Disrupt SF is set to be the biggest tech conference that TechCrunch has ever hosted. So it only makes sense that we plan an agenda fit for the occasion.
That’s why we’re absolutely thrilled to announce that Ring’s Jamie Siminoff will join us on stage for a fireside chat and Jason Mars from Clinc will be demo-ing first-of-its-kind technology on the Disrupt SF stage.
Jamie Siminoff – Ring
Earlier this year, Ring became Amazon’s second largest acquisition ever, selling to the behemoth for a reported $1 billion.
But the story begins long ago, with Jamie Siminoff building a WiFi-connected video doorbell in his garage in 2011. Back then it was called DoorBot. Now, it’s called Ring, and it’s an essential piece of the overall evolution of e-commerce.
As giants like Amazon move to make purchasing and receiving goods as simple as ever, safe and reliable entry into the home becomes critical to the mission. Ring, which has made neighborhood safety and home security its main priority since inception, is a capable partner in that mission.
Of course, one doesn’t often build a successful company and sell for $1 billion on their first go. Prior to Ring, Siminoff founded PhoneTag, the world’s first voicemail-to-text company and Unsubscribe.com. Both of those companies were sold. Based on his founding portfolio alone, it’s clear that part of Siminoff’s success can be attributed to understanding what consumers need and executing on a solution.
Dr. Jason Mars – Clinc
AI has the potential to change everything, but there is a fundamental disconnect between what AI is capable of and how we interface with it. Clinc has tried to close that gap with its conversational AI, emulating human intelligence to interpret unstructured, unconstrained speech.
Clinc is currently targeting the financial market, letting users converse with their bank account using natural language without any pre-defined templates or hierarchical voice menus.
But there are far more applications for this kind of conversational tech. As voice interfaces like Alexa and Google Assistant pick up steam, there is clearly an opportunity to bring this kind of technology to all facets of our lives.
At Disrupt SF, Clinc’s founder and CEO Dr. Jason Mars plans to do just that, debuting other ways that Clinc’s conversational AI can be applied. Without ruining the surprise, let me just say that this is going to be a demo you won’t want to miss.
Over the last few years, Google and Coursera have regularly teamed up to launch a number of online courses for developers and IT pros. Among those was the Machine Learning Crash course, which provides developers with an introduction to machine learning. Now, building on that, the two companies are launching a machine learning specialization on Coursera. This new specialization, which consists of five courses, has an even more practical focus.
The new specialization, called “Machine Learning with TensorFlow on Google Cloud Platform,” has students build real-world machine learning models. It takes them from setting up their environment to learning how to create and sanitize datasets to writing distributed models in TensorFlow, improving the accuracy of those models and tuning them to find the right parameters.
As Google’s Big Data and Machine Learning Tech Lead Lak Lakshmanan told me, his team heard that students and companies really liked the original machine learning course but wanted an option to dig deeper into the material. Students wanted to know not just how to build a basic model but also how to then use it in production in the cloud, for example, or how to build the data pipeline for it and figure out how to tune the parameters to get better results.
Leah Belsky, Coursera’s VP of enterprise development, echoed this and noted that this kind of specialization with a focus on practical models will make the credential more meaningful for employers.
The target audience for the specialization is somebody who wants to build new skills — and that’s pretty much every developers, especially now that machine learning is making inroads in virtually every area of tech. And since it’s almost impossible to hire machine learning experts, this course will surely be attractive to many employers who want their existing workforce to gain these skills.
As Lakshmanan noted, there are plenty of use cases for leading-edge kind of machine learning models, but what these courses focus on are more of the “day-to-day models” that can bring additional value to many existing products. Because of the focus on real-world problems, Lakshmanan also noted that the course should be useful for newly minted graduates who may be more familiar with the theories of machine learning than building products.
He also noted that only a few years ago, getting started with a course like this would have been rather cumbersome, not in the least because you need relatively powerful hardware with a dedicated GPU to work productively. Now, however, thanks to the various cloud platforms that offer GPU access or even specialized hardware like Google Cloud’s TPUs, the barrier of entry has dropped significantly.
It’s worth noting that these courses expect that you are already a somewhat competent programmer. While it has gotten much easier to start with machine learning thanks to new frameworks like TensorFlow, this is still an advanced skill. It’ll surely still be a while before we see a “get started with programming in Python by building a machine learning model” course.
