Imagine the manager of an aircraft maintenance operation is researching production costs, and instead of logging into an app or checking a dashboard, could simply text through Facebook Messenger and ask, “How many technicians are currently working in Oklahoma in Hangar 3B?”
Chatbots make that possible. In fact, that scenario is one that AAR, a global aviation services company that operates in more than 100 countries, is exploring. AAR’s prototype shows one big idea behind chatbots: Let company employees use their preferred messaging platforms, whether it’s a text-based one like Messenger or a voice-based assistant, to ask conversational questions of an intelligent bot, and get answers based on data from the company’s back-end systems.
Founded in 1951, AAR shows that chatbots aren’t just for born-on-the-cloud startups. Its chatbots need to integrate with custom-coded, on-premises applications that may be decades old but are still trusted to handle millions of company transactions a day.
“You don’t really need to reinvent your back-end to modernize or to digitally transform. You need to unlock your back-end’s potential,” says Serdar Yorgancigil, vice president of MRO Solutions and User Experience at AAR. “We can’t just replace a legacy solution overnight or pretend that it doesn’t exist.”
AAR offers a range of aviation services to commercial and government customers, from aircraft maintenance to replacement parts supply to logistics to airlift services. AAR’s initial chatbots are being tested in its aircraft maintenance business, in which the world’s airlines outsource the upkeep of their planes to AAR in what’s called a maintenance, repair, and overhaul (MRO) facility. AAR is the largest airframe MRO provider in the Americas and one of the top three globally.
Learning and Understanding
AAR’s IT team is using Oracle’s intelligent bot builder to create chatbot prototypes. That bot-building capability combines natural language processing and artificial intelligence so that developers can create bots that understand questions—and then learn from those questions to get smarter as they’re used.
“We continuously look for ways to make information easier to access and more available to the aviation industry,” says Kevin Larson, chief information officer at AAR. Larson sees chatbots built with Oracle’s intelligent bot builder as complementing AAR’s web and mobile solutions, providing a promising new channel in the company’s efforts to give employees and customers “the right information at the right time, whether they are an airline or AAR executive, a buyer, an aircraft technician, or a warehouse clerk.”
AAR’s first chatbot prototype aims to let its business leaders ask ad hoc questions about their operations without having to ask IT to build them another report or dashboard. “I see chatbots as a great opportunity to provide information to people who don’t typically use our systems, like our senior management, VPs, and general managers,” Yorgancigil says. That kind of leader wants to ask “unique and one-off questions that he or she may never ask again in that same context,” he says.
Oracle’s intelligent bot-building capability is part of Oracle Mobile Cloud Enterprise, an expanded suite for development, operation, and analysis across an organization’s bot, mobile, and web channels. And as part of Oracle Cloud Platform, it can leverage enterprise integration capabilities to link mobile apps and chatbots with enterprise systems via an API.
AI to Understand Context
Oracle’s bot building capability includes a dialog engine on which developers create initial scripts for customer or employee interactions with chatbots, plus natural language processing that can understand the various ways people might ask the same question. Yorgancigil has been impressed by the chatbots’ abilities to learn and understand the context of conversations.
“Once you start having a chatbot conversation, and it’s established that you’re dealing with the Oklahoma facility, if a couple of questions later you just say, ‘Give me the aircraft status,’ it knows that we’re in the context of Oklahoma and applies it as a filter, as opposed to asking you again “would you like to get status for Miami, Oklahoma, Indianapolis,” he says. “It carries over the context of where you were in the conversation throughout the follow-up questions, which I did not expect.”
Developers can also use the bot builder’s dialog engine to write scripts once that then work across multiple messaging channels. Initially, AAR developers will probably offer bots via Facebook Messenger and an internal messaging app, but they do plan to expand to many other channels, which is why Oracle’s “write-once, run-many” capability is so important, Yorgancigil says. “It’s almost like you don’t care about the delivery channel, which is the beauty of it,” he says.
AAR will test chatbots to explore a variety of possible applications, like giving mechanics a hands-free way to order parts on the shop floor. Yorgancigil notes that AAR is building a growing REST/JSON API catalog as part of its ongoing ERP modernization, which it can leverage in conjunction with the Oracle Mobile Cloud Enterprise platform. It took about a month to create AAR’s Interactive Assistant, or AARIA, which now can interact with its aircraft maintenance ERP system to provide status updates. “We’re looking forward to expanding AARIA’s capabilities into our supply chain ERP system,” Yorgancigil says.
By using a cloud service, AAR can take a small-scale, incremental approach as they modernize and take advantage of cutting-edge digital transformation technologies like AI-powered bots. Oracle Mobile Cloud Enterprise will help AAR developers scale up and integrate bots at their pace and also support a global rollout if those prototypes pan out.