“It’s f–ing idiotic. Siri is nearly useless” – Slack founder and CEO Stewart Butterfield
I understand why people might not be as happy with Siri (or any one of the other personal assistants that comes pre-installed with our smart phones) as they had hoped for. I felt much like Stewart did when I got my first Siri powered iPhone in late 2011. And yet I would suggest that the disappointment comes from poor messaging, not necessarily poor technology. If you reframe Siri to that of a “horizontal” agent, whose primary purpose is not to do any job, but rather to enable other agents, you immediately put the responsibility somewhere else – on what I would call “vertical” agents. These agents are truly intelligent, but extremely specialized, in the sense that they can only do one job, such as scheduling meetings, but they do that one job so well, that humans need not apply!
With that world-view, two outcomes are eminent:
- An immediate change in your expectations of your horizontal agent and the way you work with him or her. The new relationship will look more like this: “Hey Siri, could you have Amy set up a meeting between Michael, Dennis and I early next week please.” Amy (x.ai’s intelligent agent) being the verticalized AI doing the actual job.
- The arrival of an intelligent agent market place. Remember how the iPhone (pre App Store) came with a few dozen apps all created by Apple? Today we have millions of Apps available, by various creators. In hindsight, it was quite naive for Apple to believe that it could create and publish all the Apps ever needed. I think it is equally naive to believe that any horizontal agent will hold all the answers to the world’s questions and the ability to do all the jobs needed.
I’m writing this having seen Stewart’s Siri quote in a recent WSJ article, and his commentary on the real market lust being more aligned towards a Samantha, the agent depicted in Spike Jonze’s movie “Her.” – I could not agree more. While we are not very close to human level AI, that does not remove the possibility of humanizing the technology – and we’ve spent an immense amount of time atx.ai to do exactly that.
I certainly applaud Slack’s desire for Slackbot to be smarter, and I understand why Stewart suggested that it should support natural language and utilize machine learning to do often tedious, time-consuming task like scheduling meetings. And I think we’d all agree by now that scheduling meetings should not be a human task.
That said, ~50 scientists, engineers and AI trainers at x.ai worked in the basement for a year and a half to bring this one verticalized AI to life and we still have quite of way to go. This fact, together with Stewart’s comment that “his company will need partners with major AI capability,” suggests to me that much like Siri, Slack could become an enabler, making use of various intelligent agents that integrate with Slackbot.
No matter what the strategy might be, it is worth the travel though, given the massive productivity gains delivered by this world of agents. Stewart himself suggested Slackbot could boost a company’s productivity by 20-30%, and that’s something I’m be happy to echo with a few recent x.ai customer quotes:
“I’m not exaggerating when I say Amy has saved me close to 10 hours a week.”
“I predict that we can increase our sales efficiency by 30-40% (which translates into well over $250K per month) just by using Amy.”
Do you agree with the idea of horizontal and vertical agents and the arrival of an intelligent agent marketplace? If not, tell me why I am crazy 🙂
Dennis R. Mortensen
CEO and Founder of x.ai
*Originally appeared on LinkedIn Pulse, here.