In our new blog series, “Meet our Data Champions,” we showcase the exciting work Google Cloud customers are doing with data and AI/ML. In this edition, we talk to Bob Summers, CEO and founder of Goodcall, a company whose phone AI service leverages Google Cloud Speech AI technologies to bring the power of AI to Main Street businesses.
Q: Tell us a bit about yourself. Where did you grow up? What did your journey to tech look like?
Bob Summers: Both of my parents served in the U.S. Air Force in the computer career field. My mom was a pioneer in the Air Force as one of only a few women working in computers, and she went to school at night over a 10-year period to get her computer science degree. Her drive to overcome adversity in a tech field has always served as an inspiration.
We lived in Hawaii, Maine, and Virginia. Most of my life was spent in Virginia, where I attended Thomas Jefferson High School for Science and Technology and Virginia Tech. With two parents working with computers, I started coding early, at age 8. I was drawn to how quickly software could be built to help others, and there were many mentors along the way who helped to direct my energies.
Q: What inspired you to start Goodcall?
Bob Summers: Most of my career has been focused on making technology more accessible to those who need it. This has included bringing high-end compression technology to desktop computers for live video calling with iSpQ Videochat (predating services like Skype or FaceTime), and using machine vision to enable personal trainers to 10X their ability to coach their clients with Fit.net.
While working at Google on Ads to help small businesses grow, I realized that many local businesses struggled to answer phone calls, one of their most important channels for growth. This reflected my own experience calling to get a dinner reservation or hire a plumber. It can be a real challenge to reach and transact with a local business owner who cannot afford to have a full-time person answering calls—they cannot answer calls while simultaneously doing what they do best.
We were inspired by advances like Google Duplex, which was an early demonstration of how conversational AI could be used for practical purposes, like making table reservations and salon appointments on behalf of users. Other first-to-market Google Cloud technologies for natural language processing (NLP), such as BERT, were really inspiring in terms of how large language models (LLM) could impact the world. It was easy to see that as computation increased, there would be newer models with more and more input parameters—some have hundreds of billions now. These are all foundational works from Google that are the basis of current conversational AI trends.
But, how can we make complex LLM technologies available to every local business and helpful for their customers? Why not apply this technology toward main street businesses? Goodcall was born.
Q: Why was having a data and AI strategy important in developing your company?
Bob Summers: There are two dimensions of data that are critically important to our business: conversational quality and business intelligence (BI). Quality conversational AI depends on having thousands or more data points to train the AI models so that the machine can understand a request and properly respond in a trusted and reliable manner.
Measuring every aspect of a conversation—from time of day, duration, actions, sentiment, and more—enables us to communicate and demonstrate return on investment to the small business.
We use both types of data to inform product prioritization and business objectives. Our business depends on this data and AI strategy to serve our customers and grow.
Q: Speaking of having a data and AI strategy, which Google Cloud products are a critical part of that strategy and how are those products being leveraged? And what benefits has it brought to the business?
Bob Summers: When building Goodcall, we evaluated a number of cloud providers for compute, storage, and AI capabilities. We were drawn to Google Cloud initially because of its strength in AI. More specifically, the Speech API, which includes technologies such as speech-to-text, natural language understanding (NLU), and text-to-speech (TTS). Our team was familiar with these technologies, and when we compared them head-to-head against other providers, it was clear that Google Cloud’s products provided low latency, reliability, and quality—all requirements for our real-time conversational agents.
Beyond the Speech AI capabilities, we needed infrastructure for scale. To deliver our AI phone service to over 14,000 small businesses in the US, we deployed several dozen APIs and services from Google Cloud. The primary Google Cloud services we use are App Engine, Spanner, BigQuery, CDN, and networking services like load balancers. We also use Looker for visualization of key metrics.
Q: What’s the coolest thing you and/or your team has accomplished by leveraging AI?
Bob Summers: Conversational AI is a tool that all businesses should have access to because it can dramatically increase productivity. Our product challenge is to make wonderful AI and cloud services accessible to anyone: affordable, easy, and helpful.
The breadth of entrepreneurs that we help is by far the coolest thing we have accomplished. Goodcall is actively helping local businesses in all fifty U.S. states and in all business verticals. If you are a plumber, pizza shop owner, web designer, dry cleaner, baker, yoga instructor, etc., there is a Goodcall solution waiting to help grow revenue, customer service levels, and efficiency. In all cases, the entrepreneur does not need to have any technical expertise to teach the AI about their business.
Our team is succeeding in this challenge to make this powerful technology accessible for local business. One of our customers started when their 10-year-old set up Goodcall for the family business using only their phone! Now that is cool!
Q: What was the best advice you received as you were starting your AI journey?
Bob Summers: Some of the best advice we received on our journey was to always keep our AI models fresh with the latest customer data. There is essentially a shelf life for many data points that feed a model.
Q: What’s an important lesson you learned along the way to becoming more data/AI-driven? Were there challenges you had to overcome?
Bob Summers: Human labeling is an art and a science. AI algorithms depend on labeled training data, and there are many ways to get this labeled data. If there are low quantities, then labeling yourself as a team is a great way to start. As the data pipeline grows, using a classification algorithm or human powered service to label the data is an absolute requirement. The art here is how to train your labelers by asking the right questions about the data. For example, keeping questions simple such as yes/no, versus providing five or more options, can make a big difference in label quality. The quality of the question can dramatically impact the quality and efficiency of labeled data. Better label quality is directly related to higher AI model performance.
Q: Which leaders and/or companies have inspired you along your journey?
Bob Summers: We are inspired by the consumer AI teams behind products like the AI assistants that have brought conversational AI to the masses. They have proven that this technology can be helpful to everyone. The deployment of voice assistants at scale is changing the societal perception of how machines can help us. Goodcall is inspired and motivated by the breadth of capabilities and number of people globally who use these assistants daily to make their lives better.
Q: Thinking ahead 5-10 years, what possibilities with AI are you most excited about?
Bob Summers: AI will drive new levels of productivity across all aspects of the economy as it takes over repetitive and tedious tasks, similar to the early Industrial Age, when large machines were used to amplify human capability. Sure, ten people can work all day to dig a hole, or you can use a backhoe to get it done in a fraction of the time. AI will do similar things to knowledge-type tasks. Autonomous agents that help organizations of all sizes will be an everyday part of doing business. You can absolutely expect every product that you use will be impacted by AI in the next 5-10 years.
Q: What’s something people don’t know about you?
Bob Summers: When I was 14, I worked on Bulletin Board Systems (an early version of the Internet). I was also very eager to learn to drive a car, and an opportunity came to trade code for a car. So at age 14, I traded software for a car: a 1962 Pontiac Tempest Lemans. I spent two years restoring it, and the first day I was able to drive it, someone rear ended me.It was a Honda that hit me—my car had just a small scratch, but the Honda did not fare so well.
To learn more about the Speech AI technologies Google Cloud customers like Goodcall are using, check out this collection of Google Cloud’s top Speech AI stories from 2022 and our new generative AI offerings.
By: Google Cloud Content & Editorial
Originally published at: Google Cloud Blog
Source: Cyberpogo