Year: 2018

First, background checks at startups, then Huawei’s finance chief is arrested, SoftBank’s IPO is subscribed, and I am about to record our next edition of TechCrunch Equity. It’s Thursday, December 6, 2018.

TechCrunch is experimenting with new content forms. This is a rough draft of something new – provide your feedback directly to the author (Danny at danny@techcrunch.com) if you like or hate something here.

The dilemma of continuous background checks

My colleague John Biggs covered the Series A round for Israel-based Intelligo, a startup that provides “Ongoing Monitoring” — essentially a continuous background check that can detect if (when?) an employee has suddenly become a criminal or other deviant. That’s a slight pivot from the company’s previous focus of using AI/ML to conduct background checks more efficiently.

Background checks are a huge business. San Francisco-based Checkr, perhaps the most well-known startup in the space, has raised $149 million according to Crunchbase, driven early on by the need to on-board thousands of contingent workers at companies like Uber. Checkr launched what it calls “Continuous Check” which also actively monitors all employees for potential problems, back in July.

Now consider a piece written a few weeks ago by Olivia Carville at Bloomberg that explored the rise of “algorithmic auditors” that actively monitor employee expenses and flags ones it feels are likely to be fraudulent:

U.S. companies, fearing damage to their reputations, are loath to acknowledge publicly how much money they lose each year on fraudulent expenses. But in a report released in April, the Association of Certified Fraud Examiners said it had analyzed 2,700 fraud cases from January 2016 to October 2017 that resulted in losses of $7 billion.

Here’s a question that bugs me though: we have continuous criminal monitoring and expense monitoring. Most corporations monitor web traffic and email/Slack/communications. Everything we do at work is poked and prodded to make sure it meets “policy.”

And yet, we see vituperative attacks on China’s social credit system, which …. monitors criminal records, looks for financial frauds, and sanctions people based on their scores. How long will we have to wait before employers give us “good employee behavior” scores and attach it to our profiles in Slack?

The conundrum of course is that no startup or company wants (or can) avoid background checks. And it probably makes sense to continually monitor your employees for changes and fraud. If Bob murders someone over the weekend, it’s probably good to know that when you meet Bob at Monday’s standup meeting.

But let’s not pretend that this continuous monitoring isn’t ruinous to something else required from employees: trust. The more heavily monitored every single activity is in the workplace, the more that employees feel that if the system allows them to get away with something, it must be approved. Without any checks, you rely on trust. With hundreds of checks, policy is essentially etched into action — if I can do it, it must meet policy.

In China, where social trust is extremely low, it likely makes sense to have some sort of scoring mechanism to substitute. But for startups and tech companies, building a culture of trust — of doing the right thing even when not monitored — seems crucial to me for success. So before signing up for one of these continuous services, I’d do a double take and consider the potentially deleterious consequences.

If I was a startup employee, I would think twice (maybe thrice?) before traveling to China

Last weekend, Trump and Xi agreed to delay the implementation of tariffs on Chinese goods, which led to buoyant Chinese (tech) stocks Monday in Asia time zones. I wrote about how that doesn’t make any sense, since delaying tariffs doesn’t do anything to solve the structural issues in the US/China conflict:

To me the market is deeply misjudging not only the Chinese economy, but also the American leadership as well.

And specifically, I wrote about constraints on Huawei and ZTE:

In what world do these prohibitions disappear? The U.S. national security agencies aren’t going to allow Huawei and ZTE to deploy their equipment in America. Like ever. Quite frankly, if the choice was getting rid of all of China’s non-tariff barriers and allowing Huawei back into America, I think the U.S. negotiators would walk out.

So it was nice to learn (for me, not for her) that the head of finance of Huawei was arrested last night in Canada at the United States’ request. From my colleague Kate Clark:

Meng Wanzhou, the chief financial officer of Huawei, the world’s largest telecom equipment manufacturer and second-largest smartphone maker, has been arrested in Vancouver, Canada on suspicion she violated U.S. trade sanctions against Iran, as first reported by The Globe and Mail.

