Keen telly addicts among you may have noticed the re-emergence of a nineties phenomenon recently: Robot Wars. For the uninitiated, the programme offers keen roboteers the chance to build intricate and complex robots out of bits of old tumble dryer, abandoned road signs and power tools. These robots are then pitted against each other in a fight to the robo-death.
Another nineties phenomenon which has become increasingly present in the press is artificial intelligence which is seeing a resurgence in interest thanks to a wave of innovation. One key strand of this innovation is the development of chatbots (software designed to simulate conversations with human users). Their main purpose to date has been to automate some key front line contact functions away from human operators and towards machines.
But just as in Robot Wars, not all chatbots are equal. In the TV show, contestants pull together a range of components from hardware stores. A circular saw here, a cordless drill there, a flame thrower made out of a hijacked can of Lynx sticking out the back. Their aim is to build the best robot they can. For financial services firms developing chatbots, a similar array of choice is emerging from developers around the world and the challenge is now one of welding this technology together.
A chatbot is essentially made up of 3 components
- The platform
While many firms have deployed bots direct to their own websites, increasingly brands are deploying bots onto users’ trusted platforms. These are typically either distinct messenger applications (whatsapp, facebook messenger, Slack, KIK etc.) or a range of core native integrations (SIRI, Google Now or Microsoft’s Cortana). Picking which platform a bot will be available on has a huge influence on its success and there are stark demographic differences between the users of some of the main options.
- The Input mechanism
While the earliest chatbots have largely been built to accept text input there have been great advancements made recently in speech recognition, the use of avatars, face and gesture recognition and virtual presences. Alongside this enhanced ability for bots to interpret varied input types, the increasing number of intelligent connected home devices like Amazon Echo will lead to a proliferation of people accessing intelligent assistants around their home without the need for a phone or a tablet.
- The language processing
Of course if it’s your chatbot, the special sauce is that it talks to users in your language, and offers them options which are unique to your proposition. Early developers working with chatbots had their work cut out for them, coding up core language functions and trying to second guess how humans interact with their devices, how they really use language, and what their actual intentions are. As with many innovations, their work has lowered the barrier to entry for developing applications which make use of natural language processing and conversation management. Components like wit.ai allow users to build out conversation patterns and context associations in a developer and business user friendly UI while init.ai offers a complete bot building framework including easy integration to external APIs for data gathering.
The key to success for financial services firms lies in their ability to successfully integrate these components. The best solutions will utilise existing libraries and frameworks to minimise heavy lifting and focus on delivering value which is specific to their user base. Supporting this, the consumerisation of AI development we are witnessing with wit.ai and init.ai is being complemented by the proliferation of cloud services which offer a range of AI functions (see Watson in the cloud for some examples of this).
Chatbots and other AI technologies will soon sit alongside workflow engines, CRMs, BI tools and document management systems as components in the increasingly complex digital ecosystems which financial services firms need to manage. Just like their peer systems, chatbots will present a range of challenges to firms, through both their own functionality, and the way that they interact with other systems in the landscape. Challenges like managing redundant conversation paths and ensuring the correct corporate tone is maintained will line up alongside integration with business systems holding disparate sets of customer data and managing the user experience journey across multiple product lines.
Ultimately, victory in this arena will be taken by firms who are able to select the right component mix for their business, their technology estate and importantly, their user base.
This blog was originally published in the Altus Consulting Blog. You can find the original blog, and many more great pieces here: https://www.altus.co.uk/consulting/blog/archive/chatbot-wars/