Weeding through the hype, Gartner offers three use cases for voice technology that the enterprise can actually use right now, covering Salesforce Einstein and Amazon Alexa.
At TrailheaDX 2019, Salesforce’s Michael Machado spoke with TechRepublic about the unique challenges of building Einstein Voice Assistant within the company’s highly-customizable platform.
While voice assistants such as Siri, Cortana, Google Home, and Alexa have been embraced by the consumer, their adoption within enterprise is still in its embryonic phase. And rightfully so, Gartner senior director analyst Kyle Davis, would argue.
“There’s a lot of, unfortunately, barriers to entry for getting into voice,” Davis said, pointing to frustrations users may experience when learning how exactly to communicate with the devices.
Presenting at the Gartner Application Architecture, Development, and Integration Summit in Sydney last month, Davis offered three actual use cases of the technology.
1. Logging client interactions
Davis painted a picture of a busy sales rep in the field. If the individual is vigilant, they’re taking notes as they go, but those notes still need to be entered into a computer at some stage.
Demonstrating Salesforce’s Einstein Assistant, Davis, using the mic on his iPhone, dictated to the app and had his verbal notes automatically organised into sections that had headings auto-populated from his dummy company’s CRM.
“If you’re missing data, then you’re getting inaccurate results,” he said. “So we see this as one opportunity for being able to improve experience overall, and be able to enable your sales reps.”
Salesforce handles the organisation of the text and the auto-population of the system in use in the office.
Einstein is what Salesforce uses for analytics; the tech giant has a bit of a head start in the chatbot space, but voice assistant is a brand new play currently in private preview.
“It’s something you can start piloting … this is really not something that’s completely intuitive yet, I still feel like Salesforce has some stuff to work out, but they’re doing some pretty good advances as far as making it to where you can capture some information,” Davis said.
2. Ordering office supplies
Ordering paper, notebooks, coffee, pens — the regular office items that everyone uses but no one seems to replace until the supplies have been bled dry — is another practical use case Gartner has identified.
Davis used Amazon Alexa for his demonstration, giving the assistant a simple chore: “Alexa, please order paper”.
“It’s really just one common solution that’s always being leveraged, building a custom, private skill that’s location aware,” he explained.
He used a new concept from Alexa for Business — a shared device.
“I have this Echo provisioned as a shared device and what makes it different than what you have with either your Google Home at your house, or your Echo Dot at your house, is that this is on our enterprise accounts, and so it doesn’t have anything like the ability to do account making, it’s not going to have any personal information on there, it can’t do voice profiling,” Davis said.
“It’s not meant for sensitive information, it’s meant for being placed in the common area available to employees to do things.”
The device is locked down in the backend to a specific room, and the item is being ordered to a specific cost centre. What this means is that the user would not need to know anything except what needs replenishing.
“Sometimes it’s hard to figure out how the equipment works in one room compared to another room. We also know that equipment will break, maybe you’re missing a cable, or the cables were swapped out, and so you don’t know what is the right order to get those cables to get everything working again,” Davis said, painting a picture of a standard conference room within any office environment.
“So being able to create things like support tickets and whatnot is difficult … this is another opportunity, another area, we see a lot of people using voice.”
Again using Alexa for Business, Davis touted its use as it doesn’t require any custom coding due to Amazon already building them for enterprise to use right out of the box, with minimal configuration.
This allows for devices such as projectors, speakers, and phones to be hooked up, and synced to the likes of Office 365 to allow for voice commands to be used to book a meeting room, as one example.
There’s always a ‘but’
While these three use cases may appear simple, about 95% of the total effort is a development effort. Davis said there’s a lot of backend code that needs writing, of which, only about 5% is “magical”.
“And that’s what you’ll get from Amazon, so there’s not a lot there that you’re going to get from [others] at this point, there’s still a whole bunch of code that you’re writing,” he reiterated.
However, using a concept of a room for a shared device allows for devices to be provisioned at scale.
“But even for just being able to do that one little thing, I ended up hitting … pretty close to 150 lines of code,” he said, referring back to the example of ordering office supplies. “You can see that there’s still a fair amount of investment that you’re going to do from a developer standpoint, just to provide an experience like that.”
To just order paper, thought needs to be given to the multitude of ways someone could ask — boxes, reams, copy paper, paper, printer paper — all of these need to be accounted for.
“If you start to think about every possible way someone can ask a question, and the fact that questions might have a question asked as a response versus an answer, you start to get into what we call dialogue,” he explained.
“We have to mimic that, keep track of that we’re in this world of voice experiences, and so your lines of code very quickly get into thousands.”
Davis said while it’s a good idea to dabble in some of these capabilities, it’s an even better idea to be mindful that it’s going to be a significant time investment to reach a place where the end result is actually valuable.
“Is voice technology good enough to bring into the enterprise? I would say it is,” Davis concluded.
“But with that said, it’s not a blanket approval …. I still think there are some safeguards that we have to think about.”