Huawei confirmed the news with TechCrunch, adding that Meng, the daughter of Huawei founder Ren Zhengfei, faces unspecified charges in the Eastern District of New York, where she had transferred flights on her way to Canada.

If you wanted to know how the Trump administration was going to continue to fight the trade war outside of tariffs, you now have your answer. This is a bold move by the administration, targeting not just one of China’s most prominent tech companies, but the daughter of the founder of the company to boot.

China has since demanded her return.

Here is how this is going to play out. China is preventing the two American children of Liu Changming from leaving the country, essentially holding them hostage until their father returns to the mainland to face a criminal justice process related to an alleged fraud case. America now has a prominent daughter of a major Chinese company executive in their hands. That’s some nice tit-for-tat.

For startup founders and tech executives migrating between the two countries, I don’t think one has to literally worry about exit visas or extradition.

But, I do think the travel security operations centers at companies that regularly have employees moving between these countries need to keep very keen and cautious eyes on these developments. It’s entirely possible that these one-off “soft hostages” could flare to much higher numbers, making it much more complicated to conduct cross-border work.

Quick Bites

SoftBank’s IPO raises a lot of dollars

Takahiko Hyuga at Bloomberg reports that SoftBank has sold its entire book of shares for its whopping $23.5 billion IPO. The shares will officially price on Monday and then will trade on December 19. This is a critical and important win for Masayoshi Son, who needs the IPO of his telecom unit to deleverage some of the risk from SoftBank’s massive debt pile (and also to continue funding his startup dreams through Vision Fund, etc.)

SoftBank Vision Fund math, part 2

Arman and I talked yesterday about the complicated math behind just how many dollars are in SoftBank’s Vision Fund. More details, as Jason Rowley pointed out at Crunchbase News:

In an annual Form D disclosure filed with the Securities and Exchange Commission this morning, SBVF disclosed that it has raised a total of approximately $98.58 billion from 14 investors since the date of first sale on May 20, 2017. The annual filing from last year said there was roughly $93.15 billion raised from 8 investors, meaning that the Vision Fund has raised $5.43 billion in the past year and added six new investors to its limited partner base.

I said yesterday that the fund size should be “$97 billion or $96.7 billion with precision, assuming this $5 billion reaches a final close.” So let’s revise this number again to $99 billion or $98.6 billion with precision, since it seems the $5 billion did indeed close.

What’s next

I am still obsessing about next-gen semiconductors. If you have thoughts there, give me a ring: danny@techcrunch.com.

Thoughts on Articles

Hopefully more reading time tomorrow.

Reading docket

What I’m reading (or at least, trying to read)

• Huge long list of articles on next-gen semiconductors. More to come shortly.

If you’ve got an Echo device at home, you’ve almost certainly played a few rounds of “stump Alexa,” asking Amazon’s assistant increasingly arcane questions in an attempt to elicit some baffled response. Likely you found it was all that difficult to confuse the AI. After all, Amazon doesn’t have the advantage of Google’s deep Knowledge Graph.

One quick workaround is going the Wikipedia route, drawing upon the knowledge of users to help build a deeper base of knowledge. Amazon’s just opened that door, with an invite-only program that asks customers to submit answers to Alexa’s more difficult questions.

Amazon has been testing Alexa Answers internally, adding more than 100,000 responses in the past month alone. Now it’s opening the program up to a small cross section of the public, via email invites. Those who get asked to join can answer questions for Alexa via a website hosting a broad range of different topics.

For example, Amazon’s offered up the following (admittedly bizarre) suggestions. “Where was Barbara Bush buried?” or “Who wrote the score for Lord of the Rings?” or “What’s cork made out of?” or “Where do bats go in the winter?” I’d like to image they were all asked by the same weirdo in quick succession.

Once responses are added, Alexa will have access to them, noting that they’re from “an Amazon customer” — one assumes to take a bit of the heat off of the assistant. But what could possibly go wrong?

