natural language processing
Auto Added by WPeMatico
Twitter bags deep learning talent behind London startup, Fabula AI
Twitter has just announced it has picked up London-based Fabula AI. The deep learning startup has been developing technology to try to identify online disinformation by looking at patterns in how fake stuff vs genuine news spreads online — making it an obvious fit for the rumor-riled social network.
Social media giants remain under increasing political pressure to get a handle on online disinformation to ensure that manipulative messages don’t, for example, get a free pass to fiddle with democratic processes.
Twitter says the acquisition of Fabula will help it build out its internal machine learning capabilities — writing that the UK startup’s “world-class team of machine learning researchers” will feed an internal research group it’s building out, led by Sandeep Pandey, its head of ML/AI engineering.
This research group will focus on “a few key strategic areas such as natural language processing, reinforcement learning, ML ethics, recommendation systems, and graph deep learning” — now with Fabula co-founder and chief scientist, Michael Bronstein, as a leading light within it.
Bronstein is chair in machine learning & pattern recognition at Imperial College, London — a position he will remain while leading graph deep learning research at Twitter.
Fabula’s chief technologist, Federico Monti — another co-founder, who began the collaboration that underpin’s the patented technology with Bronstein while at the University of Lugano, Switzerland — is also joining Twitter.
“We are really excited to join the ML research team at Twitter, and work together to grow their team and capabilities. Specifically, we are looking forward to applying our graph deep learning techniques to improving the health of the conversation across the service,” said Bronstein in a statement.
“This strategic investment in graph deep learning research, technology and talent will be a key driver as we work to help people feel safe on Twitter and help them see relevant information,” Twitter added. “Specifically, by studying and understanding the Twitter graph, comprised of the millions of Tweets, Retweets and Likes shared on Twitter every day, we will be able to improve the health of the conversation, as well as products including the timeline, recommendations, the explore tab and the onboarding experience.”
Terms of the acquisition have not been disclosed.
We covered Fabula’s technology and business plan back in February when it announced its “new class” of machine learning algorithms for detecting what it colloquially badged ‘fake news’.
Its approach to the problem of online disinformation looks at how it spreads on social networks — and therefore who is spreading it — rather than focusing on the content itself, as some other approaches do.
Fabula has patented algorithms that use the emergent field of “Geometric Deep Learning” to detect online disinformation — where the datasets in question are so large and complex that traditional machine learning techniques struggle to find purchase. Which does really sound like a patent designed with big tech in mind.
Fabula likens how ‘fake news’ spreads on social media vs real news as akin to “a very simplified model of how a disease spreads on the network”.
One advantage of the approach is it looks to be language agnostic (at least barring any cultural differences which might also impact how fake news spread).
Back in February the startup told us it was aiming to build an open, decentralised “truth-risk scoring platform” — akin to a credit referencing agency, just focused on content not cash.
It’s not clear from Twitter’s blog post whether the core technologies it will be acquiring with Fabula will now stay locked up within its internal research department — or be shared more widely, to help other platforms grappling with online disinformation challenges.
The startup had intended to offer an API for platforms and publishers later this year.
But of course building a platform is a major undertaking. And, in the meanwhile, Twitter — with its pressing need to better understand the stuff its network spreads — came calling.
A source close to the matter told us that Fabula’s founders decided that selling to Twitter instead of pushing for momentum behind a vision of a decentralized, open platform because the exit offered them more opportunity to have “real and deep impact, at scale”.
Though it is also still not certain what Twitter will end up doing with the technology it’s acquiring. And it at least remains possible that Twitter could choose to make it made open across platforms.
“That’ll be for the team to figure out with Twitter down the line,” our source added.
A spokesman for Twitter did not respond directly when we asked about its plans for the patented technology but he told us: “There’s more to come on how we will integrate Fabula’s technology where it makes sense to strengthen our systems and operations in the coming months. It will likely take us some time to be able to integrate their graph deep learning algorithms into our ML platform. We’re bringing Fabula in for the team, tech and mission, which are all aligned with our top priority: Health.”
