out again in different ways, exploring what might have happened had it checked here instead of raised, folded instead of called, and so on. (Since it didn’t really happen, it’s counterfactual.)

A Monte Carlo tree is a way of organizing and evaluating lots of possibilities, akin to climbing a tree of them branch by branch and noting the quality of each leaf you find, then picking the best one once you think you’ve climbed enough.

If you do it ahead of time (this is done in chess, for instance) you’re looking for the best move to choose from. But if you combine it with the regret function, you’re looking through a catalog of possible ways the game could have gone and observing which would have had the best outcome.

So Monte Carlo counterfactual regret minimization is just a way of systematically investigating what might have happened if the computer had acted differently, and adjusting its model of how to play accordingly.

traverserj

The game originall played out as you see on the left, with a loss. But the engine explores other avenues where it might have done better.

Of course the number of games is nigh-infinite if you want to consider what would happen if you had bet $ 101 rather than $ 100, or you would have won that big hand if you’d had an eight kicker instead of a seven. Therein also lies nigh-infinite regret, the kind that keeps you in bed in your hotel room until past lunch.

The truth is these minor changes matter so seldom that the possibility can basically be ignored entirely. It will never really matter that you bet an extra buck — so any bet within, say, 70 and 130 can be considered exactly the same by the computer. Same with cards — whether the jack is a heart or a spade doesn’t matter except in very specific (and usually obvious) situations, so 99.999 percent of the time the hands can be considered equivalent.

This “abstraction” of gameplay sequences and “bucketing” of possibilities greatly reduces the possibilities Pluribus has to consider. It also helps keep the calculation load low; Pluribus was trained on a relatively ordinary 64-core server rack over about a week, while other models might take processor-years in high-power clusters. It even runs on a (admittedly beefy) rig with two CPUs and 128 gigs of RAM.

Random like a fox

The training produces what the team calls a “blueprint” for how to play that’s fundamentally strong and would probably beat plenty of players. But a weakness of AI models is that they develop tendencies that can be detected and exploited.

In Facebook’s writeup of Pluribus, it provides the example of two computers playing rock-paper-scissors. One picks randomly while the other always picks rock. Theoretically they’d both win the same amount of games. But if the computer tried the all-rock strategy on a human, it would start losing with a quickness and never stop.

As a simple example in poker, maybe a particular series of bets always makes the computer go all in regardless of its hand. If a player can spot that series, they can take the computer to town any time they like. Finding and preventing ruts like these is important to creating a game-playing agent that can beat resourceful and observant humans.

To do this Pluribus does a couple things. First, it has modified versions of its blueprint to put into play should the game lean towards folding, calling, or raising. Different strategies for different games mean it’s less predictable, and it can switch in a minute should the bet patterns change and the hand go from a calling to a bluffing one.

It also engages in a short but comprehensive introspective search looking at how it would play if it had every other hand, from a big nothing up to a straight flush, and how it would bet. It then picks its bet in the context of all those, careful to do so in such a way that it doesn’t point to any one in particular. Given the same hand and same play again, Pluribus wouldn’t choose the same bet, but rather vary it to remain unpredictable.

These strategies contribute to the “consistent randomness” I alluded to earlier, and which were a part of the model’s ability to slowly but reliably put some of the best players in the world.

The human’s lament

There are too many hands to point to a particular one or ten that indicate the power Pluribus was bringing to bear on the game. Poker is a game of skill, luck, and determination, and one where winners emerge after only dozens or hundreds of hands.

And here it must be said that the experimental setup is not entirely reflective of an ordinary 6-person poker game. Unlike a real game, chip counts are not maintained as an ongoing total — for every hand, each player was given 10,000 chips to use as they pleased, and win or lose they were given 10,000 in the next hand as well.

interface

The interface used to play poker with Pluribus. Fancy!

Obviously this rather limits the long-term strategies possible, and indeed “the bot was not looking for weaknesses in its opponents that it could exploit,” said Facebook AI research scientist Noam Brown. Truly Pluribus was living in the moment the way few humans can.

But simply because it was not basing its play on long-term observations of opponents’ individual habits or styles does not mean that its strategy was shallow. On the contrary, it is arguably more impressive, and casts the game in a different light, that a winning strategy exists that does not rely on behavioral cues or exploitation of individual weaknesses.