Movies Anywhere, the Disney-led movie service operated in partnership Universal, WB, Sony Pictures and 20th Century Fox, is adding its first pay TV partner. Starting today, Comcast Xfinity TV customers will be able to sync their accounts with Movies Anywhere in order to access their movie purchases from the Xfinity Digital Store alongside those from other digital retailers.

The movies can be accessed on Xfinity X1, in the Xfinity Stream app and on other Xfinity TV platforms.

The Movies Anywhere service first launched in 2014, as a way for Disney fans to aggregate their Disney, Pixar and Marvel film purchases in one place, including those bought through iTunes as well as those they owned on DVD or Blu-ray that came with a digital copy.

Last year, the service dropped “Disney” from its branding and branched out to include a variety of other digital retailers, including Amazon Prime Video, Google Play and VUDU. It also then allowed customers to aggregate more than just their Disney, Marvel, Star Wars and Pixar films, to include movies they own from Sony, 20th Century Fox, Universal and Warner Brothers.

That made the service much more useful, as it became a one-stop shop for accessing your broader digital movie collection.

Now, Comcast customers can participate as well, bringing in to the app their own digital movie purchases. To encourage sign-ups, Movies Anywhere is again giving away free movies as part of its launch promotion. This time, customers who sync for the first time will receive a copy of “Happy Feet,” and those who sync for the first time with at least two digital retailers will also get “The Martian” and “The Fate of the Furious.”

After signing up and syncing with the various digital stores, Comcast customers will be able to access their full movie collection for the first time through Comcast’s own platforms, as well. This includes streaming the movies on the TV through Xfinity on Demand, and on devices through the Xfinity Stream app and the web portal. Customers can also use the Movies Anywhere app or web portal across platforms to access their movie collection.

“Comcast is one of the country’s leading pay TV providers with a customer base that, like ours, consists of people who are passionate about the movies they love,” said Karin Gilford, General Manager, Movies Anywhere, in a statement. “We are thrilled to now include Comcast’s Xfinity TV customers among those who can benefit from Movies Anywhere’s ability to bring their favorite movies together in one place that can be accessed across a multitude of devices using the Movies Anywhere app and across Xfinity TV platforms.”

Qualcomm’s big Snapdragon Summit has become yet another battleground for smartphone oneupmanship, as companies duke it out around the latest flagship chips. OnePlus, which has carved out a name for itself by producing high quality handsets on a budget, clearly intended to use the Maui-based event to establish itself among innovation leaders in the category. Ultimately, however, it appears the Chinese manufacturer is leaving the show with a bit of egg on its face.

CEO Pete Lau hit the stage this week backed by a slide proclaiming that the company’s 2019 handset would be the first to market with the Snapdragon 855. Huge if true. After all, OnePlus has long taken a wait and see approach to new tech. The company has rarely been the first, but that kind of patience has paid off in lower cost, solidly built phones.

That claim, however, was apparently the result of an unfortunate miscommunication — the last thing you want when a company like Qualcomm invites you to their big party. “We appreciate the opportunity to clarify that we will be one of the first to have access to, and use, the Qualcomm Snapdragon 855, and apologize for the miscommunication,” the company said in a statement provided to Engadget.

OnePlus blames the error on a slide deck produced by a non-native English speaker. The note should have read something along the lines of “one of the first to feature.” These things happen, I suppose, though the mistake looks to be alive and well in this tweet, boasting Lau’s show stealing “surprise announcement.”

Perhaps OnePlus simply didn’t have the clearance to make such a claim on stage and only learned about it in hindsight. The company may in fact even be the first to offer the feature, but anyone who’s been involved in one of these sorts of corporate presentation will tell you the company in charge is extremely touchy when it comes to exact language — especially when it comes to playing favorites with other hardware partners.

Quite frankly, I’m surprised Qualcomm wasn’t micromanaging the A/V portion of the show to a greater degree.

