Guest author: Frank Chang, Co-Founder & Managing Partner, Flying Fish
Technology expert, Frank Chang, discusses the importance of specialization in the application of speech and chat.
As is the case with any new technology, there’s an adoption curve with artificial intelligence (AI)—particularly the applications that leverage natural language processing (NLP).
Speech is the most natural communication method—people have been interacting in this way for thousands of years. Only recently have we been getting comfortable with using speech and chat to communicate with our devices. Like the keyboard, mouse, and touchscreen that came before them, speech and chat are just other modalities—or interfaces—that are coming into their own.
With all the investment in AI over the past decade, one might wonder why it’s taking so long to get us to a point where we can converse effectively with AI-powered devices. The reason is that it’s extremely hard to get right—especially for broad applications like personal assistants.
That said, there’s a huge opportunity for technology companies to focus on a niche area and go really deep with NLP.
Being all things to all people
Siri, Cortana, Alexa, and Google are very broad—they can tell you a little bit about a lot of things. This works reasonably well in the consumer space and the more people talk to these systems, the better the models will become and the deeper the interactions will be. Questions that can’t be answered today are used as inputs to train the systems to “learn” how to answer the question next time. These chatbots are helping to drive adoption and, critically, they’ve made it socially acceptable to talk to your phone (even when you’re not on your phone) which drives more usage.
Eventually, these AI assistants will become highly personalized to the individual. They will understand the nuances of how you speak and predict how you might respond to certain scenarios. We have a long way to go to get to this point, and the barriers to entry are high, but that doesn’t mean new players can’t make an impact if they narrow their focus.
Specialization is key in NLP
There are so many opportunities for technology startups to move the needle on NLP. The key is to reduce scope and go deep on one very specific topic. Any roles that involve sifting through mounds of legacy data are great targets and with the right models, can be automated and communicated with in a natural way.
The Financial Services segment includes many roles like this. That’s why we’re seeing a wave of successful fintech companies popping up—the industry is ripe for disruption.
Case-in-point: Finn AI
You may not be able to ask Finn AI’s banking chatbot to tell you how many calories are in a pepperoni pizza but you can ask it absolutely anything about the state of your finances. By honing in on several very specific use cases, Finn AI and its conversational AI banking chatbot is becoming the “Alexa of personal finance.”
Finn AI is also capitalizing on a zeitgeist—young people don’t want to interact with people. This sounds strange but it’s true. More and more, young people are demonstrating their preference to interact with their phones and do everything digitally.
Finn AI is helping traditional banks and new digital banks deliver amazing service without the overhead of people and physical branches. By starting simple and focusing on customer service enquiries, their banking chatbot can now answer 95 percent of the questions most commonly asked of banking call center agents. As trust is built, users engage more deeply with the virtual assistant, performing transactional banking tasks and asking the chatbot for guidance on financial products and planning.
Eventually the AI-Powered chatbot will become a personal credit coach to help people manage their financial wellbeing. Think about it. That’s the level of trust you have with an individual banker or advisor who knows your family situation, your spending habits, your debt profile, and even your retirement plans.
There are many companies trying to figure out how to apply AI and NLP to different scenarios. The answer is simple: identify an area ripe for disruption, scope the problem down as much as possible, gather a ton of data, build trust, and watch your business grow from there.
About the author
Frank Chang is Co-Founder and Managing Partner at Flying Fish Partners – a venture capital firm run by technology operators and focused on software startups in the Pacific Northwest. He holds over 20 years of experience serving in executive roles at some of the largest, enterprise companies. At Amazon, he was VP of Technical Program Management for Amazon’s Audible subsidiary. During his time at Microsoft, he led the core Speech Recognition and Natural Language teams. Frank holds a BA in Computer Science from Princeton University.