Powered by WPeMatico
Live transcription and captioning in Android are a boon to the hearing-impaired
A set of new features for Android could alleviate some of the difficulties of living with hearing impairment and other conditions. Live transcription, captioning and relay use speech recognition and synthesis to make content on your phone more accessible — in real time.
Announced today at Google’s I/O event in a surprisingly long segment on accessibility, the features all rely on improved speech-to-text and text-to-speech algorithms, some of which now run on-device rather than sending audio to a data center to be decoded.
The first feature to be highlighted, live transcription, was already mentioned by Google. It’s a simple but very useful tool: open the app and the device will listen to its surroundings and simply display as text on the screen any speech it recognizes.
We’ve seen this in translator apps and devices, like the One Mini, and the meeting transcription highlighted yesterday at Microsoft Build. One would think that such a straightforward tool is long overdue, but, in fact, everyday circumstances like talking to a couple of friends at a cafe can be remarkably difficult for natural language systems trained on perfectly recorded single-speaker audio. Improving the system to the point where it can track multiple speakers and display accurate transcripts quickly has no doubt been a challenge.
Another feature enabled by this improved speech recognition ability is live captioning, which essentially does the same thing as above, but for video. Now when you watch a YouTube video, listen to a voice message or even take a video call, you’ll be able to see what the person in it is saying, in real time.
That should prove incredibly useful not just for the millions of people who can’t hear what’s being said, but also those who don’t speak the language well and could use text support, or anyone watching a show on mute when they’re supposed to be going to sleep, or any number of other circumstances where hearing and understanding speech just isn’t the best option.
Captioning phone calls is something CEO Sundar Pichai said is still under development, but the “live relay” feature they demoed onstage showed how it might work. A person who is hearing-impaired or can’t speak will certainly find an ordinary phone call to be pretty worthless. But live relay turns the call immediately into text, and immediately turns text responses into speech the person on the line can hear.
Live captioning should be available on Android Q when it releases, with some device restrictions. Live transcribe is available now, but a warning states that it is currently in development. Live relay is yet to come, but showing it onstage in such a complete form suggests it won’t be long before it appears.
Powered by WPeMatico
The team behind Baidu’s first smart speaker is now using AI to make films
The HBO sci-fi blockbuster Westworld has been an inspiring look into what humanlike robots can do for us in the meatspace. While current technologies are not quite advanced enough to make Westworld a reality, startups are attempting to replicate the sort of human-robot interaction it presents in virtual space.
Rct studio, which just graduated from Y Combinator and ranked among TechCrunch’s nine favorite picks from the batch, is one of them. The “Westworld” in the TV series, a far-future theme park staffed by highly convincing androids, lets visitors live out their heroic and sadistic fantasies free of consequences.
There are a few reasons why rct studio, which is keeping mum about the meaning of its deliberately lower-cased name for later revelation, is going for the computer-generated world. Besides the technical challenge, playing a fictional universe out virtually does away the geographic constraint. The Westworld experience, in contrast, happens within a confined, meticulously built park.
“Westworld is built in a physical world. I think in this age and time, that’s not what we want to get into,” Xinjie Ma, who heads up marketing for rct, told TechCrunch. “Doing it in the physical environment is too hard, but we can build a virtual world that’s completely under control.”
Rct studio wants to build the Westworld experience in virtual worlds. / Image: rct studio
The startup appears suitable to undertake the task. The eight-people team is led by Cheng Lyu, the 29-year-old entrepreneur who goes by Jesse and helped Baidu build up its smart speaker unit from scratch after the Chinese search giant acquired his voice startup Raven in 2017. Along with several of Raven’s core members, Lyu left Baidu in 2018 to start rct.
“We appreciate a lot the support and opportunities given by Baidu and during the years we have grown up dramatically,” said Ma, who previously oversaw marketing at Raven.