Assuming all goes well, OnePlus will still be able to claim its position as one of the first companies to bring 5G to market. Though that push to innovate will come with a hefty price tag. Here’s Lau again, this time speaking with the Verge, “It’s hard to know because there’s a lot of specifics still to look at, but it’s likely in the neighborhood of $200-300 more.”

That’s a pretty high premium for a company for a company whose growth has been so closely tied to cost. Given the 6T’s $549 starting price, we’re definitely pushing well into a premium flagship price point here. And for many consumers in many regions, the effects of 5G won’t be immediately apparently. 

What OnePlus does in the next year or so will have a profound effect on the company going forward. The smartphone maker is at a crossroads, and will have to determine where it plans to operate in the future, as king of the midrange or a premium manufacturer competing for Apple and Samsung’s high-end marketshare.

Fire TV Cube owners can now use their media streaming device to broadcast announcements around the home, similar to a one-way intercom. Amazon says it’s launching a feature called “Alexa Announcements,” which allows customers to speak to Alexa on the Fire TV Cube, then have their message announced in their own voice to all other Alexa-enabled devices in their household.

For example, you could announce that “the movie is about to start,” or say “Alexa, tell everyone that dinner is ready,” and have the message sent to your Echo devices or other compatible Alexa-enabled devices registered to your Amazon account.

To issue your broadcast, you preface it by saying either “Alexa, announce…,” “Alexa tell everyone…,” or “Alexa broadcast…”, Amazon notes.

In addition to being able to send out announcements yourself, the Fire TV Cube starting this month will also support announcements that are broadcast from select smart home devices – specifically, those where the doorbell is ringing or motion is detected.

In this case, Alexa will announce things like “motion detected at the front door” or “someone is at the front door,” and you can respond by saying “Alexa, show me the front door” to see your camera feed displayed on your TV.

You can customize which smart home announcements you’d like to see through the new Alexa Announcements section in the Alexa app.

Amazon didn’t say if or when the feature would come to other Fire TV devices.

However, this is not the first time the Fire TV Cube has been updated with a feature its other media players others don’t have.

In October, the Fire TV Cube was updated with a feature that allows you to use Alexa to navigate inside apps like Netflix and Hulu, which is especially helpful for accessibility purposes. It aids those with limited dexterity in their hands and fingers by allowing them to navigate by voice instead of pushing buttons. But it can also help if you’ve just misplaced the remote.

The Alexa Announcements feature is rolling out now to all Fire TV Cube customers in the U.S.

The Linux Foundation’s OpenChain project helps company find ways to comply with open source licensing requirements. Today at the Open Compliance Summit in Yokohama, Japan it announced Google, Facebook and Uber have joined the project as platinum members.

As platinum members the three companies become part of the governing board. Shane Coughlan, OpenChain General Manager says as the project has matured, this a logical point for three large technology companies to come on board.

“Facebook, Google and Uber are perfect new additions for this point, as we move towards becoming a formal industry standard and scaling very significantly across multiple markets. In particular, we are making sure that we can clearly communicate the advantages of OpenChain, and we can clearly show that diversity and the knowledge of our board, as well,” Coughlan told TechCrunch.

As companies who use significant amounts of open source code, having these three organizations in particular really raises the profile of the project. “Facebook, Google and Uber are really interesting because they are in strong competition with each other and are coalescing around a single solution. And I think that’s probably one of the most important messages today. Companies which are intense rivals in many markets all see one single coherent approach to open source compliance in the supply chain and are putting their weight behind it,” he said.

Google has been building its own compliance processes for years, and it sees this project, which launched in 2016, as a way to help standardize open source governance. “The OpenChain Project is changing that by defining the core requirements of a quality compliance program and developing curriculum to help with training and management. It’s hard to overstate the importance of this work now that open source is a critical input at every step in the supply chain, both in hardware and software,” Google’s Max Sills and Josh Simmons wrote in a blog post on the company joining OpenChain.

While OpenChain works toward a standard way of helping companies comply with open source licensing requirements, regardless whether it’s a tech company or any other industry, the project is about helping companies, rather than prescribing a particular way to comply. They look for best efforts and let members deal with the specifics of how they do it.

