On Tuesday, Tencent Music Entertainment filed for an IPO in the US that is expected to value it in the $25-30 billion range, on par with Spotify’s IPO in April. The filing highlights just how different its social interaction and digital goods business is from the subscription models of leading music streaming services in Western countries.

That divergence suggests an opportunity for Spotify or one of its rivals to gain a competitive advantage.

Tencent Music is no small player: As the music arm of Chinese digital media giant Tencent, its four apps have several hundred million monthly active users, $1.3 billion in revenue for the first half of 2018, and roughly 75 percent market share in China’s rapidly growing music streaming market. Unlike Spotify and Apple Music, however, almost none of its users pay for the service, and those who do are mostly not paying in the form of a streaming subscription.

Its SEC filing shows that 70 percent of revenue is from the 4.2 percent of its overall users who pay to give virtual gifts to other users (and music stars) who sing karaoke or live stream a concert and/or who paid for access to premium tools for karaoke; the other 30 percent is the combination of streaming subscriptions, music downloads, and ad revenue.

At its heart, Tencent Music is an interactive media company. Its business isn’t merely providing music, it’s getting people to engage around music. Given its parent company Tencent has become the leading force in global gaming—with control of League of Legends maker Riot Games and Clash of Clans maker Supercell, plus a 40 percent stake in Fortnite creator Epic Games, and role as the top mobile games publisher in China—its team is well-versed in the dynamics of in-game purchasing.

At first glance, the fact that Tencent Music has a lower subscriber rate than its Western rivals (3.6 percent of users paying for a subscription or digital downloads vs. 46 percent paying for a premium subscription on Spotify) is shocking given it has the key ingredient they each crave: exclusive content. Whereas subscription video streaming services like Netflix, Hulu, and Amazon Prime Video have anchored themselves in exclusive ownership of must-see shows in order to attract subscribers, the music streaming platforms suffer from commodity content. Spotify, Apple Music, Amazon Music, YouTube Music, Pandora, iHeartRadio, Deezer… they all have the same core library of music licensed from the major labels. There’s no reason for any consumer to pay for more than one music streaming subscription in the way they do for video streaming services.

In China, however, Tencent Music has exclusive rights to the most popular Western music from the major labels. The natural strategy to leverage this asset would be to charge a subscription to access it. But the reality is that piracy is still enough of a challenge in China that access to that music isn’t truly “exclusive.” Plus while incomes are rising, there’s extraordinary variance in what price point the population can afford for a music subscription. As a result, Tencent Music can’t rely on a subscription for exclusive content; it sublicenses that content to other Chinese music services as an additional revenue stream instead.

“Online music services in China have experienced intense competition with limited ability to differentiate by content due to the widespread piracy.” Tencent Music, SEC Form F-1

This puts it in a position like that of the Western music streaming services—fighting to differentiate and build a moat against competitors—but unlike them it has successfully done so. By integrating live streams and social functionality as core to the user experience, it’s gaining exclusive content in another form (user-generated content) and the network effects of a social media platform.

Some elements of this are distinct to Tencent’s core market—the broader popularity of karaoke, for instance—but the strategy of gaining competitive advantage through interactive and live content is one Spotify and its rivals would be wise to pursue more aggressively. It is unlikely that the major record labels will agree to any meaningful degree of exclusivity for one of the big streaming services here, and so these platforms need to make unique experiences core to their offering.

Online social activities like singing with friends or singing a karaoke duet with a favorite musician do in fact have a solid base of participants around the world: San Francisco-based startup Smule (backed by Shasta Ventures and Tencent itself) has 50 million monthly active users on its apps for that very purpose. There is a large minority of people who care a lot about singing songs as a social experience, both with friends and strangers.

Spotify and Apple Music have experimented with video, messaging, and social streams (of what friends are listening to). But these have been bonus features and none of them were so integrated into the core product offering as to create serious switching costs that would stop a user from jumping to the other.

The ability to give tips or buy digital goods makes it easier to monetize a platform’s most engaged and enthusiastic users. This is the business model of the mobile gaming sector: A minority percentage of users get emotionally invested enough to pay real money for digital goods that enhance their experience, currency to tip other members of the community, or access to additional gameplay.

As the leading music platform, it is surprising that Spotify hasn’t created a pathway for superfans of music to engage deeper with artists or each other. Spotify makes referrals to buy concert tickets or merchandise —a very traditional sense of what the music fan wants—but hasn’t deepened the online music experience for the segment of its user base that would happily pay more for music-related experiences online (whether in the form of tipping, digital goods, special digital access to live shows, etc.) or for deeper exposure to the process (and people) behind their favorite songs.