Let AI write the script
Immersive films, or games, depending on how one wants to classify the emerging field, are already available with pre-written scripts for users to pick from. Rct wants to take the experience to the next level by recruiting artificial intelligence for screenwriting.
At the center of the project is the company’s proprietary engine, Morpheus. Rct feeds it mountains of data based on human-written storylines so the characters it powers know how to adapt to situations in real time. When the codes are sophisticated enough, rct hopes the engine can self-learn and formulate its own ideas.
“It takes an enormous amount of time and effort for humans to come up with a story logic. With machines, we can quickly produce an infinite number of narrative choices,” said Ma.
To venture through rct’s immersive worlds, users wear a virtual reality headset and control their simulated self via voice. The choice of audio came as a natural step given the team’s experience with natural language processing, but the startup also welcomes the chance to develop new devices for more lifelike journeys.
“It’s sort of like how the film Ready Player One built its own gadgets for the virtual world. Or Apple, which designs its own devices to carry out superior software experience,” explained Ma.
On the creative front, rct believes Morpheus could be a productivity tool for filmmakers as it can take a story arc and dissect it into a decision-making tree within seconds. The engine can also render text to 3D images, so when a filmmaker inputs the text “the man throws the cup to the desk behind the sofa,” the computer can instantly produce the corresponding animation.
Path to monetization
Investors are buying into rct’s offering. The startup is about to close its Series A funding round just months after banking seed money from Y Combinator and Chinese venture capital firm Skysaga, the startup told TechCrunch.
The company has a few imminent tasks before achieving its Westworld dream. For one, it needs a lot of technical talent to train Morpheus with screenplay data. No one on the team had experience in filmmaking, so it’s on the lookout for a creative head who appreciates AI’s application in films.
Rct studio’s software takes a story arc and dissects it into a decision-making tree within seconds. / Image: rct studio
“Not all filmmakers we approach like what we do, which is understandable because it’s a very mature industry, while others get excited about tech’s possibility,” said Ma.
The startup’s entry into the fictional world was less about a passion for films than an imperative to shake up a traditional space with AI. Smart speakers were its first foray, but making changes to tangible objects that people are already accustomed to proved challenging. There has been some interest in voice-controlled speakers, but they are far from achieving ubiquity. Then movies crossed the team’s mind.
“There are two main routes to make use of AI. One is to target a vertical sector, like cars and speakers, but these things have physical constraints. The other application, like Alpha Go, largely exists in the lab. We wanted something that’s both free of physical limitation and holds commercial potential.”
The Beijing and Los Angeles-based startup isn’t content with just making the software. Eventually, it wants to release its own films. The company has inked a long-term partnership with Future Affairs Administration, a Chinese sci-fi publisher representing about 200 writers, including the Hugo award-winning Cixin Liu. The pair is expected to start co-producing interactive films within a year.
Rct’s path is reminiscent of a giant that precedes it: Pixar Animation Studios . The Chinese company didn’t exactly look to the California-based studio for inspiration, but the analog was a useful shortcut to pitch to investors.
“A confident company doesn’t really draw parallels with others, but we do share similarities to Pixar, which also started as a tech company, publishes its own films, and has built its own engine,” said Ma. “A lot of studios are asking how much we price our engine at, but we are targeting the consumer market. Making our own films carry so many more possibilities than simply selling a piece of software.”
Powered by WPeMatico
Skymind raises $11.5M to bring deep learning to more enterprises
Skymind, a Y Combinator-incubated AI platform that aims to make deep learning more accessible to enterprises, today announced that it has raised an $11.5 million Series A round led by TransLink Capital, with participation from ServiceNow, Sumitomo’s Presidio Ventures, UpHonest Capital and GovTech Fund. Early investors Y Combinator, Tencent, Mandra Capital, Hemi Ventures, and GMO Ventures, also joined the round/ With this, the company has now raised a total of $17.9 million in funding.
The inclusion of TransLink Capital gives a hint as to how the company is planning to use the funding. One of TransLink’s specialties is helping entrepreneurs develop customers in Asia. Skymind believes that it has a major opportunity in that market, so having TransLink lead this round makes a lot of sense. Skymind also plans to use the round to build out its team in North America and fuel customer acquisition there.