As applications make use of open source libraries, it’s up to the developers and their companies to make sure they use the code as defined in whatever license the project uses. The Linux Foundation, which is an umbrella organization for many popular open source projects, created the OpenChain project as a way to encourage developers to comply with these requirements.

After going public in the UK last year, Boku has made an acquisition to expand its carrier billing services, which let users bill mobile content purchases from companies like Apple, Microsoft, Spotify, and 152 other app and other content purveyors to their mobile bills. Today, Boku announced the acquisition of Danal, Inc, a specialist in mobile identity and authentication services, so that it can offer more sophisticated transaction services and also to move into new areas.

Boku will pay up to $68 million for Danal, the company said. Specifically, the financial terms are described by Boku as a “reverse triangular merger” and include 26.7 million Boku common shares of $0.0001 each (“Common Shares”), $3 million of Boku warrants exercisable at 141p each and $1 million in cash, along with a deferred consideration of up to $64 million, “satisfied in Common Shares and warrants, dependent on Danal’s future performance,” which Boku also described as “challenging performance targets for Danal, thereby allowing both parties to share the benefits of efficiencies and growth.”

Danal, Boku said, will become a part of a US subsidiary of Boku.

The market is not particularly excited by the deal it seems: the company’s stock has dropped by more than 23 percent in trading today. Boku currently has a market cap of around £168 million ($216 million), and it says that total payment volume in the 10 months to October was up 124 percent to $2.8 billion (versus $1.3 billion the year before), and monthly active users were 12.2 million in October, up 83 percent on a year before.

This is not Danal’s first transaction with a carrier billing service. In 2016, it sold a portion of its business, BilltoMobile, to Bango for $3.5 million.

Boku is buying the rest of the business left behind, with a view to building a bridge between the data that carriers have about their users and services that those users might engage with either on their mobile devices, or through other digital channels. This could include expanding the range of purchases that you can make through carrier billing, but it could potentially also be applied to any service that either has a risk of fraud — such as financial or government-run services — or could use a carrier data to help authenticate the identity of the user.

“Charging purchases to your phone bill has proved a great way for the world’s largest digital companies to acquire and retain users, but has had fairly limited application outside digital content,” said Jon Prideaux, CEO of Boku, in a statement. “This Acquisition allows us to offer services that go further and to improve user quality for our customers while at the same time improving the mobile experience for users. Mobile commerce is booming, yet many tools were developed to support PC-based commerce. Danal has shown that MNO data can also combat fraud, reduce friction in signup and ensure regulatory compliance on mobile. These problems are relevant not just to our existing digital customers but also in other sectors including e-commerce, finance, transportation and government.”

Notably, this potentially could help Boku grow revenues in developed markets alongside the emerging markets where it is currently active.

Danal, based in San Jose, already counts financial institutions, government agencies and retailers as customers, including Western Union, BNP Paribas, PayPal, Square, Moneygram, Login.gov and USAA.

Boku said Danal  generated revenue of $5.1 million and a loss before interest, taxation, depreciation and amortisation of $5.2 million for the full year that ended December 31, 2017. Liabilities as of that date were $10.3 million.

The bigger picture for mobile payments are that while they continue to grow, they are still just around one-third of all e-commerce transactions, according to recent figures collected over the opening weekend of holiday sales.

Within that, billing to carriers is just one part of the overall mix, and after accounting for others in the transaction chain, it makes for thin margins. This explains partly why Boku would be working on adding in new revenue streams. But in emerging markets, carrier billing is a popular alternative among users who may not have bank accounts and payment cards. This latest deal for Boku should help it in that area, too.

In its 3,000-square-foot warehouse in San Francisco’s Mission District, Farmstead founders Pradeep Elankumaran and Kevin Li, a pair of former Yahoo product managers, plot the future of grocery shopping.