The pros who had their lunch money taken by the implacable Pluribus were good sports, however. They praised the system’s high level play, its validation of existing techniques, and inventive use of new ones. Here’s a selection of laments from the fallen humans:

I was one of the earliest players to test the bot so I got to see its earlier versions. The bot went from being a beatable mediocre player to competing with the best players in the world in a few weeks. Its major strength is its ability to use mixed strategies. That’s the same thing that humans try to do. It’s a matter of execution for humans — to do this in a perfectly random way and to do so consistently. It was also satisfying to see that a lot of the strategies the bot employs are things that we do already in poker at the highest level. To have your strategies more or less confirmed as correct by a supercomputer is a good feeling. -Darren Elias

It was incredibly fascinating getting to play against the poker bot and seeing some of the strategies it chose. There were several plays that humans simply are not making at all, especially relating to its bet sizing. -Michael ‘Gags’ Gagliano

Whenever playing the bot, I feel like I pick up something new to incorporate into my game. As humans I think we tend to oversimplify the game for ourselves, making strategies easier to adopt and remember. The bot doesn’t take any of these short cuts and has an immensely complicated/balanced game tree for every decision. -Jimmy Chou

In a game that will, more often than not, reward you when you exhibit mental discipline, focus, and consistency, and certainly punish you when you lack any of the three, competing for hours on end against an AI bot that obviously doesn’t have to worry about these shortcomings is a grueling task. The technicalities and deep intricacies of the AI bot’s poker ability was remarkable, but what I underestimated was its most transparent strength – its relentless consistency. -Sean Ruane

Beating humans at poker is just the start. As good a player as it is, Pluribus is more importantly a demonstration that an AI agent can achieve superhuman performance at something as complicated as 6-player poker.

“Many real-world interactions, such as financial markets, auctions, and traffic navigation, can similarly be modeled as multi-agent interactions with limited communication and collusion among participants,” writes Facebook in its blog.

Yes, and war.


TechCrunch

Google has responded to a report this week from Belgian public broadcaster VRT NWS, which revealed that contractors were given access to Google Assistant voice recordings, including those which contained sensitive information — like addresses, conversations between parents and children, business calls, and others containing all sorts of private information. As a result of the report, Google says it’s now preparing to investigate and take action against the contractor who leaked this information to the news outlet.

The company, by way of a blog post, explained that it partners with language experts around the world who review and transcribe a “small set of queries” to help Google better understand various languages.

Only around 0.2 percent of all audio snippets are reviewed by language experts, and these snippets are not associated with Google accounts during the review process, the company says. Other background conversations or noises are not supposed to be transcribed.

The leaker had listened to over 1,000 recordings, and found 153 were accidental in nature — meaning, it was clear the user hadn’t intended to ask for Google’s help. In addition, the report found that determining a user’s identity was often possible because the recordings themselves would reveal personal details. Some of the recordings contained highly sensitive information, like “bedroom conversations,” medical inquiries, or people in what appeared to be domestic violence situations, to name a few.

Google defended the transcription process as being a necessary part of providing voice assistant technologies to its international users.

But instead of focusing on its lack of transparency with consumers over who’s really listening to their voice data, Google says it’s going after the leaker themselves.

“[Transcription] is a critical part of the process of building speech technology, and is necessary to creating products like the Google Assistant,” writes David Monsees, Product Manager for Search at Google, in the blog post. “We just learned that one of these language reviewers has violated our data security policies by leaking confidential Dutch audio data. Our Security and Privacy Response teams have been activated on this issue, are investigating, and we will take action. We are conducting a full review of our safeguards in this space to prevent misconduct like this from happening again,” he said.

As voice assistant devices are becoming a more common part of consumers’ everyday lives, there’s increased scrutiny on how tech companies are handline the voice recordings, who’s listening on the other end, what records are being stored, and for how long, among other things.

This is not an issue that only Google is facing.