“TransLink is the perfect lead for this round, because they know how to make connections between North America and Asia,” Skymind CEO Chris Nicholson told me. “That’s where the most growth is globally, and there are a lot of potential synergies. We’re also really excited to have strategic investors like ServiceNow and Sumitomo’s Presidio Ventures backing us for the first time. We’re already collaborating with ServiceNow, and Skymind software will be part of some powerful new technologies they roll out.”
It’s no secret that enterprises know that they have to adapt AI in some form but are struggling with figuring out how to do so. Skymind’s tools, including its core SKIL framework, allow data scientists to create workflows that take them from ingesting the data to cleaning it up, training their models and putting them into production. The promise here is that Skymind’s tools eliminate the gap that often exists between the data scientists and IT.
“The two big opportunities with AI are better customer experiences and more efficiency, and both are based on making smarter decisions about data, which is what AI does,” said Nicholson. “The main types of data that matter to enterprises are text and time series data (think web logs or payments). So we see a lot of demand for natural-language processing and for predictions around streams of data, like logs.”
Current Skymind customers include the likes of ServiceNow and telco company Orange, while some of its technology partners that integrate its services into their portfolio include Cisco and SoftBank .
It’s worth noting that Skymind is also the company behind Deeplearning4j, one of the most popular open-source AI tools for Java. The company is also a major contributor to the Python-based Keras deep learning framework.
Powered by WPeMatico
Google’s new voice recognition system works instantly and offline (if you have a Pixel)
Voice recognition is a standard part of the smartphone package these days, and a corresponding part is the delay while you wait for Siri, Alexa or Google to return your query, either correctly interpreted or horribly mangled. Google’s latest speech recognition works entirely offline, eliminating that delay altogether — though of course mangling is still an option.
The delay occurs because your voice, or some data derived from it anyway, has to travel from your phone to the servers of whoever operates the service, where it is analyzed and sent back a short time later. This can take anywhere from a handful of milliseconds to multiple entire seconds (what a nightmare!), or longer if your packets get lost in the ether.
Why not just do the voice recognition on the device? There’s nothing these companies would like more, but turning voice into text on the order of milliseconds takes quite a bit of computing power. It’s not just about hearing a sound and writing a word — understanding what someone is saying word by word involves a whole lot of context about language and intention.
Your phone could do it, for sure, but it wouldn’t be much faster than sending it off to the cloud, and it would eat up your battery. But steady advancements in the field have made it plausible to do so, and Google’s latest product makes it available to anyone with a Pixel.
Google’s work on the topic, documented in a paper here, built on previous advances to create a model small and efficient enough to fit on a phone (it’s 80 megabytes, if you’re curious), but capable of hearing and transcribing speech as you say it. No need to wait until you’ve finished a sentence to think whether you meant “their” or “there” — it figures it out on the fly.
So what’s the catch? Well, it only works in Gboard, Google’s keyboard app, and it only works on Pixels, and it only works in American English. So in a way this is just kind of a stress test for the real thing.
“Given the trends in the industry, with the convergence of specialized hardware and algorithmic improvements, we are hopeful that the techniques presented here can soon be adopted in more languages and across broader domains of application,” writes Google, as if it is the trends that need to do the hard work of localization.
Making speech recognition more responsive, and to have it work offline, is a nice development. But it’s sort of funny considering hardly any of Google’s other products work offline. Are you going to dictate into a shared document while you’re offline? Write an email? Ask for a conversion between liters and cups? You’re going to need a connection for that! Of course this will also be better on slow and spotty connections, but you have to admit it’s a little ironic.
Powered by WPeMatico
WellSaid aims to make natural-sounding synthetic speech a credible alternative to real humans
Many things are better said than read, but the best voice tech out there seems to be reserved for virtual assistants, not screen readers or automatically generated audiobooks. WellSaid wants to enable any creator to use quality synthetic speech instead of a human voice — perhaps even a synthetic version of themselves.
