The ubiquity of conversational user interfaces including text-based messaging and voice-enabled devices is well documented at this point. The ability to use these conversational channels for digital self-service is solidly moving from a delight to an expectation, across industries. As this once nice-to-have interface becomes a need-to-have, Conversational AI is not a question of IF for your financial institution (FI), but WHEN. As big boys like Bank of America, JP Morgan, Capital One and Ally Bank lean hard into their Virtual Financial Assistant (VFA) offerings in 2020 and make it rain new features, more and more competing banks are realizing that WHEN = NOW.
As your FI moves Conversational AI from a roadmap item to an on-deck project, it’s time to hit the streets and find a provider to help – unless of course you have a few years and a hundred million bucks to spare to build it yourself (BofA’s rumored resources to build Erica!).
While there are many things that should be considered in any new technology project – especially customer facing ones – we want to highlight five that we feel are vital learned from our years of experience working with FIs ranging in size from community banks and credit unions, all the way up to top ten U.S. banks.
Machine Learning vs. Decision Trees
Big tech has made it accessible (some would even say easy) to grab a few developers, wire up a few one-line commands on a smart speaker, and trigger corresponding APIs. FIs/providers utilizing these one-line commands simply change pressing a button in the app to pressing a button with your voice. While the novelty of using a smart speaker with these commands seems like a win for now, FIs with VFAs relying on a decision tree of commands risk quickly hitting a plateau and losing momentum in adoption.
The key to advanced use cases (and a pleasant experience with the foundational stuff) is going beyond one-line commands into natural conversations. Multi-turn flexible conversations, ones where the VFA detects entities and variables on the fly without the need for a formulaic sequence of questions, is what makes a VFA, well, human-like. Ask to see this when getting demos. See how a VFA handles back-and-forth (multi-turn) conversations or the customer changing their mind (like real customers will), “Transfer 40 dollars to my backup checking”…”actually make it 75 bucks.” A flexible VFA that can handle requests in a human-like way across many channels (in-app, SMS, Alexa, Google, Facebook) is what will drive adoption and usage, not one-liners on just a shiny Alexa.
Finance Domain-Specific Experience
Every few months it seems as if there is a new Conversational AI startup in the market touting their ability to easily deploy a VFA and save you tons of money. One problem though – you have to agree to be one of their first two or three customers. But wait, how can that be? The big flashy demo at the conference? The dizzying talk of neural networks and machine learning and unicorns? Noise is noise and there is plenty of it in this space, especially from newcomers. In addition to new Conversational AI startups, you have several live-agent chat/support providers jumping into AI (It doesn’t make sense to us either) because they figure that since their plumbing for a chat window already exists, it should be easy to build Conversational AI into it – it’s not. It’s vital that FIs find an experienced provider (like Abe.ai) who understands the unique challenges of a finance domain specific AI. A successful deployment takes a deep understanding of machine learning, model management in a regulatory environment, integration into digital banking API layers, and many other technical considerations. With that said, cool tech is not enough. Understanding how FIs deploy and manage change comes with experience and is crucial to deployments.
Deployment of Advanced Use Cases
While future-proofing technology is next to impossible, there are still a few things that can help keep you ready. Roadmap, roadmap, roadmap, and not just features on a slide, but in development (or better yet already built!). This is difficult technology to get right, so consider the “HAVE-built-it” folks vs. the “WILL-build-it” ones. A provider may make it sound easy to build the use case you are focused on, but don’t be so sure. Those new use cases take time and you want to partner with a provider who has already invested time in developing advanced use cases (proactive engagement, product origination, advanced payments, human hand-off, etc). At Abe.ai, we propose a deliberate rollout plan with each of our FI customers, starting with foundational capabilities so you build customer adoption and trust. The advanced use cases are built and waiting to be utilized when the FI’s roadmap, and customer adoption/appetite, warrant them.
Also worth noting is the provider’s availability of tools for conversational design and data science. Having the freedom to customize how your VFA responds and the ability to build unique use cases for your FI may be important to you as you go further on this journey. Partnering with a provider who not only has the tools, but also the resources available to help you means not needing to hire an army of data scientists and conversational designers.
Ability & Willingness to Collaborate with Your Partners
In addition to having it built already and having the tools to let you DIY, your likelihood of success can hinge on your conversational AI provider’s ability to incorporate and work with your existing and future partners. Even if the provider is not actively collaborating with any of your current partners, ensure they have managed relationships with other technology and banking partners in the industry, and done so successfully and responsibly. The success of partnerships is critical as each partner becomes more and more specialized. As VFAs become ubiquitous across FIs, those able to incorporate customer insights from a variety of sources and systems (partners!) to trigger conversations, will deliver the most value through conversational AI.
Staying Power for Long-Term Conversational Banking Success
In our years in the industry, we have seen competitors come and go and ancillary service providers trying to dabble in and out of Conversational AI with little success. The truth is there are very few Conversational AI providers who can meet the standards detailed above. Even less when you add all the other considerations of a successful new technology project. We have been on the receiving end of multiple calls from FIs who attempted to build VFAs themselves or with a provider that did not meet the standards above, and failed. We have even fielded a panicked call from an FI who had a hard outage – their VFA stopped responding, as did the provider, because they no longer existed.
Abe.ai is a leader in the Conversational AI space with nearly four years experience meeting and exceeding the standards detailed above of advanced AI, experience, foresight and partnership. We are backed by the strength and resources of a parent company (as of January 2019) with more than 20 years of experience in financial technology (Envestnet Yodlee). All reasons why we are determined to continue to be the preeminent provider of Conversational AI in the financial services space and would appreciate the opportunity to discuss with you further.