Machine learning is on the verge of making a major impact upon business communications, in terms of customer facing and by enhancing the role of the employee within an organization. This includes the means by which information is delivered and with providing improved insights and predictive analytics for decision making.
An example of a platform to deliver this is Chorus.ai. This solution uses an in house AI-based framework that leverages deep learning to automatically generate a “voice fingerprint”, allowing listeners to separate speakers. This is particularly of use with sales calls. Additionally, the system is specially adapted to recognize a company’s competition and niche industry terms that general AI solutions don’t understand.
To understand more about how AI is shaping this area of business, Digital Journal spoke with Micha Breakstone, Co-Founder and Chief Scientist at Chorus.ai.
Digital Journal: How sophisticated is AI becoming?
Micha Breakstone: That’s a tricky question. On the one hand, we see AI successfully accomplishing increasingly challenging tasks at an increased accuracy, speed, and effectiveness, with examples ranging from automatic speech transcription, image recognition, logical inference, and process optimization. On the other hand, AI has a long road ahead of it, in the sense that it still requires huge amounts of data to learn from. While we’re seeing the learning itself become much more powerful and accurate, it is still lacking the ability to extrapolate and deduce from relatively few examples, as humans do, and so much of AI’s progress still depends on enormous amounts of manually-tagged data.
DJ: What are the key developments with AI for linguistics?
Breakstone: The two major key AI developments in recent years as applied to language include the following:
Speech - One of the first big wins for Neural Nets and Deep Learning was for Speech, specifically Automatic Speech Recognition (ASR), or the task of automatically transcribing voice recordings into written words. Advances in AI increased the accuracy of transcription engines by over 50% since 2012, with current precision hovering around 90+% depending on the language and task. Researchers estimate that manual transcription by humans is slightly higher than 95% so AI researchers still have a long way to go to achieve human parity.
Word embeddings – Roughly speaking, word embeddings are a mapping of words to relatively low-dimension vectors in a way that preserves interesting characteristics of the words, such as co-occurrence or meaning. One can think of them as a mathematical representation of words that retains syntactic or semantic relations. Embeddings are a key component in improving various Natural Language Processing (NLP) tasks such as Question Answering, automatic text summarization, and logical inference from text.
DJ: What types of practical applications are there with this?
Breakstone:The applications run wide and deep, including the various technologies behind Virtual Personal Assistants (e.g. Siri, Alexa, Google Now), automatic transcription, and information extraction of insights from conversations (e.g. Chorus.ai).
DJ: What does Chorus.ai do?
Breakstone:Chorus.ai is the market-leading platform for transforming conversations into data and insights, ensuring more effective communication between companies and their customers. Chorus.ai’s technology transcribes and analyzes conversations in real time, sending users results in minutes. The proprietary algorithms detect High-Value Moments that serve as the foundation of an effective strategy for sales teams, thus freeing sales representatives to focus on building relationships. Every cloud in a company, whether it be sales, marketing, or business development, becomes more valuable when powered by Chorus.ai’s technology.
Unlike many other companies in the Conversation Intelligence space, Chorus.ai’s proprietary algorithms allow near-immediate transcription and analysis. Chorus.ai’s in-house development allows for vertical-specific customization, increasing accuracy and boosting results. It also ensures privacy, as the user’s data is in one known location and not sent to a third party for analysis. Due to these factors, Chorus.ai is able to produce more accurate, quick results and include more advanced features than competitors.
DJ: How did you go about developing Chorus.ai?
Breakstone:Our three founders’ combined experience made the process of starting Chorus.ai seamless. Our CEO was a Manager with Bain for 8 years, focusing on the impact of harvesting huge troves of untapped organizational data for insights to unlock business potential. OurCTO has been doing VOIP and conferencing for over a decade. With this background, Chorus.ai was a natural choice. Prior to becoming Chorus.ai’s Chief Scientist, I helped build and sell two successful startups in the space of Natural Language Processing and Speech, the more recent one a Virtual Personal Assistant Platform business unit he founded and sold to Intel.
DJ: What were the main technological challenges?
Breakstone:Countless conversations between employees and their customers, internal meetings, and business-centered discussions are conducted with no proper record or analysis, even in an era where Artificial Intelligence and Deep Learning are on the rise. While large companies have attempted to create Automatic Speech Recognition (ASR) engines that address the need for accurate records of conversations, such processors are still thrown off by background noises, speakers’ accents, and non-standard vocabularies like names of companies and products.
Also, companies still struggle with taking in information gathered on a sales call and effectively putting it into their CRMs. Finally, and perhaps most importantly, doing all this in a robust, streamlined way, and then extracting insights to guide better conversations is a challenge that has eluded even the most advanced AI companies to date.
DJ: Which companies have expressed an interest in the technology?
Breakstone:Chorus.ai has a large array of customers who have already utilized our AI technology and seen substantial results. Software company Procore shorted its ramp time after implementing Chorus’s technology and continues to use it to support its rapid scaling. Marketing company Engagio implemented the technology to align sales development reps, account executives, marketing, and customer success. Marketing consultant EverString increased its close rate after putting Chorus.ai to use, as well as better organized its revenue teams. Multiple other companies, whether it be customer engagement platforms, SaaS-based enterprises, and content marketing hubs, have implemented Chorus.ai’s technology and seen increases in productivity and efficiency with their sales teams.
DJ: What do they intend to do with the technology?
Breakstone:Chorus.ai has paved the way for better customer insight for all of these companies. Every employee should understand his or her customers and Chorus.ai technology allows you to have the consumer’s voice right in your inbox. Chorus seamlessly and automatically populates a company’s CRM system with call details, ensuring that sales reps will have access to the updated information they need. Companies are using this real-time information to make sales teams more effective, productive, and efficient.
Also, Chorus’s proprietary algorithms are trained on the collective learnings of the top B2B revenue teams, as well as a sales team’s own best practices. This creates a Coaching Network, or a democratization of successful methods, that companies are taking advantage of, and results in quicker ramping of new hires and increased sales. Overall, Chorus.ai capitalizes on crucial data across a buyer’s entire journey to help sales representatives reach their full potential.