Tencent Music has an advantage in creating social music experiences because it is part of the same company that owns the country’s leading social apps and is integrated into them. It has been able to build off the social graph of WeChat and QQ rather than building a siloed social network for music. Even Spotify’s main corporate rivals, Apple Music and Amazon Music, aren’t attached to leading social platforms. (Another competitor, YouTube Music, is tied to YouTube but the video service’s social features are secondary aspects of the product compared to the primary role of social interaction on Facebook, Instagram, and WhatsApp).

Spotify could build out more interactive products itself or could buy social-music startups like Smule, but Tencent Music’s success also suggests the benefits of a deal that’s sometimes speculated about by VCs and music industry observers: a Facebook acquisition of Spotify. As one, the leading social media company and the leading music streaming company could build out more valuable video live streaming, group music sharing, karaoke, and other social interactions around music that tap Facebook’s 2 billion users to use Spotify as their default streaming service and lock existing Spotify subscribers into the service that integrates with their go-to social apps.

Deeper social functionality doesn’t seem to be the path Spotify is prioritizing, though. It has removed several social features over the years and is anchoring itself in professional content distribution (rather than user-generated content creation), becoming the new pipes for professional musicians to put their songs out to the world (and likely aiming to disrupt the role of labels and publishers more than they will publicly admit). To that point, the company’s acquisitions—of startups like Loudr, Mediachain, and Soundtrap—have focused on content analytics, content recommendation, royalty tracking, and tools for professional creators.

This is the same race its more deep-pocketed competitors are running, however, and it doesn’t lock consumers into the platform like the network effects of a social app or the exclusivity of a mobile game do. It recently began opening its platform for musicians to add their songs directly—something Tencent Music has allowed for years—but this seems less like a move to a YouTube or SoundCloud-style user-generated content platform and more like a chess move in the game of eventually displacing labels. Ultimately, though, building out more social interaction around music will be critical to it in escaping the race with Apple Music and the rest by achieving more defensibility.

Backed with nearly $87 million in venture capital funding from GV, Oak HC/FT and F-Prime Capital, Quartet Health was founded in 2014 by Arun Gupta, Steve Shulman and David Wennberg to improve access to behavioral healthcare. Its mission: “enable every person in our society to thrive by building a collaborative behavioral and physical health ecosystem.”

Recent shakeups within the New York-based company’s c-suite and a perusal of its Glassdoor profile suggest Quartet’s culture is not fully in line with its own philosophy.  

In the last few weeks, chief product officer Rajesh Midha has left the company and president and chief operating officer David Liu is on his way out, TechCrunch has learned and confirmed with Quartet. Founding chief executive officer Arun Gupta, meanwhile, has stepped into the executive chairman role, relinquishing responsibility of the company’s day-to-day operations to former chief science officer David Wennberg, who’s taken over as CEO.

“I’m focusing on our external growth,” Gupta told TechCrunch on Friday. “David has really stepped up as CEO.”

Gupta and Wennberg said Liu’s role was no longer needed because Wennberg had assumed his responsibilities. Liu will formally exit the company at the end of the month. As for its product chief, the pair say Midha had “transitioned out” of the role and that an unnamed internal candidate was tapped to replace him.

When asked whether other employees had left in recent weeks,  Wennberg provided the following indeterminate statement: “We are always having people coming in. I don’t think we’ve had any unusual turnover. We’re hiring and people’s roles change and that’s just part of growth.”

Quartet, which provides a platform that allows providers to collaborate on treatment plans, currently has 150 employees, according to its executives.

In a LinkedIn status update published this week — after TechCrunch’s initial inquiries — Gupta announced his transition to executive chairman:

“Still full-time, though focused largely on our opportunity to further evangelize our mission, [I will] drive the change we want to see in this world, and expand our reach … I have tremendous confidence in David’s ability to lead our many talented Quartetians to deliver this next phase.”

Several former employees seemed less than pleased with Gupta’s performance, writing in a number of Glassdoor reviews that he was “abominable,” “kind of a monster” and “by far the worst executive.”

When asked for comment on those reviews, Gupta and Wennberg shrugged it off: “Glassdoor is Glassdoor.” They agreed its important to pay attention to but impossible to vet.

Gupta began his career as a management consultant at McKinsey and served as a consultant to The World Bank before joining Palantir, Peter Thiel’s data-mining company, as an advisor in 2014. Wennberg, for his part, was the CEO of The High Value Healthcare Collaborative, a consortium of 15 healthcare delivery systems, before co-founding Quartet.

In January, Quartet raised a $40 million Series C to expand throughout the U.S. F-Prime Capital and Polaris Partners led the round, with participation from GV and Oak HC/FT. The financing valued the company at $300 million, according to PitchBook.