Looking ahead, Lakshmanan also noted that the team is looking at a next course that would build upon the existing one, but with a focus on working with unstructured data. That’s a different class of problem with its own skill set and one that’ll allow the graduates of the first course to apply their knowledge to a whole different set of data.
Chatbot startup Hugging Face has raised a $4 million seed round led by Ronny Conway from a_capital. Existing investors Betaworks, SV Angel and Kevin Durant are also participating.
I already reviewed Hugging Face so I won’t write the same thing again. But the startup has been building a chatbot app with a strong personality for bored teenagers. Instead of focusing on customer support or convenience, Hugging Face is focusing on emotions and entertainment.
It’s been available in the App Store as a standalone app and on Kik. Today, the company is also launching Hugging Face on Messenger. It should help bring new users.
Even without Messenger, Hugging Face now handles 1 million messages per day. In total, Hugging Face has received over 100 million messages.
It’s also worth noting that Hugging Face accepts text messages, photos, emojis, everything. So you can take a selfie, send a sad emoji, and the chatbot will know how you feel.
And it’s clear that Hugging Face is betting on surprise and enjoyment. The app doesn’t have to be perfect to be entertaining.
Beyond the consumer app, the team behind Hugging Face has written a couple of research papers about artificial intelligence. It’s clear that the startup plans on building a great team of engineers when it comes to natural language conversations. The team will double over the coming months.
IBM CEO Virginia Rometty is currently having lunch with French President Emmanuel Macron in Paris. And Rometty talked with Le Monde and announced some new investments in France as part of the Tech for Good Summit organized by Macron.
The company plans to hire 1,800 new people in France over the next couple of years. While this isn’t really groundbreaking as IBM has 380,000 employees around the world, it’s interesting to see the focus of these hires.
IBM plans to put together a research team focused on blockchain projects, artificial intelligence and the internet of things. This hiring plan still represents pocket change for such a big company with hundreds of thousands of people.
Le Monde also noticed that IBM has reduced its team in France for years. Since 2012, IBM has cut more or less as many people as the company plans to hire. IBM clients in France include Orange Bank, SNCF and LVMH.
It’s not enough in this day and age that we have to deal with fake news, we also have to deal with fake prescription drugs, fake luxury goods, and fake Renaissance-era paintings. Sometimes all at once! IBM’s Verifier is a gadget and platform made (naturally) to instantly verify that something is what it claims to be, by inspecting it at a microscopic level.
Essentially you stick a little thing on your phone’s camera, open the app, and put the sensor against what you’re trying to verify, be it a generic antidepressant or an ore sample. By combining microscopy, spectroscopy, and a little bit of AI, the Verifier compares what it sees to a known version of the item and tells you whether they’re the same.
The key component in this process is an “optical element” that sits in front of the camera (it can be anything that takes a decent image) amounting to a specialized hyper-macro lens. It allows the camera to detect features as small as a micron — for comparison, a human hair is usually a few dozen microns wide.
At the micron level there are patterns and optical characteristics that aren’t visible to the human eye, like precisely which wavelengths of light it reflects. The quality of a weave, the number of flaws in a gem, the mixture of metals in an alloy… all stuff you or I would miss, but a machine learning system trained on such examples will pick out instantly.
For instance a counterfeit pill, although orange and smooth and imprinted just like a real one if one were to just look at it, will likely appear totally different at the micro level: textures and structures with a very distinct pattern, or at least distinct from the real thing — not to mention a spectral signature that’s probably way different. There’s also no reason it can’t be used on things like expensive wines or oils, contaminated water, currency, and plenty of other items.
IBM was eager to highlight the AI element, which is trained on the various patterns and differentiates between them, though as far as I can tell it’s a pretty straightforward classification task. I’m more impressed by the lens they put together that can resolve at a micron level with so little distortion and not exclude or distort the colors too much. It even works on multiple phones — you don’t have to have this or that model.
The first application IBM is announcing for its Verifier is as a part of the diamond trade, which is of course known for fetishizing the stones and their uniqueness, and also establishing elaborate supply trains to ensure product is carefully controlled. The Verifier will be used as an aide for grading stones, not on its own but as a tool for human checkers; it’s a partnership with the Gemological Institute of America, which will test integrating the tool into its own workflow.
By imaging the stone from several angles, the individual identity of the diamond can be recorded and tracked as well, so that its provenance and trail through the industry can be tracked over the years. Here IBM imagines blockchain will be useful, which is possible but not exactly a given.