There’s been a series of major advances in voice synthesis over the last couple of years as neural network technology improves on the old highly manual approach. But Google, Apple and Amazon seem unwilling to make their great voice tech available for anything but chirps from your phone or home hub.
As soon as I heard about WaveNet, and later Tacotron, I tried to contact the team at Google to ask when they’d get to work producing natural-sounding audiobooks for everything on Google Books, or as a part of AMP, or make it an accessibility service, and so on. Never heard back. I considered this a lost opportunity, as there are many out there who need such a service.
So I was pleased to hear that WellSaid is taking on this market, after a fashion, anyway. The company is the first to launch from the Allen Institute for AI (AI2) incubator program announced back in 2017. They do take their time!
Talk the talk
I talked with the co-founders CEO Matt Hocking and CTO Michael Petrochuk, who explained why they went about creating a whole new system for voice synthesis. The basic problem, they said, is that existing systems not only rely on a lot of human annotation to sound right, but they “sound right” the exact same way every time. You can’t just feed it a few hours of audio and hope it figures out how to inflect questions or pause between list items — much of this stuff has to be spelled out for them. The end result, however, is highly efficient.
“Their goal is to make a small model for cheap [i.e. computationally] that pronounces things the same way every time. It’s this one perfect voice,” said Petrochuk. “We took research like Tacotron and pushed it even further — but we’re not trying to control speech and enforce this arbitrary structure on it.”
“When you think about the human voice, what makes it natural, kind of, is the inconsistencies,” said Hocking.
And where better to find inconsistencies than in humans? The team worked with a handful of voice actors to record dozens of hours of audio to feed to the system. There’s no need to annotate the text with “speech markup language” to designate parts of sentences and so on, Petrochuk said: “We discovered how to train off of raw audiobook data, without having to do anything on top of that.”
So WellSaid’s model will often pronounce the same word differently, not because a carefully manicured manual model of language suggested it do so, but because the person whose vocal fingerprint it is imitating did so.
And how does that work, exactly? That question seems to dip into WellSaid’s secret sauce. Their model, like any deep learning system, is taking innumerable inputs into account and producing an output, but it is larger and more far-reaching than other voice synthesis systems. Things like cadence and pronunciation aren’t specified by its overseers but extracted from the audio and modeled in real time. Sounds a bit like magic, but that’s often the case when it comes to bleeding-edge AI research.
It runs on a CPU in real time, not on a GPU cluster somewhere, so it can be done offline as well. This is a feat in itself, as many voice synthesis algorithms are quite resource-heavy.
What matters is that the voice produced can speak any text in a very natural-sounding way. Here’s the first bit of an article — alas, not one of mine, which would have employed more mellifluous circumlocutions — read by Google’s WaveNet, then by two of WellSaid’s voices.
The latter two are definitely more natural sounding than the first. On some phrases the voices may be nearly indistinguishable from their originals, but in most cases I feel sure I could pick out the synthetic voice in a few words.
That it’s even close, however, is an accomplishment. And I can certainly say that if I was going to have an article read to my by one of these voices, it would be WellSaid’s. Naturally it can also be tweaked and iterated, or effects applied to further manipulate the sound, as with any voice performance. You didn’t think those interviews you hear on NPR are unedited, did you?
The goal at first is to find the creatives whose work would be improved or eased by adding this tool to their toolbox.
“There are a lot of people who have this need,” explained Hocking. “A video producer who doesn’t have the budget to hire a voice actor; someone with a large volume of content that has to be iterated on rapidly; if English is a second language, this opens up a lot of doors; and some people just don’t have a voice for radio.”
It would be nice to be able to add voice with a click rather than just have block text and royalty-free music over a social ad (think the admen):
I asked about the reception among voice actors, who of course are essentially being asked to train their own replacements. They said that the actors were actually positive about it, thinking of it as something like stock photography for voice; get a premade product for cheap, and if you like it, pay the creator for the real thing. Although they didn’t want to prematurely lock themselves into future business models, they did acknowledge that revenue share with voice actors was a possibility. Payment for virtual representations is something of a new and evolving field.