As part of the funding, Quartet announced it was adding three new directors to its board: F-Prime’s executive partner Carl Byers; Ken Goulet, an executive vice president at health insurance provider Anthem; and former Rackspace CEO and BuildGroup co-founder Lanham Napier. Other outside board members include Oak HC/FT’s managing partner Annie Lamont, GV partner Krishna Yeshwant, Polaris managing partner Brian Chee and former U.S. Congressman Patrick Kennedy.

Quartet previously raised a $40 million Series B in April 2016 led by GV. The investment marked the venture capital investment arm of Google’s first in a mental health startup. Before that, the startup brought in a $7 million Series A led by Oak HC/FT’s managing partner Annie Lamont.

For now, Quartet remains committed to growth.

“We learn from what we are doing and we continue to learn,” Wennberg said. “That is part of growth. It’s hard and you just keep working and growing because we have a huge mission.”

When Brazilian-born Henrique Dubugras and Pedro Franceschi met at 16 years old, they bonded over a love of coding and mutual frustrations with their strict mothers, who didn’t understand their Mark Zuckerberg-esque ambitions. 

To be fair, their moms’ fear of their hacking habits only escalated after their pre-teen sons received legal notices of patent infringements in the mail. A legal threat from Apple, which Franceschi received after discovering the first jailbreak to the iPhone, is enough to warrant a grounding, at the very least.

Their parents implored them to quit the hacking and stop messing around online.

They didn’t listen.

Today, the now 22-year-olds are announcing a $125 million Series C for their second successful payments business, called Brex, at a $1.1 billion valuation. Greenoaks Capital, DST Global and IVP led the round, which brings their total raised to date to about $200 million.

San Francisco-based Brex provides startup founders access to corporate credit cards without a personal guarantee or deposit. It’s also supported by the likes of PayPal founders Peter Thiel and Max Levchin, the former chief executive officer of Visa Carl Pascarella and a handful of leading venture capital firms. 

“Brex is off to one of the most exciting starts we’ve ever seen,” IVP’s Somesh Dash said in a statement.

The financing makes them some of the youngest unicorn founders in history and puts them in a rare class of startups that have galloped into unicorn territory at such a fast clip. Brex was founded in the winter of 2017. It only launched publicly in June 2018.

How’d they do it?

“I’ve had two failed attempts, one successful attempt and one on the way to being a successful attempt,” Brex CEO Dubugras told TechCrunch while reciting a lengthy resume.

At 14, when most of us were worrying about what the first year of high school would bring us, Dubugras was more concerned about what his next business attempt would be. He had already built a successful online game but was forced to shut it down after receiving those patent infringement notices.

Naturally, he used the cash he earned from the game to start a company — an education startup meant to help Brazilian students apply to American schools. He himself was hoping to get into Stanford and had learned quickly how little Brazilian students understood of the U.S. college application process.

In some respects, the company was a success. It garnered 800,000 users but failed to make any money. His small fortune wasn’t enough to scale the business.

“There aren’t a lot of VCs in Brazil that are willing to fund 15-year-olds,” Dubugras told TechCrunch.

Shortly after folding the edtech, he met Franceschi, a Brazilian teen from Rio — Dubugras is from São Paulo — who understood his appetite for innovation and was just as hungry for success. The pair got to talking and because of Franceschi’s interest in payments, they started Pagar.me, the “Stripe of Brazil.”

Pagar.me raised $30 million, amassed a staff of 100 and was processing up to $1.5 billion in transactions when it sold. Finally, they had a real success under their belt. Now it was time to relocate. 

“We wanted to come to Silicon Valley to build stuff because everything here seemed so big and so cool,” Dubugras said.

And come to Silicon Valley they did. In the fall of 2016, the pair enrolled at Stanford. Shortly after that, they entered Y Combinator with big dreams for a virtual reality startup called Beyond. 

“I think three weeks in we gave it up,” Dubugras said. “We realized we aren’t the right founders to start this business.”

He credits Y Combinator with helping him realize what they were good at — payments.

As founders themselves, Dubugras and Franceschi were hyper-aware of a huge problem entrepreneurs face: access to credit. Big banks see small businesses as a risk they aren’t willing to take, so founders are often left at a dead-end. Dubugras and Franceschi not only had a big network of startup entrepreneurs in their Rolodex, but they had the fintech acumen necessary to build a credit card business designed specifically for founders.

So, they scrapped Beyond and in April 2017, Brex was born. The startup picked up momentum quickly, so much so that the pair decided to drop out of Stanford and pursue the business full time.

Simplifying financial access

Brex doesn’t require any kind of personal guarantee or security deposit and it doesn’t use third-party legacy technology; its software platform is built from scratch.

