Deep Dive: Chatbots and conversational AI struggle to keep up during pandemic
- AI and chatbots have been vaunted as customer service solutions for years.
- But the pandemic showed just how far these tools are from really making a difference for customers.

Call volumes into banks have gone up 10x during the pandemic. Wait times have been hours at the peak of the crisis. Chatbots and other forms of conversational AI have been deployed to help. But they're not quite there yet.
Tearsheet's Sara Toth Stub recently wrote a story about where chatbots are useful and where they aren't. After a lot of hype, they are providing some real value but it will be years before the technology approaches human customer service reps.
Sara Toth Stub is my guest today for a deep dive podcast to discuss what's happening inside the customer service queue in leading banks and financial institutions.
Subscribe: Apple Podcasts I SoundCloud I Spotify I Google Podcasts
The following excerpts were edited for clarity.
Why the interest in chatbots now?
I was actually speaking to the Kabbage co-founder Kathryn Petralia about a month ago about the PPP loan program. She mentioned then that they were overwhelmed with the call volume. The company needed to rely on automation to manage all that. I began to see it as a trend.
If you go to a financial institution or fintech, they almost all have notes on their homepages saying they are experiencing longer than usual wait-times on the phone. Things are so uncertain -- people have extra questions about their finances and that's compounded by the PPP and other forms of aid offered through financial institutions. It's a confusing process, especially for small business owners.
SPONSORED
What banks are using to handle inbound volume
Most banks rely on some type of routing system that clarifies a consumer's needs and questions so that they're answered by a person or a chatbot. Most banks have this and those that didn't were scurrying to update their system once it became clear that this pandemic wasn't going away. I spoke with a company called Kasisto, which designs customer service solutions driven by conversations. They were inundated by requests from banks that didn't have sufficient resources for customer service.
In addition, a lot of places needed to shut down their call centers or had them working at limited capacity. Because of security, you can't necessarily have people sitting at home, answering calls for a financial institution. This also exacerbated the situation. As the CEO of Kasisto told me, this is no longer a fun thing to have -- the pandemic has made this seem very necessary for banks and financial institutions.
Use cases
A very valuable use for these tools is routing -- where people express what they need or get an answer to an easy question. That's a big use for a lot of institutions. In addition to these systems helping customers, the systems also analyze all the questions they're getting and you can look for trends, see what people are asking about.
The CEO of Kasisto said this was interesting and by looking at this data, banks could serve customers better because they can see what new questions are being asked. At some point, a lot of people started calling in to see how often ATM machines were cleaned and sanitized. That wasn't a question banks were asked before. This way, banks could quickly see that they needed to have an answer to this. Without conversational systems like this, gathering this information would have taken a lot longer.
Is AI prime time?
Some people would say that AI is overhyped, but in fairness, it's something that's developing all the time. I spoke with Daniel Faggella, who's the CEO of Emerj Artificial Intelligence Research. He was saying that we're not at a point where you can have a real, in-depth conversation with AI, other than asking what your balance is. He did say, though, that the number of staff required to answer calls is dropping. AI has cut out some of the wait time.
It's also interesting to point out that a lot of these assistants are on apps. Banks are allowing people to do many of the things they could already do sitting on their laptops, but to do it hands-free and get answers more quickly. It's not necessarily that they're getting new information, but they're getting it in a new way.
How far away is AI from really making a difference
One thing that needs to happen is more interaction and training from humans -- like putting together data scientists with customer service professionals to work with the software and develop more practical solutions.
When I was speaking to Daniel Faggella at Emerj, he said that in five years there will be even less wait time, but AI still won't be there. He also said, that at the end of the day, high value clients will almost always get routed to a person because nothing can completely replace the human touch.