In recent years, there has been a surge of investment in artificial intelligence technology for banks. For instance, in 2020, Cornerstone found that 38% of banks and 51% of credit unions had discussed implementing AI technology at the board or executive level.
The desire to enhance customer service is frequently a key driver behind these AI initiatives. In 2021, Capgemini recorded that 94% of banking and insurance firms saw improving customer service as a key objective in their AI initiatives. Rapid progress continues to be made, and more than half of these firms report that at least 40% of their customer interactions utilise AI tech.
Virtual agents, powered by conversational AI technology, are a crucial component of these customer service automation initiatives. In fact, Gartner predicts that 1 in 10 of all agent interactions, across all industries, will be automated by conversational AI in 2026.
Banks are on the frontline of a wider digital transformation that is seeing the expansion of AI technology into almost all aspects of daily life. From omnipresent algorithms to smart cities, healthcare diagnostics, and virtual assistants, the future is personalised, predictive, and engaging.
Many prominent banks, including HSBC, Bank of America, and Wells Fargo, have launched or announced conversational AI offerings to customers. Virtual assistants are quickly becoming a new face of banking.
Key decisions
If your financial organisation is planning a conversational AI initiative, you’ve likely already faced an elementary dilemma. Do you build your own conversational technology in-house from the ground up? Or do you purchase custom and domain-specific technology from a specialist provider?
At this point, a bank will usually take one of two paths forward. The first is to put together an in-house team to create a custom conversational AI platform that will power their virtual assistants. The second is to work with a dedicated, specialist conversational AI supplier to build a bespoke solution for their bank.
Building conversational AI in-house
It’s easy to see why an in-house solution would be appealing to many banking organisations. Building a solution from the ground up gives you maximum control over the scope, features, and functionality of your virtual agents. Trust and stability are often essential in banking, and by building the technology yourself you have complete control over the branding, tone, and security of the product. You can also make any necessary changes yourself without the need to consult with a third party.
However, in reality, it is incredibly difficult to realise these benefits if a sophisticated solution is required. The journey toward an in-house solution is long and arduous. You’ll need to build an expert team who have expertise in the fields of computational linguistics, machine learning, natural language processing, and automatic speech recognition if you plan on supporting inbound calls. These are some of the most in-demand skills in the AI industry. If even the biggest tech companies are struggling to attract this kind of talent, does your bank have the necessary budget and resources to do it? Further, there is the need for considerable know-how that only comes from building effective Conversational AI services; success requires experience allied with a deep appreciation of the theoretical field. The number of people who have been engaged in building sophisticated automated services is vanishingly small.
The lead time on such a project is likely to be significant. To build a truly competitive conversational AI solution your experts need to be at the cutting edge of their field. Before you can even get started, you need to have a comprehensive understanding of the fields mentioned above. By action.ai’s estimates, a bank would need to undertake at least two years of research before launching an in-house solution; and that’s even if you choose to use NLP transformer models to jumpstart matters.
Conversational AI technology can require a voluminous amount of training data, and this can take years to gather, collect, and analyse. Even working out what kind of training data you need can be a challenge in itself. Data collection is costly and the analysis is time-consuming. All of this needs to be factored into any proposal for an in-house project.
What about no-code platforms?
Banks who want an in-house solution, but who don’t want to commit to such a lengthy and costly project, might be tempted to turn to a low- or no-code platform. Services such as through IBM Watson, Google Dialogflow, or Azure’s Bot Service enable their clients to create a simple conversational AI solution relatively cheaply. These bots can be deployed over a range of instant messaging and voice channels.
On the other hand, there can be substantial drawbacks to relying on no-code tools. Firstly, the resulting bots are likely to lack sophistication and have limited support for voice interactions. No-code tools are simply too inflexible to meet the needs of an agile financial organisation. Secondly, these tools are often not as easy to use as initially promised. Poorly written documentation and limited customer support can leave clients confused and unable to achieve their objectives.
Finally, low-code solutions lock your bank into the ecosystem of one particular tech company. You cannot pick and choose the best-in-class components, and you’re unlikely to be able to build a truly competitive conversational AI product with these platforms.
The benefits of buying conversational AI tech
Suppliers of specialist banking-focused conversational AI tech have already done much of the hard work. They’ve assembled a team of experts who are at the peak of their fields, and they’ve done the extensive research necessary to create great conversational technology. They will likely have already tailored their platform to the specific needs of the banking industry, and they know how best to effectively commercialise the results. The very best suppliers will already accommodate your security standards, and they will have a plan in place to scale the technology as needed.
A specialist will also already have a large amount of training data in place, and will have the networks necessary to commission new data to the client’s specification when required. Unlike most low- and no-code platforms, conversational AI specialist solutions allow for on-premises hosting as well as cloud deployments. An on-premises solution puts your bank’s customer data securely in your hands.
In short, buying this technology, rather than building it, is the best way to ensure that your bank can deploy a sophisticated, reliable, and cost-effective conversational AI service.
Why do banks need conversational AI?
While the best solution will vary from bank to bank, the successful deployment of conversational AI and customer experience automation is vital to the future of banking.
Virtual assistants have the ability to power a more cohesive digital ecosystem where the customer is presented with an engaging and unified experience at every touch point. From making bank transfers and checking balances, to opening new accounts and reporting fraud risk, conversational AI helps customers to achieve their goals in a consistent and conversational manner.
By engaging customers and exceeding their expectations, conversational technology has an enormous potential to improve customer loyalty and trust. And by automating customer service interactions over the phone, virtual agents can provide extraordinary cost savings.
When your organisation purchases conversational AI technology, you’re buying into improved customer service, reduced contact centre costs, and the future of the banking industry itself.