It simplifies a lot of the frustrating parts of corporate expenses by providing companies with a consolidated look at their spending. At the end of each month, for example, a CEO can easily see how much the entire company spent on Uber or Amazon. 

Plus, Brex can give entrepreneurs a credit limit that’s as much as 10 times higher than what they’d receive elsewhere and they can issue cards, virtual cards at least, moments after the online application is complete.

“We have a very similar effect of what Stripe had in the beginning, but much faster because Silicon Valley companies are very good at spending money but making money is harder,” Dubugras explained.

As part of their funding announcement, Brex said it will launch a rewards program built with the needs and spending patterns of founders in mind. Beyond that, they plan to use the capital to hire engineers and figure out how to grow the business’s client base beyond only tech startups.

“We want to dominate corporate credit cards,” Dubugras said. “We want every single company in the world, whenever they do businesses expenses, to do it on a Brex card.”

Amazon has sold a lot of Echo Dots. Like a crazy, silly, unfathomable number of the things. Over the past two generations, it has arguably become the single-largest driver of the smart speaker craze.

There’s nothing exceptional about the product, of course. It’s a simply designed hockey puck of a product, designed to mostly stay out of sight. But it’s a hard thing to resist — even for those who’ve been reluctant to embrace the category.

It’s a dip in the water, a gateway drug into the strange new world of smart speakers. So, how to improve upon Amazon’s best-selling device? The trick is adding to the experience while not impacting its biggest selling point: the fact that it’s $50. That sort of price point gives you considerably less wiggle room than with, say, a $1,000 phone.

Announced at an event in Seattle last month alongside eight million other new Alexa products, the new Dot marks more than just a simple upgrade to the line. It represents a way forward for the Echo line. It’s a product that bears Google’s unmistakable influence, while pointing toward the place the modular speaker system will occupy in the smart home going forward.

It’s the mark of Google that really strikes you right out of the box. The first two generations of the product were utilitarian. They weren’t much to look at, but rather a gateway to Alexa, designed to be hidden away. Granted, fabric covers are all the rage now in consumer electronics, but the new Echo’s cloth perimeter bears more than a passing resemblance to Google’s Home Mini.

Amazon was understandably shaken by Google’s rapid ascent in the category. Days before the Alexa event, Strategy Analytics noted that the Home Mini had surpassed the Echo Dot as the best-selling smart speaker for the quarter. It’s not exactly panic mode, but it’s a pretty clear indication that it’s time for an upgrade.

While the new Dot draws some clear aesthetic influence from the Home line, I prefer Amazon’s take. It splits the difference between old and new in a nice way. The fabric cover doubles as a speaker grille, running along the outside of the product. The top, meanwhile, maintains a familiar design language, with a rounded matte black top bearing a quartet of physical buttons. The light-up status ring runs flush between these two surfaces.

The new Dot is notably larger than its predecessor — a bit of a surprise, given that the more compact size was the second-gen Dot’s biggest selling point. That said, the fact that the new device looks nice enough to be displayed out in the open no doubt emboldened the company to make it a bit larger. It’s a solid thing, too. I was a bit surprised by the heft of the puck — you could do some serious damage with the thing.

One of the upshots of the larger footprint is the volume increase. The new Dot is capable of getting 70 percent louder than its predecessor (by Amazon’s count). The move finds Amazon putting a stronger emphasis on the second part of the “smart speaker equation.” The sound system on earlier Dots wasn’t built for much beyond giving Alexa voice. That’s why the company built in an auxiliary output.

That’s still here, of course, but the built-in sound output is much improved. It’s also a lot less distorted at top volume. I still wouldn’t use it as my default speaker, but the Dot’s role in Amazon’s new à la carte sound system is an interesting one.

The company sent along two Dots for the sole purpose of trying out the new stereo pairing feature — and I’m glad they did. It’s probably the most interesting addition to the line. In the revamped Alexa app, you’ll find the Create a Speaker Set option under the Settings tab. From here, you can turn two Dots into a stereo pair. The setup is simple — though I did run into some trouble on our office Wi-Fi. Both Echos need to be on the same network in order for the feature to work properly, and the app wasn’t quite able to discern that they, in fact, were.

The app will walk you through the process and let you determine which device will handle which channel of the stereo track. Paired together, it’s a nice experience — kind of a small-scale home theater experience. Add in the new Echo Sub and it’s even better. Keep in mind, of course, that you’ve just spent $230. Things add up fast. Of course, that’s still $100 cheaper than the HomePod.

