AI-Driven Customer Service Is Gaining Steam – Forbes

(Original Caption) 1895 – Telephone traffic operators seated at a line of early manual switchboards.
For companies that are looking for a practical AI application – AI customer service is here – and it’s only getting better.
Klarna – one of the highest profile fintech companies on the planet, recently announced that their AI customer service tools are able to do the work of roughly 700 employees. Many other companies are following suit, with Microsoft’s recent announcement that it plans to roll out AI in its call centers. So how did we get here and what can we expect as AI customer gains more and more traction?
Chatbots, which gained popularity in the mid-2010s, were primarily rule-based systems designed to handle straightforward questions. They relied on predefined scripts and decision trees, often leading to pretty frustrating customer experiences. I personally worked on a customer chatbot company, and the technology has come a very long way in a very short amount of time.
Today’s AI customer service relies more on inherent NLU – Natural Language Understanding – that makes interactions seem more natural and less scripted. You actually feel like you are talking to a person rather than a ‘robot’.
The major leap in AI customer service comes from its ability to learn and adapt continuously. Unlike traditional chatbots, modern AI systems analyze vast amounts of data to understand context, sentiment, and user intent more accurately. This capability not only makes for a better experience, but it helps AI customer service potentially anticipate customer needs rather than wait for a customer to be angry at a chatbot.
There are many high-profile entrepreneurs working on AI customer service and related areas. Bret Taylor, the Chairman of Open AI, is currently working on a new company called Sierra. Other companies in the space include: Sierra, Ada, Kustomer and Gladly.
Decagon – that came out of Stealth today with a $35M round from Accel and A16Z – already has significant traction and an impressive list of backers. Decagon’s cap table includes Box CEO Aaron Levie, Lattice CEO Jack Altman, Rippling COO Matt MacInnis and a slew of other Silicon Valley heavy-hitters. Their customer roster includes Bilt, Substack and Eventbrite amongst others.
I recently asked Decagon’s CEO, Jesse Zhang, about the infamous ‘data issue’. Many adopters of AI are concerned that their data may be used incorrectly, or trained for the benefit of other companies.
Zhang’s answer for Decagon: “I can’t speak for all companies but customer data is the top priority for us. By default, customer data is never used for training and we have agreements in place with our 3rd-party providers to ensure that they do not train on the data. If the customer opts in, we are able to fine-tune models based on their data, and those models will only ever be used for them. Finally, we also have pipelines to process data beforehand and strip out sensitive data before they reach the LLM, which creates an airgap.”
As the base LLMs get better, AI customer service should also get better and solutions like Decagon (alongside human reinforcement learning) should give companies more leverage over time.
With all the options out there, it can be confusing for companies to select the right solutions. There are vertical specific solutions like Gorgias (which is designed specifically for ecommerce), there are solutions designed to help AI contact centers (including by Google), and numerous other options for different channels. It is also hard to discern and vet whether the technology is real, which is an important consideration when making a purchasing decision.
I asked Zhang his perspective on how companies should evaluate potential solutions in the wake of all the choices that are currently available on market:
“The main innovation of true AI-based customer support solutions is that LLMs are used from the ground up to generate answers, take actions, etc instead of the old way of using decision trees and basic AI models. Buyers should certainly be clear of sort of AI solution they are getting into. Is it an old system that is pivoting into generative AI or is it built from the ground up with LLMS? LLMs yield a ton of benefits but also come with things that buyers should definitely think about:

One thing is clear – with the amount of technologists in the space, the venture money that has poured in, and the clear efficiencies AI customer service is driving for companies – it is very much here to stay. If you are looking for a first AI use case to tackle, AI customer service is a potential great candidate.
*Disclosure – Decagon’s CEO is a small shareholder through a fund in one of the author’s business ventures.

One Community. Many Voices. Create a free account to share your thoughts. 
Our community is about connecting people through open and thoughtful conversations. We want our readers to share their views and exchange ideas and facts in a safe space.
In order to do so, please follow the posting rules in our site’s Terms of Service.  We’ve summarized some of those key rules below. Simply put, keep it civil.
Your post will be rejected if we notice that it seems to contain:
User accounts will be blocked if we notice or believe that users are engaged in:
So, how can you be a power user?
Thanks for reading our community guidelines. Please read the full list of posting rules found in our site’s Terms of Service.