“Think of us as if Whole Foods was rebuilt from scratch by tech founders,” Elankumaran, Farmstead’s chief executive officer, told TechCrunch. “Of course we do delivery because it’s 2018 and no one wants to go to the store anymore.”

Elankumaran launched San Francisco-based Farmstead in 2016 after Amazon and Instacart’s food delivery services repeatedly disappointed him. The startup leverages artificial intelligence-powered predictive analytics and machine learning to accurately predict supply and demand of its inventory, a move Elankumaran says has helped the company significantly reduce waste, as well as complete deliveries to Bay Area residents in less than an hour.

“I had a lot of trouble getting food delivered consistently,” he said. “My daughter had just turned two and she started drinking a lot of milk and I found myself going to the grocery store three to four times a week to buy the same things.”

“So I posted on Nextdoor asking if anyone was interested in a milk, eggs and bread delivery service and in two days, 200 people said yes.”

Two-plus years later, the company is today announcing an additional seed round of $2.2 million, bringing its total raised to date to $7.5 million. ARTIS Labs, Resolute Ventures and Red Dog Capital participated in the round, along with Y Combinator . Farmstead completed the Silicon Valley accelerator program in 2016 shortly before its initial launch, similar to Instacart, which graduated from Y Combinator in 2012. Elankumaran said the company plans to use the capital to hire aggressively and expand beyond the Bay Area in 2019. 

Farmstead’s business may sound a lot like Instacart, a very well-funded grocery delivery service worth an astounding $7.6 billion, but the startup says the differences are notable. Instacart is a tech layer on top of a supermarket that provides delivery, whereas Farmstead is the supermarket and the delivery service. Elankumaran says this — storing groceries in large, centralized warehouses and making the deliveries — is a highly scalable model destined to defeat Instacart.

Resolute Ventures general partner Mike Hirshland said in a statement that Farmstead could “become a monster company.”

“To replace a trip to the grocery store, so many things have to go right, from ordering the right inventory to last-mile delivery. Farmstead has cracked the code on making grocery delivery profitable and rapidly scalable,” he said.

The company has also recently partnered with Udelv, an autonomous vehicle startup, to make deliveries via the company’s modified GEM eL XD electric trucks.

Amazon Music, reportedly the third-largest subscription-based music service worldwide, is doubling down on voice. The company is today announcing the launch of new set of features that will allow Amazon Music customers to chat with Alexa in a more natural way in order to find and play the music they want to hear. The features allow Amazon Music customers and Alexa to engage in back-and-forth conversations to narrow down playlist options, and will allow users to tell Alexa what songs they like or hate, and have her remember.

Alexa’s integration with Amazon Music isn’t new. The company has been working to add a variety of voice control options for its music service since launch so customers could easily play songs on their Echo speakers and other Alexa-powered devices.

Over the years, it has offered the ability for customers to ask Alexa for music by mood, genre, or even lyrics, and more. But these have been simple voice commands, like “Alexa, play pop music,” for example, or “Alexa, play the song that goes…”, followed by some lyrics.

Now, Alexa will be able to have more of a discussion with the customer.

At launch, this feature will help customers fight the right playlist – like a holiday playlist or one for a dinner party, Amazon suggests.

To get started, customers can say, “Alexa, help me find a playlist,” “Alexa, help me find music for a holiday party,” or “Alexa, help me find dinner music.”

For help finding a playlist by genre, you have to start by saying “Alexa, help me find a playlist,” then direct the assistant to the genre you want. (Asking her, “Alexa, help me find a workout playlist,” didn’t work to trigger this feature during testing, unfortunately).

Amazon says Alexa can also be directed to playlists by mood or tempo, in addition to genre or occasion.

After the initial voice command, Alexa responds with follow-up questions to customize the playlist selection for the individual user.

When answering, the assistant sometimes injects her personality into the conversation, as well, by way of her typical corny jokes and comments.

In testing the feature, the conversations often began in different ways.

For example:

Me: Alexa, help me find a holiday playlist.
Alexa: We’ll be rocking around the Christmas tree. Any vibe in particular? Mellow? Or maybe upbeat?