Of course, it’s unfair to compare the two. Amazon and Apple’s speakers are in entirely different leagues. But the new Dot and other additions to the Echo home stereo system represent a very Amazon approach to the category, giving users the ability to mix and match devices, while still maintaining a low price point.

The third-generation Dot isn’t a complete reinvention of the wheel, but it’s big enough to warrant an upgrade for many users. Though perhaps “upgrade” isn’t the operative word here. Given Amazon’s ultimate goal of an Alexa device in every room, it’s easy to see it becoming yet another addition to your growing collection.

After a few months of hype, version 9.0 of the software that runs on Teslas is rolling out starting today.

There’s a bit of a catch, though: one of the bigger expected features — one that would let Tesla’s autopilot handle passing cars and automatically make highway interchanges — is being delayed for now.

So what’s new?

• Newer Teslas (those built after August 2017) can now use the built-in forward-facing cameras as a Dash cam. If you need to save some footage of what’s going on in front of your car, a new dash cam icon will save the last 10 minutes of footage.
• There’s a new UI on the center display across all of the vehicles, cleaning things up while making the user experience a good bit more consistent between the Model S/X and the 3.
• The Model 3’s center display is getting a web browser (something the Model S and X have had for a while, but the 3 has been missing), along with a calendar and real-time energy-monitoring tool.
• Turn-by-turn now shows more information (particularly about your next steps), and you can tell the car whether or not you’re currently allowed to use the carpool lane and it’ll factor that into its routing.
• Autopilot will now disable full-speed acceleration when you’re at low speeds when there’s something in front of the car. If you’re parking in a parking garage, for example, Autopilot will try to stop you from rocketing into the wall if you mix up the accelerator/brake or mistakenly think you’re in reverse.
• You can now push a navigation destination to the car via the Tesla mobile app, saving you the effort of punching it in once you’re in the driver’s seat.
• Those Atari games Elon Musk said they’d sneak in as Easter eggs? They’re apparently in there now — “if you can find them.”
• A new 360º visualization, built by tapping all eight external cameras into one big-meshed image, shows a bird’s-eye view of your car and vehicles around you.

Musk had previously said to expect Navigate on Autopilot to go live in this build, greatly improving Autopilot’s abilities on freeways. You’ll still have to keep your hands on the wheel, but it’ll suggest lane changes, handle freeway interchanges and take freeway exits for you.

So why the delay? It’s not quite ready yet, says Musk:

With today’s rollout, Navigate on Autopilot will be running in “Shadow Mode.” In other words, all of the new logic and calculations will be running silently in the background, but Autopilot won’t actually be utilizing it — it’s just crunching the numbers and double checking that everything Navigate on Autopilot would do is safe.

Given that just means it’ll (hopefully) be that much safer when it does go live, I’m all for it. When we’re talking about a 5,000+ pound metal box cruising itself around at 70+ MPH, shipping early and crossing your fingers isn’t an option.

Facebook Messenger could soon let you user your voice to dictate and send messages, initiate voice calls, and create reminders. Messenger for Android’s code reveals a new M assistant button atop the message thread screen that activates listening for voice commands for those functionalities. Voice control could make Messenger simpler to use hands-free or while driving, more accessible for the vision or dexterity-impaired, and perhaps one day, easier for international users whose native languages are hard to type.

Facebook Messenger was previously spotted testing speech transcription as part of the Aloha voice assistant believed to be part of Facebook’s upcoming Portal video chat screen device. But voice commands in the M assistant are new, and demonstrate an evolution in Facebook’s strategy since its former head of Messenger David Marcus told me voice “is not something we’re actively working on right now” in September 2016 on stage at TechCrunch Disrupt.

The prototype was discovered by all-star TechCrunch tipster Jane Manchun Wong, who’d previously discovered prototypes of Instagram Video Calling, Facebook’s screen time digital well-being dashboard, and Lyft’s scooter rentals before the officially launched. When reached for comment, a Facebook Messenger spokesperson confirmed to TechCrunch that Facebook is internally testing the voice command feature. The told TechCrunch “We often experiment with new experiences on Messenger with employees. We have nothing more to share at this time.”

Messenger is eager to differentiate itself from SMS, Snapchat, Android Messages, and other texting platforms. The app has aggressively adopted visual communication features like Facebook Stories, augmented reality filters, and more. Wong today spotted Messenger prototyping augmented reality camera effects being rolled into the GIFs, Stickers, and Emoji menu in the message composer.

Facebook has found that users aren’t so keen on tons of bells and whistles like prominent camera access or games getting in the way of chat, so Facebook plans to bury those more in a forthcoming simplified redesign of Messenger. But voice controls add pure utility without obstructing Messenger’s core value proposition and could end up getting users to chat more if they’re eventually rolled out.

