The Israeli startup has developed an AI-based system for sales talks that provides managers with quick insights into sales calls and sales team performances
Israeli startup Chorus.ai seems to know how to make a sales call. The company on Tuesday announced a $16 million financing round, bringing the total amount raised by the company to about $22 million. Chorus.ai is developing an AI-based analysis system.
NLP-based analysis of sales calls
Chorus.ai records all of the sales teams’ calls, and then transcribes the recordings. That is where the interesting part comes in: finding the important parts of the call. With a single click, the managers can search for certain words or select topics, such as pricing, problems, anxiety, competitors, and closing a deal. The system also finds key words by itself, and assigns them to relevant topics with the help of natural language processing (NLP) and machine learning. With the click of a button, managers can get all the questions that sales representatives asked or were asked, see the answers, and obtain insights about the difficulties and successes. The Chorus.ai system is interfaced with services like those offered by Salesforce, for example.
Founded in 2015 by CEO Roy Raanani and VP R&D Micha Breakstone, Chorus.ai serves companies such as Marketo, Qualtrics, and Dynamic Signal, among others. The current round was led by the Redpoint Ventures fund, which has invested in startups like Zendesk, Stripe, Cyanogen, Netflix, TiVo, and Israeli company SentinelOne. Also participating in the round was the Emergence Capital fund, which has invested in Box and Salesforce, and invested in Chorus.ai’s seed round. Chorus.ai has 25 employees in its offices in Tel Aviv and San Francisco.
Speaking with Geektime, Chorus.ai President and Head of R&D Micha Breakstone says that the company will use the capital it raised to expand and broaden its R&D team in Israel, and for sales and marketing. Breakstone emphasizes that up until now, the company had a small team, which grew only recently.
Four months ago, you announced your previous financing round of $6 million, which was signed in 2015. Why did you choose to raise money again?
Breakstone: “We didn’t really decide to do the current round. What happened was that investors have been asking to meet with us ever since the last financing round. For the first year and a half, we answered politely that we needed to keep things calm in order to concentrate completely on the product and the technology. When we felt ready, we accepted meetings. We didn’t plan to raise money now, either, but when we met Tomasz Tunguz from Redpoint Ventures, we realized that we had here a partner with a profound understanding of the market, who could help us progress strategically. With the blessing of Gordon Ritter and Santi Subotovsky from Emergence Capital, we decided to hold a more serious financing round, in which they of course also participated.”
We see quite a few companies specializing in this sector, such as People.ai and Israeli company Gong.io. What is special about Chorus.ai?
Breakstone: One of the things we are the proudest of is that we developed all of our systems within the company, led by Russell Levy. The entire technology is ours, from the telephony to the automated transcription to the semantic and emotional analysis we developed. It gives us enormous power in leveraging the technology and using it to analyze the calls in real time. I haven’t seen anything like it in competing companies.”
When you boil it down, selling is not only a rational process; it also includes more than a few emotional elements. How does the Chorus.ai system handle that?
Breakstone: “There’s a lot of truth in what you say. Chorus.ai also analyzes non-semantic signals involving emotional mechanisms, for example whether a voice sounds professional, charismatic, or important, whether the people talking sound like they’re concentrating fully on the conversation, are surprised, interested, and so forth. Beyond that, since our system has correctly analyzed over half a million calls to date, machine learning can be used to draw conclusions about subliminal triggers that are not known in advance, some of which could probably be categorized as emotional.
“In the end, when there is enough data, if people can understand a signal or a trigger, it is likely that computers can also learn it. Obviously, we don’t claim to have ‘solved the equation.’ Actually we don’t believe there is one single formula, but we do believe that there are strong signs that can help and empower sales teams. From our experience, if we ask a really good sales manager what the difference is between his or her best salespeople and average ones, the answer will usually be unclear. He or she might explain that good salespeople ask more questions in a better way, but if we insist on trying to clarify what ‘asking questions in a better way’ means, he or she won’t know what to answer.
“This is where Chorus.ai enters the picture. It is able to figure out exactly which questions are worthwhile to ask, when they should be asked, and which words should be used in them. Our system is able to distinguish clearly between better salespeople and average salespeople. It enables salespeople who aren’t as good to learn from the best and improve, whether it involves semantic signals or emotional triggers.”