Microsoft CEO Satya Nadella has predicted the “death of SaaS,” saying that the future of software will be dominated by AI agents that directly interface between systems and people. As we all grab our tubas and take a running leap onto the passing bandwagon, let’s pause just for a moment, to think through how to approach this.
You’re intelligent, right? Not even artificially so. So I’d like you to try an experiment: Sit down in front of any transactional web site or tool. Try Expedia, for example. Now call a good friend whom you know well on the phone – audio only – and ask her to pretend she’s going on a business trip. She has to decide to where, when, etc. Your job – intelligent being that you are – will be to act as her Agent, interfacing between her and the data (Expedia). G’head. Do it. I’ll wait…
“I want to go to Atlanta on Monday for 3 days.”
“Do you mean, returning Wednesday?”
“No, Thursday.”
“Ok, let me read you some flight options. There’s a flight on Delta at —“
“I hate Delta.”
“Ok, no Delta flights. There’s a flight on Pan Am at 8am —“
“You think I’m going to wake up at 4am to make an 8am flight? Are you nuts?!”
“Sorry, ok, let me start with 10am flights”
“Oh, for G-d’s sake, can you just text me the list of flights? Maybe just print it and fax it to me!”
“You know what? If you don’t want my help, you can stuff it!”
“I NEVER ASKED YOU FOR YOUR HELP! And YOU know what? I’ve never liked you anyway!”
I’m sorry I destroyed a lovely friendship, but I needed to make a point: Here you were, undeniably an Intelligent Agent. Maybe you should’ve known she wasn’t the type who’d wake up at 4am. Maybe, over time, you’d understand her travel patterns well enough that you’d be able to do what my assistants at Warner Bros. or Fox could do and anticipate my needs well. But even under the best of circumstances, Data Presentation and Option Menus MATTER.
So, with all respect to Microsoft’s Satya, I think he has this assertion that AI is the death of SaaS or applications and web sites wrong. The holy grail of applications and web sites has ALWAYS been customization; understanding the user and orienting the human-computer interface – visual or otherwise – to his or her preferences, habits, etc. AI is a huge leap forward for achieving THAT goal. But it is NOT – nor should it be – a mechanism for replacing tools that humans use to make selections and give commands.
It comes down to this: In pretty much any context, a human being operates best when she understands the context, the environment, the options. This applies as much when booking a flight as it does when shopping on Amazon or selecting your health plan. And, for that matter, the same applies when walking into a supermarket or entering a nightclub.
As your company begins to apply AI to what it does – whether for your end-consumer, business partners, or internal employees – ask yourself these key questions:
- Does the user know why they’re here, what this application is FOR?
- Will the user be aware of all of the important things they can do or options they have?
- Will the user clearly understand what they have chosen to do and feel confident in their decision?
- Will the same actions taken by the user have predictable and reproducible results each time?
- Does the user feel “In Control”?
If your AI implementation does not provide the key assurances I have described here, you are taking a huge chance that the he will lose trust in your product or application. Maybe do it anyway, but think it through.
Now I’m going to go into Uber Eats and just say “I’M HUNGRY”… and see what arrives!
And anyway, who you gonna trust? The CEO of Microsoft? Or ME???
Podcast: Play in new window | Download