It’ll be a while before you can have one of your own, but here’s hoping this type of tech becomes popular enough that you can check the quality or makeup of something at least without having to visit some lab.
Excited to announce that this year’s The Europas Unconference & Awards is shaping up! Our half day Unconference kicks off on 3 July, 2018 at The Brewery in the heart of London’s “Tech City” area, followed by our startup awards dinner and fantastic party and celebration of European startups!
The event is run in partnership with TechCrunch, the official media partner. Attendees, nominees and winners will get deep discounts to TechCrunch Disrupt in Berlin, later this year.
The Europas Awards are based on voting by expert judges and the industry itself. But key to the daytime is all the speakers and invited guests. There’s no “off-limits speaker room” at The Europas, so attendees can mingle easily with VIPs and speakers.
What exactly is an Unconference? We’re dispensing with the lectures and going straight to the deep-dives, where you’ll get a front row seat with Europe’s leading investors, founders and thought leaders to discuss and debate the most urgent issues, challenges and opportunities. Up close and personal! And, crucially, a few feet away from handing over a business card. The Unconference is focused into zones including AI, Fintech, Mobility, Startups, Society, and Enterprise and Crypto / Blockchain.
We’ve confirmed 10 new speakers including:
Eileen Burbidge, Passion Capital
Carlos Eduardo Espinal, Seedcamp
Richard Muirhead, Fabric Ventures
Sitar Teli, Connect Ventures
Nancy Fechnay, Blockchain Technologist + Angel
George McDonaugh, KR1
Candice Lo, Blossom Capital
Scott Sage, Crane Venture Partners
Andrei Brasoveanu, Accel
Tina Baker, Jag Shaw Baker
How To Get Your Ticket For FREE
We’d love for you to ask your friends to join us at The Europas – and we’ve got a special way to thank you for sharing.
Your friend will enjoy a 15% discount off the price of their ticket with your code, and you’ll get 15% off the price of YOUR ticket.
That’s right, we will refund you 15% off the cost of your ticket automatically when your friend purchases a Europas ticket.
The Awards celebrates the most forward thinking and innovative tech & blockchain startups across over some 30+ categories.
Startups can apply for an award or be nominated by anyone, including our judges. It is free to enter or be nominated.
What is The Europas?
Instead of thousands and thousands of people, think of a great summer event with 1,000 of the most interesting and useful people in the industry, including key investors and leading entrepreneurs.
• No secret VIP rooms, which means you get to interact with the Speakers
• Key Founders and investors speaking; featured attendees invited to just network
• Expert speeches, discussions, and Q&A directly from the main stage
• Intimate “breakout” sessions with key players on vertical topics
• The opportunity to meet almost everyone in those small groups, super-charging your networking
• Journalists from major tech titles, newspapers and business broadcasters
• A parallel Founders-only track geared towards fund-raising and hyper-networking
• A stunning awards dinner and party which honors both the hottest startups and the leading lights in the European startup scene
• All on one day to maximise your time in London. And it’s PROBABLY sunny!
That’s just the beginning. There’s more to come…
Interested in sponsoring the Europas or hosting a table at the awards? Or purchasing a table for 10 or 12 guest or a half table for 5 guests? Get in touch with:
Phone: +44 (0) 20 3239 9325
Speaking in front of EU lawmakers today Facebook’s founder Mark Zuckerberg namechecked the GDPR’s core principles of “control, transparency and accountability” — claiming his company will deliver on all that, come Friday, when a new European Union data protection framework, GDPR, starts being applied, finally with penalties worth the enforcement.
However there was little transparency or accountability on show during the session, given the upfront questions format which saw Zuckerberg cherry-picking a few comfy themes to riff on after silently absorbing an hour of MEPs’ highly specific questions with barely a facial twitch in response.
The questions MEPs asked of Zuckerberg were wide ranging and often drilled deep into key pressure points around the ethics of Facebook’s business — ranging from how deep the app data misuse privacy scandal rabbithole goes; to whether the company is a monopoly that needs breaking up; to how users should be compensated for misuse of their data.