For a lot of consumers, Pokémon GO wasn’t their first exposure to augmented reality — it was the dog-selfie lens inside Snapchat.

In the past few years, consumer use hasn’t evolved too heavily when it comes to what people are actually using AR for, even though technical capabilities have taken some giant leaps. Snap was an early leader, but now the industry is much more crowded with Apple, Google, Facebook and others all staffing up extensive teams focused on smartphone-based AR capabilities.

At our one-day TC Sessions: AR/VR event in LA on October 18, we’ll be chatting with Eitan Pilipski, the VP of Snap’s Camera Platform, a role that would seem to be pretty central to the long-term vision of a company that has long referred to itself as “a camera company.”

Snap has been throwing some updates to their developer tools as of late, especially for their Lens Studio product, which gives developers access to tools to create AR masks and experiences. There’s a lot of room to grow, and it will be interesting to see how much depth Snap can pull from these short experiences and whether it sees “lenses” evolving to bring users more straightforward utility in the near term.

The company hasn’t had the easiest bout as a public company lately, but it’s clear that it sees computer vision and augmented reality as key parts of the larger vision it hopes to achieve. At our LA event we’ll look to dive deeper into how they’re approaching these technologies and what it can bring consumers beyond a little added enjoyment.

As a special offer to TechCrunch readers, save 35 percent on $149 General Admission tickets when you use this link or code TCFAN. Student tickets are just $45 and can be booked here.

For a lot of consumers, Pokémon GO wasn’t their first exposure to augmented reality — it was the dog-selfie lens inside Snapchat.

In the past few years, consumer use hasn’t evolved too heavily when it comes to what people are actually using AR for, even though technical capabilities have taken some giant leaps. Snap was an early leader, but now the industry is much more crowded with Apple, Google, Facebook and others all staffing up extensive teams focused on smartphone-based AR capabilities.

At our one-day TC Sessions: AR/VR event in LA on October 18, we’ll be chatting with Eitan Pilipski, the VP of Snap’s Camera Platform, a role that would seem to be pretty central to the long-term vision of a company that has long referred to itself as “a camera company.”

Snap has been throwing some updates to their developer tools as of late, especially for their Lens Studio product, which gives developers access to tools to create AR masks and experiences. There’s a lot of room to grow, and it will be interesting to see how much depth Snap can pull from these short experiences and whether it sees “lenses” evolving to bring users more straightforward utility in the near term.

The company hasn’t had the easiest bout as a public company lately, but it’s clear that it sees computer vision and augmented reality as key parts of the larger vision it hopes to achieve. At our LA event we’ll look to dive deeper into how they’re approaching these technologies and what it can bring consumers beyond a little added enjoyment.

As a special offer to TechCrunch readers, save 35 percent on $149 General Admission tickets when you use this link or code TCFAN. Student tickets are just $45 and can be booked here.

Outside the crop of construction cranes that now dot Vancouver’s bright, downtown greenways, in a suburban business park that reminds you more of dentists and tax preparers, is a small office building belonging to D-Wave. This office, squat, angular, and sun-dappled one recent cool Autumn morning, is unique in that it contains an infinite collection of parallel universes.

Founded in 1999 by Geordie Rose, D-Wave company worked in relatively obscurity on esoteric problems associated with quantum computing. When Rose was PhD student at the University of British Columbia he turned in an assignment that outlined a quantum computing company. His entrepreneurship teacher at the time, Haig Farris, found the young physicists ideas compelling enough to give him $1,000 to buy a computer and a printer to type up a business plan.

The company consulted with academics until 2005 when Rose and his team decided to focus on building usable quantum computers. The result, the Orion, launched in 2007 and was used to classify drug molecules and play Sodoku. The business now sells computers for up to $10 million to clients like Google, Microsoft, and Northrop Grumman.

“We’ve been focused on making quantum computing practical since day one. In 2010 we started offering remote cloud access to customers and today, we have 100 early applications running on our computers (70% of which were built in the cloud),” said CEO Vern Brownell. “Through this work, our customers have told us it takes more than just access to real quantum hardware to benefit from quantum computing. In order to build a true quantum ecosystem, millions of developers need the access and tools to get started with quantum.”

Now their computers are simulating weather patterns and tsunamis, optimizing hotel ad displays, solving complex network problems, and, thanks to a new, open source platform, could help you ride the quantum wave of computer programming.

Inside the box

When I went to visit D-Wave they gave us unprecedented access to the inside of one of their quantum machines. The computers, which are about the size of a garden shed, have a control unit on the front that manages the temperature as well as queuing system to translate and communicate the problems sent in by users.