A closed beta launches today, which you can sign up for at the company’s site. They’re going to be launching with five voices to start, with more voices and options to come as WellSaid’s place in the market becomes clear. Part of that process will almost certainly be inclusion in tools used by the blind or otherwise disabled, as I have been hoping for years.
Sounds familiar
And what comes after that? Making synthetic versions of users’ voices, of course. No brainer! But the two founders cautioned that’s a ways off for several reasons, even though it’s very much a possibility.
“Right now we’re using about 20 hours of data per person, but we see a future where we can get it down to one or two hours while maintaining a premium lifelike quality to the voice,” said Petrochuk.
“And we can build off existing data sets, like where someone has a back catalog of content,” added Hocking.
The trouble is that the content may not be exactly right for training the deep learning model, which advanced as it is can no doubt be finicky. There are dials and knobs to tweak, of course, but they said that fine-tuning a voice is more a matter of adding corrective speech, perhaps having the voice actor reading a specific script that props up the sounds or cadences that need a boost.
They compared it with directing such an actor rather than adjusting code. You don’t, after all, tell an actor to increase the pauses after commas by 8 percent or 15 milliseconds, whichever is longer. It’s more efficient to demonstrate for them: “say it like this.”
Even so, getting the quality just right with limited and imperfect training data is a challenge that will take some serious work if and when the team decides to take it on.
But as some of you may have noticed, there are also some parallels to the unsavory world of “deepfakes.” Download a dozen podcasts or speeches and you’ve got enough material to make a passable replica of someone’s voice, perhaps a public figure. This of course has a worrying synergy with the existing ability to fake video and other imagery.
This is not news to Hocking and Petrochuk. If you work in AI, this kind of thing is sort of inevitable.
“This is a super important question and we’ve considered it a lot,” said Petrochuk. “We come from AI2, where the motto is ‘AI for the common good.’ That’s something we really subscribe to, and that differentiates us from our competitors who made Barack Obama voices before they even had an MVP [minimum viable product]. We’re going to watch closely to make sure this isn’t being used negatively, and we’re not launching with the ability to make a custom voice, because that would let anyone create a voice from anyone.”
Active monitoring is just about all anyone with a potentially troubling AI technology can be expected to do — though they are looking at mitigation techniques that could help identify synthetic voices.
With the ongoing emphasis on multimedia presentation of content and advertising rather than written, WellSaid seems poised to make an early play in a growing market. As the product evolves and improves, it’s easy to picture it moving into new, more constrained spaces, like time-shifting apps (instant podcast with five voices to choose from!) and even taking over territory currently claimed by voice assistants. Sounds good to me.
Powered by WPeMatico
Gong.io nabs $40M investment to enhance CRM with voice recognition
With traditional CRM tools, sales people add basic details about the companies to the database, then a few notes about their interactions. AI has helped automate some of that, but Gong.io wants to take it even further using voice recognition to capture every word of every interaction. Today, it got a $40 million Series B investment.
The round was led by Battery Ventures, with existing investors Norwest Venture Partners, Shlomo Kramer, Wing Venture Capital, NextWorld Capital and Cisco Investments also participating. Battery general partner Dharmesh Thakker will join the startup’s board under the terms of the deal. Today’s investment brings the total raised so far to $68 million, according to the company.
Indeed, $40 million is a hefty Series B, but investors see a tool that has the potential to have a material impact on sales, or at least give management a deeper understanding of why a deal succeeded or failed using artificial intelligence, specifically natural language processing.
Company co-founder and CEO Amit Bendov says the solution starts by monitoring all customer-facing conversation and giving feedback in a fully automated fashion. “Our solution uses AI to extract important bits out of the conversation to provide insights to customer-facing people about how they can get better at what they do, while providing insights to management about how staff is performing,” he explained. It takes it one step further by offering strategic input like how your competitors are trending or how are customers responding to your products.