Made clear to Mark Zuckerberg that digital platforms have to guarantee full protection of our citizens' privacy. We cannot accept illicit use of personal data to manipulate elections. Democracy cannot be turned into a marketing operation. pic.twitter.com/Nk0MB5IK8u
Why did it refuse a public meeting with the EU parliament? Why has it spent “millions” lobbying against EU privacy rules? Will the company commit to paying taxes in the markets where it operates? What’s it doing to prevent fake accounts? What’s it doing to prevent bullying? Does it regulate content or is it a neutral platform?
Zuckerberg made like a sponge and absorbed all this fine-grained flak. But when the time came for responses the data flow was not reciprocal; Self-serving talking points on self-selected “themes” was all he had come prepared to serve up.
But when it comes to Facebook’s own operations, the company maintains a highly filtered, extremely partial ‘newsfeed’ on its business empire — keeping a tight grip on the details of what data it collects and why.
Yes, you can download the data you’ve willingly uploaded to Facebook. Just don’t expect Facebook to give you a download of all the information it’s gathered and inferred about you.
The EU parliament’s political group leaders seemed well tuned to the myriad concerns now flocking around Facebook’s business. And were quick to seize on Zuckerberg’s dumbshow as further evidence that Facebook needs to be ruled.
Thing is, in Europe regulation is not a dirty word. And GDPR’s extraterritorial reach and weighty public profile looks to be further whetting political appetites.
So if Facebook was hoping the mere appearance of its CEO sitting in a chair in Brussels, going through the motions of listening before reading from his usual talking points, that looks to be a major miscalculation.
“It was a disappointing appearance by Zuckerberg. By not answering the very detailed questions by the MEPs he didn’t use the chance to restore trust of European consumers but in contrary showed to the political leaders in the European Parliament that stronger regulation and oversight is needed,” Green MEP and GDPR rapporteur Jan Philipp Albrecht told us after the meeting.
The MEP had also asked Zuckerberg to commit to no exchange of data between the two apps. Zuckerberg determinedly made no such commitment.
Claude Moraes, chair of the EU parliament’s civil liberties, justice and home affairs (Libe) committee, issued a slightly more diplomatic reaction statement after the meeting — yet also with a steely undertone.
“Trust in Facebook has suffered as a result of the data breach and it is clear that Mr. Zuckerberg and Facebook will have to make serious efforts to reverse the situation and to convince individuals that Facebook fully complies with European Data Protection law. General statements like ‘We take privacy of our customers very seriously’ are not sufficient, Facebook has to comply and demonstrate it, and for the time being this is far from being the case,” he said.
“The Cambridge Analytica scandal was already in breach of the current Data Protection Directive, and would also be contrary to the GDPR, which is soon to be implemented. I expect the EU Data Protection Authorities to take appropriate action to enforce the law.”
Damian Collins, chair of the UK parliament’s DCMS committee, which has thrice tried and failed to get Zuckerberg to appear before it, did not mince his words at all. Albeit he has little reason to, having been so thoroughly rejected by the Facebook founder — and having accused the company of a pattern of evasive behavior to its CTO’s face — there’s clearly not much to hold out for now.
“What a missed opportunity for proper scrutiny on many crucial questions raised by the MEPs. Questions were blatantly dodged on shadow profiles, sharing data between WhatsApp and Facebook, the ability to opt out of political advertising and the true scale of data abuse on the platform,” said Collins in another reaction statement after the meeting. “Unfortunately the format of questioning allowed Mr Zuckerberg to cherry-pick his responses and not respond to each individual point.
“I echo the clear frustration of colleagues in the room who felt the discussion was shut down,” he added, ending with a fourth (doubtless equally forlorn) request for Zuckerberg to appear in front of the DCMS Committee to “provide Facebook users the answers they deserve”.
In the latter stages of today’s EU parliament session several MEPs — clearly very exasperated by the straightjacked format — resorted to heckling Zuckerberg to press for answers he had not given them.
“Shadow profiles,” interjected one, seizing on a moment’s hesitation as Zuckerberg sifted his notes for the next talking point. “Compensation,” shouted another, earning a snort of laughter from the CEO and some more theatrical note flipping to buy himself time.
Then, appearing slightly flustered, Zuckerberg looked up at one of the hecklers and said he would engage with his question — about shadow profiles (though Zuckerberg dare not speak that name, of course, given he claims not to recognize it) — arguing Facebook needs to hold onto such data for security purposes.
Zuckerberg did not specify, as MEPs had asked him to, whether Facebook uses data about non-users for any purposes other than the security scenario he chose to flesh out (aka “keeping bad content out”, as he put it).