Earlier this month, Amazon responded to a U.S. senator’s inquiry over how it was handling consumers’ voice records. The inquiry had followed a CNET investigation which discovered Alexa recordings were kept unless manually deleted by users, and that some voice transcripts were never deleted. In addition, a Bloomberg report recently found that Amazon workers and contractors during the review process had access to the recordings, as well as an account number, the user’s first name, and the device’s serial number.

Further, a coalition of consumer privacy groups recently lodged a complaint with the U.S. Federal Trade Commission which claims Amazon Alexa is violating the U.S. Children’s Online Privacy Protection Act (COPPA) by failing to obtain proper consent over the company’s use of the kids’ data.

Neither Amazon nor Google have gone out of their way to alert consumers as to how the voice recordings are being used.

As Wired notes, the Google Home privacy policy doesn’t disclose that Google is using contract labor to review or transcribe audio recordings. The policy also says that data only leaves the device when the wake word is detected. But these leaked recordings indicate that’s clearly not true — the devices accidentally record voice data at times.

The issues around the lack of disclosure and transparency could be yet another signal to U.S. regulators that tech companies aren’t able to make responsible decisions on their own when it comes to consumer data privacy.

The timing of the news isn’t great for Google. According to reports, the U.S. Department of Justice is preparing for a possible antitrust investigation of Google’s business practices, and is watching the company’s behavior closely. Given this increased scrutiny, one would think Google would be going over its privacy policies with a fine-toothed comb — especially in areas that are newly coming under fire, like policies around consumers’ voice data — to ensure that consumers understand how their data is being stored, shared, and used.

Google also notes today that people do have a way to opt-out of having their audio data stored. Users can either turn off audio data storage entirely, or choose to have the data auto-delete every 3 months or every 18 months.

The company also says it will work to better explain how this voice data is used going forward.

“We’re always working to improve how we explain our settings and privacy practices to people, and will be reviewing opportunities to further clarify how data is used to improve speech technology,” said Monsees.


TechCrunch

Rivigo, a tech startup in India that wants to build a more reliable and safer logistics network, has raised $ 65 million as major investors continue to place big bet on opportunities in overhauling trucking system in the country.

The Series E round, which has not closed, for the five-year-old startup was led by existing investors Warburg Pincus and SAIF Partners.  The startup, which has raised more than $ 280 million to date, said it aims to be profitable by March next year.

Rivigo operates a tech platform that tracks and manages shipments and ensures that drivers are available at all times and trucks are as fully loaded as possible. The platform also automatically rotates drivers so that they can get enough rest and see their family while the trucks keep moving. Drivers use an app to navigate maps and accept assignments.

“Relay trucking is now very well established where relay truck pilots lead better life and customers gets exceptional service. With technology and freight marketplace, we now want to bring relay to every truck in the country,” Deepak Garg, founder and CEO of Rivigo, said in a statement.

Rivigo, which competes with heavily-backed startups such as BlackBuck, owns its own fleet of trucks while also operating a freight marketplace. This separates it from competitors that serve purely as an aggregator — or Uber for trucks, if you will.

The startup, which claims to have the largest reach in India, said it would use the fresh capital to further expand its network and tech infrastructure in the country. Financially, too, Rivigo has driven past many of its competitors. In the financial year that ended in March this year, Rivigo’s revenue jumped to $ 105 million at a 77% year-over-year growth rate. Its losses also widened to $ 35 million, according to disclosures it made to the local regulator.

“From building algorithmically complex models to accurately predicting the life journey of a consignment to creating a dynamic pricing engine for the freight marketplace, the company is working on hundreds of unique problems at scale,” said Garg.

India’s logistics market, despite being valued at $ 160 billion, remains one of the most inefficient sectors that continues to drag the economy.

Last month, Rivigo launched National Freight Index that shows live tariff rates for different lanes and vehicles in the country in a bid to bring more transparency to the ecosystem.

More to follow later today…


TechCrunch

After generally being the butt of the public market’s jokes since its IPO, Snap is having a killer 2019, with its stock price nearly tripling in value. The successes are perhaps giving the company a moment to pause and think more about generating future value.

Part of that equation is certainly the company’s Yellow accelerator that aims to invest in pre-seed startups that bring mobile users to shared experiences.

We covered Yellow’s inaugural batch back in September, now we’ve got the full rundown on Snap’s second class of bets.

