Unlike digital-first organisations, traditional businesses have a wealth of enterprise applications built up over decades, many of which continue to run core business processes.
In this series of articles we investigate how organisations are approaching the modernisation, replatforming and migration of legacy applications and related data services.
We look at the tools and technologies available encompassing aspects of change management and the use of APIs and containerisation (and more) to make legacy functionality and data available to cloud-native applications.
This post is presented by Moveworks – the company’s software combines AI with Natural Language Processing (NLP) to understand and interpret user requests, challenges and problems before then using a further degree of AI to help deliver the appropriate actions to satisfy the user’s needs. The commentary below comes from Vaibhav Nivargi, CTO and co-founder of Moveworks.
Nivargi writes as follows…
There’s no shortage of buzzwords that come to mind when speaking about IT modernisation, from data analytics to digital maturity to multi-cloud infrastructures. Although these terms have become somewhat stale, companies are achieving lower costs and greater scalability than ever before by embracing all of the above.
But the truth is – for 99% of employees – the IT team isn’t judged on any of those buzzwords.
For them, what matters is getting tech support the moment they need it, without having to jump through hoops in the process. The real challenge, then, is to speak their language.
When employees email IT to ask for troubleshooting assistance on Zoom (or some other collaboration channel) or to request a license for Salesforce, help desk agents must read through their tickets and assign them to the right subject-matter experts, who must then also read them to decide the best resolution. Often, this process will involve several back-and-forths between the submitter and multiple agents, with some emails getting lost in the inbox and some tickets getting stuck in a queue. That’s why, despite our advanced analytics and automated tools and complex cloud architectures, the average IT issue still takes three days to resolve – a delay that doesn’t suggest ‘modern’ at all.
To solve this language problem, CIOs are considering conversational AI.
Conversation can’t be hardcoded
The notion that understanding language is a key route to supporting employees is not a new idea; in recent years, simple chatbots and dialogue flows have become more and more common in the enterprise. However, these attempts to remove IT agents from the employee support experience have fallen flat for a number of reasons – most obviously, because natural language conversations can’t be hardcoded in advance.
For the sake of comparison, consider what it’s like to call an automated customer service line. First, the bot lists a number of categories and asks you to pick the most appropriate one, which begins a process of slowly narrowing down why you’re calling via a game of Twenty Questions. Often, your response to a question will send you on a loop back to the start, while other times, the list of categories to choose from will be so out-of-date that you can’t decide. At the end of this whole affair – if you’re lucky -you’ll finally get to talk with an actual human, who’s busy helping six other people and who somehow has none of the context you just gave the bot.
To summarise, the hardcoded decision trees that these call centres employ aren’t scalable, they need constant maintenance as the relevant questions and responses inevitably change and worst of all, they force users to stick to an unnatural script, which is not how we communicate. This is exactly the difficulty that’s plagued chatbots and automated scripts in the enterprise: machines can’t carry a conversation, meaning that the IT team gets burnt out reading and resolving routine issues – instead of focusing on high-impact projects.
From request to resolution
Providing autonomous IT support that employees will actually use, in lieu of emailing the help desk, means conversing with them on their terms and keeping pace as they unpredictably switch between topics. If you’re in the business of doing this, you soon realise that breaking down even a single support request requires a combination of many machine learning models – all of which must constantly be improving to combat data drift. Of course, resolving IT requests in seconds takes more than just conversational AI; you also need deep integrations with backend systems and other enterprise software.
What are the ramifications of seamless conversational AI in practice?
For employees, it means that getting IT support takes thirty seconds of painless conversation, not three days of waiting for help. For IT teams, it means more time to concentrate on high-level strategy work, as opposed to addressing routine tickets. And for organisations transitioning to work-from-home, conversational AI allows the remote workforce to get support anywhere, any time – making the home office as productive as the real one.