He also ignored a second follow-up pressing him on how non-users can “stop that data being transferred”.
“On the security side we think it’s important to keep it to protect people in our community,” Zuckerberg said curtly, before turning to his lawyer for a talking point prompt (couched as an ask if there are “any other themes we wanted to get through”).
His lawyer hissed to steer the conversation back to Cambridge Analytica — to Facebook’s well-trodden PR about how they’re “locking down the platform” to stop any future data heists — and the Zuckbot was immediately back in action regurgitating his now well-practiced crisis PR around the scandal.
What was very clearly demonstrated during today’s session was the Facebook founder’s preference for control — that’s to say control which he is exercising.
Hence the fixed format of the meeting, which had been negotiated prior to Facebook agreeing to meet with EU politicians, and which clearly favored the company by allowing no formal opportunity for follow ups from MEPs.
Zuckerberg also tried several times to wrap up the meeting — by insinuating and then announcing time was up. MEPs ignored these attempts, and Zuckerberg seemed most uncomfortable at not having his orders instantly carried out.
Instead he had to sit and watch a micro negotiation between the EU parliament’s president and the political groups over whether they would accept written answers to all their specific questions from Facebook — before he was publicly put on the spot by president Antonio Tajani to agree to provide the answers in writing.
Although, as Collins has already warned MEPs, Facebook has had plenty of practice at generating wordy but empty responses to politicians’ questions about its business processes — responses which evade the spirit and specifics of what’s being asked.
The self-control on show from Zuckerberg today is certainly not the kind of guardrails that European politicians increasingly believe social media needs. Self-regulation, observed several MEPs to Zuckerberg’s face, hasn’t worked out so well has it?
The first MEP to lay out his questions warned Zuckerberg that apologizing is not enough. Another pointed out he’s been on a contrition tour for about 15 years now.
Facebook needs to make a “legal and moral commitment” to the EU’s fundamental values, he was told by Moraes. “Remember that you’re here in the European Union where we created GDPR so we ask you to make a legal and moral commitment, if you can, to uphold EU data protection law, to think about ePrivacy, to protect the privacy of European users and the many millions of European citizens and non-Facebook users as well,” said the Libe committee chair.
But self-regulation — or, the next best thing in Zuckerberg’s eyes: ‘Facebook-shaped regulation’ — was what he had come to advocate for, picking up on the MEPs’ regulation “theme” to respond with the same line he fed to Congress: “I don’t think the question here is whether or not there should be regulation. I think the question is what is the right regulation.”
“The Internet is becoming increasingly important in people’s lives. Some sort of regulation is important and inevitable. And the important thing is to get this right,” he continued. “To make sure that we have regulatory frameworks that help protect people, that are flexible so that they allow for innovation, that don’t inadvertently prevent new technologies like AI from being able to develop.”
He even brought up startups — claiming ‘bad regulation’ (I paraphrase) could present a barrier to the rise of future dormroom Zuckerbergs.
Of course he failed to mention how his own dominant platform is the attention-sapping, app gobbling elephant in the room crowding out the next generation of would-be entrepreneurs. But MEPs’ concerns about competition were clear.
Instead of making friends and influencing people in Brussels, Zuckerberg looks to have delivered less than if he’d stayed away — angering and alienating the very people whose job it will be to amend the EU legislation that’s coming down the pipe for his platform.
Ironically one of the few specific questions Zuckerberg chose to answer was a false claim by MEP Nigel Farage — who had wondered whether Facebook is still a “neutral political platform”, griping about drops in engagement for rightwing entities ever since Facebook’s algorithmic changes in January, before claiming, erroneously, that Facebook does not disclose the names of the third party fact checkers it uses to help it police fake news.
So — significantly, and as was also evident in the US Senate and Congress — Facebook was taking flak from both left and right of political spectrum, implying broad, cross-party support for regulating these algorithmic platforms.
Actually Facebook does disclose those fact checking partnerships. But it’s pretty telling that Zuckerberg chose to expend some of his oh-so-slender speaking time to debunk something that really didn’t merit the breath.
Farage had also claimed, during his three minutes, that without “Facebook and other forms of social media there is no way that Brexit or Trump or the Italian elections could ever possibly have happened”.
Funnily enough Zuckerberg didn’t make time to comment on that.
Baidu, the Chinese search giant, is spinning out its business unit responsible for utility apps and its mobile ad business to sharpen its focus on artificial intelligence.