Yellow’s latest accelerator class definitely showcases some similarities to their inaugural group, but you’ll notice more online-to-offline startups aiming to bring users into real-world scenarios and communities like a concert subscription service and workout service reviews. This contrasts a bit to the first class which seemed a bit more focused on camera-based startups that centered around selfies, AR and photos.

From an organizational standpoint, things haven’t shifted too much inside Yellow. The broader company has had a standout 2019, building back a healthy chunk of the market cap value it has lost since debuting publicly. One wonders whether this has enabled the company’s accelerator group to push its investment ambitions beyond Snap’s mobile app focus.

Mike Su, Snap’s director of Yellow, tells me that there haven’t been any top-down directives to shift investment strategies for the accelerator and that the prevalence of offline startups in the class is just more representative of the applicants.

“[The class] continues to be an extension of our values and our thesis,” Su tells me. “Snap has always been about people making connections inside and outside the app.”

Here is Yellow’s summer 2019 class of startups.

Active Spaces

ClassPass might toss you in a random workout and say good luck, but Active Spaces is looking to give you more info when searching for your exercising fix. The New York startup is scouring its way through the NYC reviewing gyms and studios one-at-a-time. It’s less about star ratings than it is about giving you a bird’s eye view of what’s there and what’s missing. It’s all really well-done and gives you a ton of info about what you’re in for, and you can book direct from the app.

casino royale2

Cash Live

HQ Trivia might be falling on hard times but Cash Live is looking to take the daily mobile quiz show in a new direction by leaning on the laurels of gaming, some good ole fashioned casino titles. The Vancouver startup is planning to bring a live host to scheduled 15-minute poker, blackjack and bingo tournaments.

AFP PHOTO / ANGELA WEISS

Disko

Finding local concerts sucks and it’s a process that hasn’t found its startup solution yet. Disko is building a concert subscription service that helps users discover new events in their city with a flat rate $ 25 per month subscription service which will let users attend up to four concerts per month. The LA startup is starting off in its hometown but has ambitions to expand elsewhere soon.

image16 800x500

Dose of Society

We’re missing a lot of diversity in the voices and perspectives we see in the media we enjoy. Dose of Society is a London media startup looking to share “real stories from real people.” The group’s videos have had more than 18 million views since launching at the end of 2017.

Screen Shot 2019 07 10 at 11.28.05 AM

Frame

Snap still has vertical video startups firmly in its purview. Frame is a weekly newsmagazine built for mobile that’s trying to rethink how we get news delivered to us. The NY startup is looking beyond push notifications and is also supporting text updates and calendar updates so that its subscribers can make time to absorb its narrative vertical video  journalism.

Screen Shot 2019 07 10 at 11.44.14 AM

Loco Adventures

Pokémon GO brought people into physical spaces with its location-based gaming, but other startups are seeing the potential to even further localize AR experiences. Berlin-based Loco Adventures is building games that guide you through local areas with a chat message narrative style.

Muze

Muze sees the endless wave of comments on the web and wants to make things a bit noisier, the New York team is working on a way to bring audio commentary “to the always-on stream of internet video” and share it across the web.

ROBYN BECK/AFP/Getty Images

Quirktastic, Inc.

The startup has the ambitious goal of building a community for “geeks, gamers and nerds” that’s less toxic to minority groups. The Durham, NC company wants to connect these people with each other and the events they want to check out. Quirktastic says they have 15,000 users since they launched in beta in March.

nike

SNKRHUD

The sneaker business is a hefty one, but SNKRHUD is betting that it still isn’t as big as it could be. It’s trying to focus on the dormant sneaker heads who are liking shoes on Instagram and searching through online stores but haven’t delved further into communities. The Brooklyn team wants to be the glue between existing platforms.

Photo: Thomas Barwick/Getty Images

Stop, Breathe & Think

There’s a lot in the world to get stressed and anxious about, Stop, Breathe & Think is aiming to build a digital wellness platform to help people feel better. The app lets people check-in with how they’re feeling and then the app is able to recommend short activities like meditation, breathing, yoga, acupressure, guided journaling, and more.


TechCrunch

Created by R the Company. Powered by SiteMuze.