Screenshot: Gong.io
Bendov says he started the company because he has had this experience at previous startups where he wants to know more about why he lost a sale, but there was no insight from looking at the data in the CRM database. “CRM could tell you what customers you have, how many sales you’re making, who is achieving quota or not, but never give me the information to rationalize and improve operations,” he said.
The company currently has 350 customers, a number that has more than tripled since the end of 2017 when it had 100. He says it’s not only that it’s adding new customers, existing ones are expanding, and he says that there is almost zero churn.
Today, Gong has 120 employees, with headquarters in San Francisco and a 55-person R&D team in Israel. Bendov expects the number of employees to double over the next year with the new influx of money to keep up with the customer growth.
Powered by WPeMatico
Chorus.ai rings up $33M for its platform that analyses sales calls to close more deals
Chorus.ai, a service that listens to sales calls in real time, and then transcribes and analyses them to give helpful tips to the salesperson, has raised $33 million to double down on the current demand for more AI-based tools in the enterprise.
The Series B is being led by Georgian Partners, with participation also from Redpoint Ventures and Emergence Capital, previous investors that backed Israeli-founded, SF-based Chorus.ai in its $16 million Series A two years ago.
In the gap between then and now, the startup has seen strong growth, listening in to some 5 million calls, and performing hundreds of thousands of hours of transcriptions for around 200 customers, including Adobe, Zoom, and Outreach (among others that it will not name).
Micha Breakstone, the co-founder (who has a pretty long history in conversational AI, heading up R&D at Ginger Software and then Intel after it acquired the startup; and before that building the tech that eventually became Summly and got acquired by Yahoo, among other roles), says that while the platform gives information and updates to salespeople in real time, much of the focus today is on providing information to users post-conversation, based on both audio and video calls.
One of its big areas is “smart themes” — patterns and rules Chorus has learned through all those calls. For example, it has identified what kind of language the most successful sales people are using and in turn prompts those who are less successful to use it more. Two general tips Breakstone told me about: using more collaborative terms like we and us; and giving more backstory to clients, although there will be more specific themes and approaches based on Chorus’s specific customers and products.
“I’d say we are super attuned to our customers and what they need and want,” Breakstone said. Which makes sense given the whole premise of Chorus.
It also creates smart “playlists” for managers who will almost certainly never have the time to review hundreds of hours of calls but might want to hear instructive highlights or ‘red alert’ moments where a more senior person might need to step in to save or close a deal.
There are currently what seems like dozens of startups and larger businesses that are currently tackling the opportunity to provide “conversational intelligence” to sales teams, using advances in natural language processing, voice recognition, machine learning and big data to help turn every sales person into a Jerry Maguire (yes, I know he’s an agent, but still, he needs to close deals, and he’s a salesman). They include TalkIQ (which has now been acquired by Dialpad), People.AI, Gong, Voicera, VoiceOps, and I’m pulling from a long list.
“We were among the very first to start this, no one knew what conversational intelligence was before us,” Breakstone says. He describes most of what was out in the market at the time as “Nineties technology” and adds that “our tech is superior because we built it in the correct way from the ground up, with nothing sent to a third party.”
He says that this is one reason why the company has negative churn — it essentially wins customers and hasn’t lost any. And having the tech all in-house not only means the platform is smarter and more accurate, but that helps with compliance around regulations like GDPR, which also has been a boost to its business. It’s also scored well on metrics around reps hitting targets better with its tools (the company claims its products lead to 50 percent greater quota attainment and ‘ramp time’ up by 30 percent for new sales people who use it).
“Chorus.ai has helped us become a smarter sales organization as we’ve scaled. We have visibility into our sales conversations and what is working across all of our offices”, said Greg Holmes, Head of Sales for Zoom Video Communications, in a statement. “We’ve seen a drastic reduction in new hire ramp times and higher sales productivity with even more reps hitting quota. Chorus.ai is a game changer.”