Me: Alexa, help me find a holiday playlist.
Alexa: Sure, let’s sample a few holiday playlists. Shall we narrow the options down to country or maybe jazz?

Me: Alexa, help me find a holiday playlist.
Alexa: Let’s sample a few holiday playlists. Were you thinking any vibe in particular? Relaxed or upbeat, perhaps?

Me: Alexa, help me find a holiday playlist.
Alexa: It’s the most wonderful time of the year. So many options. Should we narrow it down to a classical style or maybe pop?

You can respond to Alexa’s questions naturally, as she offered her various suggestions.

For example:

Me: Alexa, help me find a holiday playlist.
Alexa: Amazon Music has a ton of merry, merry good playlists. Let’s find the right one for you. Were you thinking any vibe in particular? Relaxed or upbeat, perhap?
Me: Relaxed.
Alexa: Relaxing sounds good. Here’s one of my more popular Christmas playlists, Christmas Choirs.
[A sample of the playlist’s music begins]
Alexa: Will that work?
Me: Let’s do something else.
Alexa: Alright. Try this.
[Another playlist begins]
Alexa: An Irish Christmas. How about that?
Me: Nah. Something else.
Alexa: Okay. You might like this.
[A third playlist begins]
Alexa: Christmas Oldies. Will that work?
Me: Okay, that’s cool.
Alexa: Great. Happy Holidays.
[Music plays]

“It’s a paradigm shift because you’re getting into conversations,” explains  Kintan Brahmbhatt, Director of Product for Amazon Music, about the new feature. “Before, customers would ask Alexa to do something and Alexa would respond with music. Today, we’re shifting into deep personalization and Alexa will have an interactive conversation with the customer,” he says.

“The vision and the mission is to provide a very magical and simple customer experience when it comes to discovering music and enjoying music,” Brahmbhatt adds.

The ability to have back-and-forth chats with Alexa isn’t the only new feature today.

The voice assistant will work to help make the music experience more personalized, too.

Now, you can tell Alexa when you like or dislike the song that’s playing. There isn’t one specific command to use here. Instead, any number of phrases will work, like “Alexa, I like this song,” “Alexa, this is my favorite,” or “Alexa, I don’t like this.”

In addition to these explicit signals of interest, Alexa will also learn from your implicit actions – like which songs you play most often. These signals then feed into her ability to play something you like when you ask her to simply “Play Music.”

That command will now trigger a more personalized response that’s based on algorithms that consider factors like your personalized playlist, songs you’ve said you liked, artists you’ve asked Alexa to follow in the past who have new music out, and even forgotten songs you haven’t heard in a while, but used to like.

The features are rolling out starting to today to Amazon’s tens of millions of Amazon Music subscribers, which includes both Prime Music and Amazon Music Unlimited. It will work across any Alexa device, not just Echo speakers and screens.

Amazon is also running a deal that gives customers access to the premium streaming service, Amazon Music Unlimited’s 50+ million songs, for $0.99 per month for the first three months.

In the near future, Amazon says Alexa will be able to suggest music, too, when asked for recommendations.

She’ll do so by using cues from your previous listening habits, and by asking a few questions regarding your favorite genres, eras and other preferences. This will allow Alexa to be able to anticipate what customers are in the mood to hear, and suggest relevant music, or new releases tailored to them.

ECG/EKG was easily the new Apple Watch’s most lauded feature. It’s also been the most delayed. Of course, this kind of serious health feature is the sort of thing you need to get exactly right, for reasons that ought to be pretty obvious on their face.

Electrocardiogram finally goes live today for Series 4 owners as part of the watchOS 5.1.2 update. It’s an important feature — and one that will go a ways toward helping establish the wearable as a more serious health monitor.

The new feature builds on a hardware upgrade built into the Series 4: a pair of electrodes built into the larger back crystal on the rear of the watch and the digital crown. Once enabled, the new feature is checking for a couple of key bits of heart health: irregular heart rhythms, which the watch will passively monitor in the background, and ECG, which requires the user to actively engage with by completing the circuit with a finger tip placed on the edge of the watch’s digital crown.