Inside the machine is a tube that, when fully operational, contains a small chip super-cooled to 0.015 Kelvin or -459.643 degrees Fahrenheit or -273.135 degrees Celsius. The entire system looks like something out of the Death Star – a cylinder of pure data that the heroes must access by walking through a little door in the side of a jet black cube.

It’s quite thrilling to see this odd little chip inside of its supercooled home. As the computer revolution maintained its predilection towards room-temperature chips, these odd and unique machines are a connection to an alternate timeline where physics is wrestled into submission in order to do some truly remarkable things.

And now anyone – from kids to PhDs to everyone in between – can try it.

Into the Ocean

Learning to program a quantum computer takes time. Because the processor doesn’t work like a classic universal computer you have to train the chip to perform simple functions that your own cellphone can do in seconds. However, in some cases researchers have found the chips can outperform classic computers by 3,600 times. This trade off – the movement from the known to the unknown – is why D-Wave exposed their product to the world.

“We built Leap to give millions of developers access to quantum computing. We built the first quantum application environment so any software developer interested in quantum computing can start writing and running applications — you don’t need deep quantum knowledge to get started. If you know Python, you can build applications on Leap,” said Brownell.

To get started on the road to quantum computing D-Wave build the Leap platform. The Leap is an open source toolkit for developers. When you sign up you receive one minute’s worth of quantum processing unit time which, given that most problems run in milliseconds, is more than enough to begin experimenting. A queue manager lines up your code and runs it in order received and the answers are spit out almost instantly.

You can code on the QPU with Python or via Jupiter notebooks and it allows you to connect to the QPU with an API token. After writing your code, you can send commands directly to the QPU and then output the results. The programs are currently pretty esoteric and require a basic knowledge of quantum programming but, it should be remembered, classic computer programming was once daunting to the average user.

I downloaded and ran most of the demonstrations without a hitch. These demonstrations – factoring programs, network generators, and the like – essentially turned the ideas concepts of classical programming into quantum questions. Instead of iterating through a list of factors, for example, the quantum computer creates a “parallel universe” of answers and then collapses each one until it finds the right answer. If this sounds odd it’s because it is. The researchers at D-Wave argue all the time about how to imagine a quantum computer’s various processes. One camp sees the physical implementation of a quantum computer to be simply a faster methodology for rendering answers. The other camp, itself aligned with Professor David Deutsch’s ideas presented in The Beginning of Infinity, sees the sheer number of possible permutations a quantum computer can traverse as evidence of parallel universes.

What does the code look like? It’s hard to read without understanding the basics, a fact that D-Wave engineers factored for in offering online documentation. For example, below is most of the factoring code for one of their demo programs, a bit of code that can be reduced to about five lines on a classical computer. However, when this function uses a quantum processor, the entire process takes milliseconds versus minutes or hours.

Classical

# Python Program to find the factors of a number

print_factors(num)

Quantum


@qpu_ha
def factor(P, use_saved_embedding=True):

####################################################################################################
# get circuit
####################################################################################################

construction_start_time = time.time()

validate_input(P, range(2 ** 6))

# get constraint satisfaction problem
csp = dbc.factories.multiplication_circuit(3)

# get binary quadratic model
bqm = dbc.stitch(csp, min_classical_gap=.1)

# we know that multiplication_circuit() has created these variables
p_vars = ['p0', 'p1', 'p2', 'p3', 'p4', 'p5']

# convert P from decimal to binary
fixed_variables = dict(zip(reversed(p_vars), "{:06b}".format(P)))
fixed_variables = {var: int(x) for(var, x) in fixed_variables.items()}

# fix product qubits
for var, value in fixed_variables.items():
    bqm.fix_variable(var, value)

log.debug('bqm construction time: %s', time.time() - construction_start_time)

####################################################################################################
# run problem
####################################################################################################

sample_time = time.time()

# get QPU sampler
sampler = DWaveSampler(solver_features=dict(online=True, name='DW_2000Q.*'))
_, target_edgelist, target_adjacency = sampler.structure

if use_saved_embedding:
    # load a pre-calculated embedding
    from factoring.embedding import embeddings
    embedding = embeddings[sampler.solver.id]
else:
    # get the embedding
    embedding = minorminer.find_embedding(bqm.quadratic, target_edgelist)
    if bqm and not embedding:
        raise ValueError("no embedding found")

# apply the embedding to the given problem to map it to the sampler
bqm_embedded = dimod.embed_bqm(bqm, embedding, target_adjacency, 3.0)

# draw samples from the QPU
kwargs = {}
if 'num_reads' in sampler.parameters:
    kwargs['num_reads'] = 50
if 'answer_mode' in sampler.parameters:
    kwargs['answer_mode'] = 'histogram'
response = sampler.sample(bqm_embedded, **kwargs)