As part of the spin-out, Baidu is selling a large chunk of its equity in the ‘Global DU’ business to as-yet-undisclosed investors. The plan is to sell “a majority equity stake” in order to take Global DU independent. Once the deal is completed — it is targeted at a Q3 2018 timeframe — Baidu’s share of the business will drop to around 34 percent. Further, the business is likely to raise additional capital for growth.
Herman Yu, Baidu CFO, said this latest spin-out will give Global Du “autonomy and agility in its operation.”
It will also allow Baidu to focus more keenly on artificial intelligence. The firm said it will set up a new global business unit around its AI-powered services, including recommendation engine PopIn, keyboard app Simeji and other services in the U.S. and Southeast Asia. The plan is to allow these services to work more closely with Baidu’s AI labs, which include locations in Silicon Valley and Seattle.
Despite the AI push, Baidu has suffered as key personnel departed over the past year.
Last week, Qi Lu, Baidu’s president and COO who is also its highest-ranking AI specialist, departed the company due to personal reasons. The exit was unexpected, particularly since the former Microsoft executive only took the role less than two years ago.
India, second in population only to China with rapidly-growing internet access, is an obvious place to look, and would-be pretender to the Toutiao crown has been found in the shape of NewsDog, a Chinese company that stumbled on success in India. Today, NewsDog announced a $50 million Series C round led by Chinese internet giant Tencent.
Toutiao is a phenomenon in China. The app has around 200 million daily users, and it is one of the few new tech products to emerge in a China where Tencent and Alibaba dominate the consumer app landscape. Point in case, it is so mainstream now that it has even run into issues with China’s internet censors. Toutiao is essentially a news aggregation service that lets consumers catch their daily reads and discover stories with an experience tailored to their habits and likes.
That’s very much the style of NewsDog, which claims over 50 million users. The service has branched out to cover 10 of Indians many languages, while it recently established a platform — ‘WeMedia’ — that augments its content aggregation by allowing users to submit stories, too.
This round is a major milestone for the company. In a competitive environment, it is the largest fundraising round from a news app company in India while it more obviously brings Tencent, the $500 billion tech giant, on board with its experience and support. Other investors include Chinese VCs Danhua Capital (DHVC) and Legend Capital as well as Chinese mobile app firm DotC United.
NewsDog’s competition includes Dailyhunt — which is backed by Toutiao-owner Bytedance — Inshorts, which counts Tiger Global among its investors, and NewsPoint, which is owned by media firm Times Internet.
One other competition is UC News, a service from Alibaba-owned UC Web, which, like NewsDog, is Chinese.
NewsDog was launched in 2016 by CEO Forrest Chen Yukun, a computer science graduate from Tsinghua University graduate, and Yi Ma, who holds a PhD from Princeton University and previously worked at Baidu and Goldman Sachs .
Data from App Annie shows that NewsDog is the top news app in the Google Play Store in India — Android is the country’s dominant operating system — ahead of Dailyhunt and NewsPoint in second and third, respectively. NewsDog plans to use this new funding to pull further ahead of the competition by focusing on adding more languages and deepening its content library.
The company said it is already using machine learning to help produce an experience that is customized to users — the experience that Toutiao pioneered in China — and it plans to double down on that.
“Poly culture and multiple languages make content matching an incredibly hard problem,” Chen said in a statement. “So far, we have made good initial progress but content business is like an endless journey. There is no finish line, you have to just keep running.”
NewsDog is aiming to reach 100 million users as its next milestone as India’s internet population surges. The country is tipped to reach 500 million internet users by June 2018, according to a report from the Internet and Mobile Association of India (IAMAI) and Kantar IMRB. That’s up from 481 million six months prior, but internet penetration in rural areas is at just 20 percent compared with 65 percent in urban India which indicates even more growth potential.
For Tencent, meanwhile, this investment is another upping of its pace in India.
One of the more interesting applications of AI to the world of advertising and marketing has been in how it’s being used to help measure and ultimately shape campaigns. Now, a company providing the technology to do that has raised a round both to expand its business in adtech as well as to tackle new applications in healthcare and education.
Realeyes, a London-based startup that uses computer vision to read a person’s emotional responses when they are watching a video as short as six seconds long, and then using predictive analytics to help map that reading to the video to provide feedback on its effectiveness, has raised $16.2 million in funding, money that it plans to use to expand in engineering and business development.