Chorus has raised $55 million to date and Breakstone said he would not disclose its valuation — despite my best attempts to use some of those sales tips to winkle the information out of him. But I understand it to be “significantly higher” than in its last round, and definitely in the hundreds of millions.
As a point of reference, after its Series A two years ago, it was only valued at around $33 million post-money according to PitchBook.
“Maintaining high-quality sales conversations as you scale a sales organization is hard for many companies, but key to delivering predictable revenue growth. Chorus.ai’s Conversation Intelligence platform solves that challenge with a market-leading solution that is easy-to-use and delivers best-in-class results.” said Simon Chong, Managing Partner at Georgian Partners, in a statement. (Chong is joining the board with this round.) “Chorus.ai works with some of the best sales teams in the world and they love the product. We are very excited to partner with Chorus.ai on their next phase of growth as they help world class sales teams reach higher quota attainment and efficiency.”
Powered by WPeMatico
CV Compiler is a robot that fixes your resume to make you more competitive
Machine learning is everywhere now, including recruiting. Take CV Compiler, a new product by Andrew Stetsenko and Alexandra Dosii. This web app uses machine learning to analyze and repair your technical resume, allowing you to shine to recruiters at Google, Yahoo and Facebook.
The founders are marketing and HR experts who have a combined 15 years of experience in making recruiting smarter. Stetsenko founded Relocate.me and GlossaryTech while Dosii worked at a number of marketing firms before settling on CV Compiler.
The app essentially checks your resume and tells you what to fix and where to submit it. It’s been completely bootstrapped thus far and they’re working on new and improved machine learning algorithms while maintaining a library of common CV fixes.
“There are lots of online resume analysis tools, but these services are too generic, meaning they can be used by multiple professionals and the results are poor and very general. After the feedback is received, users are often forced to buy some extra services,” said Stetsenko. “In contrast, the CV Compiler is designed exclusively for tech professionals. The online review technology scans for keywords from the world of programming and how they are used in the resume, relative to the best practices in the industry.”
The product was born out of Stetsenko’s work at GlossaryTech, a Chrome extension that helps users understand tech terms. He used a great deal of natural language processing and keyword taxonomy in that product and, in turn, moved some of that to his CV service.
“We found that many job applications were being rejected without even an interview, because of the resumes. Apparently, 10 seconds is long enough for a recruiter to eliminate many candidates,” he said.
The service is live now and the team expects the corpus of information to grow and improve over time. Until then, why not let a machine learning robot tell you what you’re doing wrong in trying to get a job? That is, before it replaces you completely.
Powered by WPeMatico
ServiceNow to acquire FriendlyData for its natural language search technology
Enterprise cloud service management company ServiceNow announced today that it will acquire FriendlyData and integrate the startup’s natural language search technology into apps on its Now platform. Founded in 2016, FriendlyData’s natural language query (NLQ) technology enables enterprise customers to build search tools that allow users to ask technical questions even if they don’t know the right jargon.
FriendlyData’s NLQ tech figures out what they are trying to say and then answers with text responses or easy-to-understand data visualizations. ServiceNow said it will integrate FriendlyData’s tech into the Now Platform, which includes apps for IT, human resources, security operations, and customer service management. It will also be available in products for developers and ServiceNow’s partners.
In a statement, Pat Casey, senior vice president of development and operations at ServiceNow, said “ServiceNow is bringing NLQ capabilities to the Now Platform, enabling companies to ask technical questions in plain English and receive direct answers. With this technical enhancement, our goal is to allow anyone to easily make data driven decisions, increasing productivity and driving businesses forward faster.”
The acquisition of FriendlyData is the latest in ServiceNow’s initiative to reduce the friction of support requests within organizations with AI-based tools. For example, it launched a chatbot-building tools called Virtual Agent in May, which enables companies to create custom chatbots for services like Slack or Microsoft Teams to automatically handle routine inquiries such as equipment requests. It also announced the acquisition of Parlo, a chatbot startup, around the same time.
Powered by WPeMatico