Of course, getting all of this isn’t as simple as just installing a software update. There is, understandably, a pretty long opt-in here. The on-boarding process is several pages long for both of the new features, as Apple collects some vital information and repeatedly reminds you of some important information — like the fact that the watch can’t detect a heart attack. If you feel like you might be having one, call the emergency services.

The Apple Watch isn’t meant to replace a doctor either, of course. Really, it’s just a way to monitor for complications. If the smartwatch can be regarded as a potential lifesaver or even peripheral medical device, it’s due to the fact that it features a kind of always-on monitoring. After all, outside of the proliferation of these sorts of wearables, most of us won’t experience something like constant ECG monitoring until under the care of a doctor. If this feature is capable of isolating that information ahead of time, it could go a ways toward addressing complications before they turn into major issues.

The sign-up process airs on the side of caution, while attempting to not overwhelm the end user with information. It’s a tricky balance, and if TOS have taught us anything, it’s that too much information upfront will ultimately result in the user’s eyes glazing over. In the case of this information, that could potentially lead to serious consequences if not properly adhered to.

Some of the key takeaways:

• It cannot detect a heart attack (see a doctor)
• It cannot detect blood clots or a stroke (see a doctor)
• It cannot detect other heart-related conditions (see a doctor)
• [It] is not constantly looking for AFib

That last one is particularly important when distinguishing between the new features. While heart rhythm detection is a feature, the Watch isn’t regularly looking for atrial fibrillation. That’s where the ECG app and the finger detection come in. The feature is intended to be used when the heart rhythm monitor detects that something is off — like a skipped or rapid heartbeat. In which case, it will send a notification right to your wrist.

If that happens, fire up the ECG app, rest your arm on your lap or a table and hold your finger to the crown for 30 seconds. Apple will display a real-time graph of your heart rhythm while you wait. It’s strangely soothing, honestly, though Apple doesn’t recommend using the feature with much regularity, unless you have cause to.

Using it just now, I got a “This ECG does not show signs of atrial fibrillation” note, meaning the reading falls within the parameters of a sinus rhythm.

Here’s your old friends at WebMD:

Your heart’s job is to pump blood to your body. When it’s working the way it should, it pumps to a regular, steady beat. This is called a normal sinus rhythm. When it’s not, you could have an irregular heartbeat called AFib.

So, good. No need to call the doctor. If you’re still feeling unwell, however, there’s a quick link to dial emergency services on the screen. There’s also a spot for adding any symptoms you might be having if you’re feeling less than 100 percent. And while Apple promises not to share any of the info collected on-device, you can always export your findings to a PDF for your doctor to take a gander at.

Along with the new feature comes a new White Paper, detailing the technology. It’s an usual bit of transparency from Apple, but the company understandably wants to be as upfront about the technology as possible. The paper details a lot of what went into bringing the feature up to speed for general availability.

Apple started with a pre-clinical study of 2,000 subjects, including ~15 percent who have been diagnosed with heart arrhythmia. Six-hundred subjects were then involved with the clinical trial to validate the AFib.

Per Apple, “Rhythm classification of a 12-lead ECG by a cardiologist was compared to the rhythm classification of a simultaneously collected ECG from the ECG app. The ECG app demonstrated 98.3% sensitivity in classifying AFib and 99.6% specificity in classifying sinus rhythm in classifiable recordings.”

The company employed similar methods to validate the Irregular Rhythm notifications. “Of the participants that wore an Apple Watch and ECG patch at the same time,” the company writes, “almost 80 percent received the notification and showed AFib on the ECG patch, and 98 percent received the notification and showed other clinically relevant arrhythmias on the ECG patch.”

In addition to that testing, the company has also employed a number of medical doctors to help ensure the product meets the sort of exacting standards one would hope from a product like this.

More information on the research can be found in this Stanford partnered paper published earlier this month.