# convert back to the original problem space
response = dimod.unembed_response(response, embedding, source_bqm=bqm)

sampler.client.close()

log.debug('embedding and sampling time: %s', time.time() - sample_time)

“The industry is at an inflection point and we’ve moved beyond the theoretical, and into the practical era of quantum applications. It’s time to open this up to more smart, curious developers so they can build the first quantum killer app. Leap’s combination of immediate access to live quantum computers, along with tools, resources, and a community, will fuel that,” said Brownell. “For Leap’s future, we see millions of developers using this to share ideas, learn from each other, and contribute open source code. It’s that kind of collaborative developer community that we think will lead us to the first quantum killer app.”

The folks at D-Wave created a number of tutorials as well as a forum where users can learn and ask questions. The entire project is truly the first of its kind and promises unprecedented access to what amounts to the foreseeable future of computing. I’ve seen lots of technology over the years and nothing quite replicated the strange frisson associated with plugging into a quantum computer. Like the teletype and green-screen terminals used by the early hackers like Bill Gates and Steve Wozniak, D-Wave has opened up a strange new world. How we explore it us up to us.

Reelgood, a startup aimed at helping cord cutters find their next binge, is out today with its biggest update yet. The company has been developing its streaming guide over the past year to solve the issues around discovery that exist when consumers drop traditional pay TV in favor of streaming services like Netflix, Hulu, HBO, Prime Video, and others.

The company first launched as a website in the summer of 2017 before expanding to mobile last fall. During that time, it’s grown to over a million monthly active users who now check in with Reelgood to find something new to watch.

With today’s update to its iOS app, Reelgood is adding a number of features, including personalized recommendations, curated selections, alerts for shows and movies you’re tracking, advanced search and filtering, and the ability to track content over 50 more streaming services, among other things.

As discovery is Reelgood’s focus, the updated app now offers two new types of recommendations.

One is Reelgood’s own take on “Because You Watched” – a type of viewing suggestion you’ll find today on individual services, like Netflix. But those are more limited because they’ll only suggest other shows or movies they offer themselves. Reelgood’s recommendations will instead span all the services you have access to, offering a more universal set of suggestions.

This feature is tied to Reelgood’s watch history, where you track which shows and movies you’ve seen. That means you have to use Reelgood as your tracking app as well, in order for this feature to work.

The app’s other new way of offering recommendations is less personalized – in fact, it’s random. Because sometimes serendipity is a better way to find something, a feature called “Reelgood Roulette” lets you shake your device while on the Discover tab to get a non-personalized, random suggestion.

Reelgood credits Netflix Roulette, created by Andrew Sampson, as the basis for this addition. In fact, it acquired the rights to the software last year, and then updated it to support more streaming services.

The app also now offers more powerful search and filtering capabilities involving Rotten Tomatoes, IMDb scores, plus cast and crew listings. This allows you to query up things like “Meryl Streep’s top-rated movies” or “drama series with an IMDb rating of at least 8.0 that came out in the last 3 years,” for example.

Reelgood’s search and filtering mechanisms have always been the place where it excels, but it’s less useful as a simple tracker. For that, I prefer TV Time, which lets you quickly mark entire seasons or series as “Watched” and offers discussion boards for each episode where you can post photos and memes and chat with other fans.

TV Time, however, hasn’t been as useful for making recommendations – its suggestions have been off-the-mark when I’ve tried it in the past, often leaning too heavily on network’s back catalogs than pushing me to more current or trending content. It makes me wish I could combine the two apps into one for the best of both worlds – tracking and recommendations.

The updated Reelgood app also doubles down on its own curation capabilities by offering editorial collections. For example: 2018 Emmy Nominees, IMDb’s Top 250 Movies, Original Picks, Dark Comedies, British Humour, and more. This can be a good way to find something to watch when you’re really stumped.

And as you discover new shows and movies you want to see, you can set alerts so you’ll be notified when they hit one of the streaming services you’re subscribed to, similar the tracking feature on Roku OS.

Finally, Reelgood’s update includes the addition of 50+ streaming services – that means there’s now support for more niche services like IndieFlix, FilmStruck, Shudder, Fandor, CrunchyRoll, Mubi, AcornTV and Starz, among others.

“Reelgood 4.0 is the culmination of all we’ve learned about how people watch and the increasingly fragmented streaming world,” said Eli Chamberlin, Reelgood’s head of product and design. “Our aim with this release was to take all the streaming content out there, and display it in the most meaningful way possible so that people can get the most out of their existing streaming services without wasting countless hours browsing.”

The new app is rolling out to iOS today on the App Store.