The rise of “smart” and connected hardware that picks up data as much as produces it is the opportunity that Realeyes is tapping. “We are surrounded by devices with cameras and microphones in them,” CEO and founder Mihkel Jäätma said in an interview.
The Series A round comes after a strong run of growth at the company. It says that revenues have shot up 932 percent in the last four years, and it has added customers like Coca Cola, Mars, Publicis, Turner and Oath (which also owns TechCrunch) to its books.
Realeyes is not wasting time in bringing on extra talent to support the expansion. Barry Coleman, formerly at LootCrate, is coming on as COO. And Maja Pantic, a professor of affective and behavioural computing at Imperial College London who had been on the Realeyes Advisory Board, is getting “a more hands-on role.” Both will report to Jäätma, who started the company while still a student at Oxford.
The round was led by Draper Esprit, with participation also from Karma Ventures and Harbert European Growth Capital.
Draper Esprit recently had a big win in the area of marketing tech with another UK startup in its portfolio, when Oracle acquired Grapeshot for what one reliable source said was up to $400 million including earn-outs; Grapeshot had only raised $22 million in total.
Karma, meanwhile, is another notable investor: formed by the founding engineers behind Skype, the group announced a €70 million fund earlier this month focused on “deep tech”, and this is one of the first investments to come out of that fund. (And in addition to the affinity for what Realeyes is doing, there is another connection: those early Skype engineers are Estonian, and Realeyes’ Jäätma also hails from there.)
There have been a number of notable startups using advances in computer vision and big data analytics to gather more information about how people are responding to ad campaigns.
Jäätma said that Realeyes stands apart from the pack for a few reasons. The first is the size of the company’s database. “We have hand-labelled over 15 million frames of naturally-occurring emotions, with up to seven human assessments for each frame over the last decade,” he said.
The second is what the company does with that data, specifically in relation to video. “Commercially, we have invested more in the full platform around the core measurement technology,” he said, which includes predictive analytics that even provide data on how users’ responses to videos will impact sales of the item being advertised. “Marketers can actually use emotional intelligence to drive business outcomes.”
“Realeyes is changing the way marketers can measure impact through their cutting-edge technology. Artificial Intelligence will continue to change the way we understand each other – even our emotions,” said Stuart Chapman, COO at Draper Esprit, in a statement. “Realeyes is well positioned to fundamentally change the way the advertising industry can be more engaging to its audiences.”
“Driving business outcomes” is also very important right now because there has been a lot of scrutiny over how users are tracked around the web, this could give brands, agencies, media platforms and others a key way to gleaning effectiveness without having to do get too invasive.
(And to be completely clear, Jäätma said that Realeyes’ technology is applied on small pools of users, and only by way of opted-in panels. In other words, this is not tech that will suddenly start recording your responses to online ads you might come across in your daily web browsing.)
And the fact that Realeyes is able to draw conclusions from even short video clips is also interesting: it plays into the fact that a lot of video ads today are a turn-off if they last too long, and so marketers are looking for shorter and more engaging formats to offset that issue.
Notably, the company currently is not working with Google’s YouTube, but it has been running tests with other video players, a pilot with Virool, to see how opt-in campaigns on high-traffic sites might work. (You could, for example, imagine something like this being used in a consumer survey campaign format, which could also potentially position Google as a competitor to Realeyes, too.)
While a lot of Realeyes’ focus will continue on the marketing and advertising world, it’s also starting to look at other areas — specifically health tech — specifically looking at how to help detect depression — and education, here specifically looking at ways of improving how students stay engaged with digital learning content.
Even with a squarely opt-in model behind what Realeyes does now, there are, of course, still challenges that the company will need to overcome.
For example, one big story in the news in recent weeks has been about resignations of employees at Google who were unhappy with how the company would potentially start working on AI projects with government groups. Even if Realeyes is largely still in the domain of deep tech — more than half of its employees are engineers and working in R&D — and figuring out new ways of gleaning psychological information from small fragments of video and facial responses, there is will always be a question mark for any new tech company about how it would feel about how this might work across any and all applications.
“It’s a great question but it hasn’t come up with us yet,” Jäätma said. “We’re not having such conversations and use cases. We’re still a 65-person company and that is an important enough discussion that we would take a [collective] point of view, but we haven’t had to yet. The smartest people in science are also pondering those things and don’t have the final answers.”