Slack uses AI to outsmart ‘cockroach’ emails in offices

Emails are the cockroaches of the internet and almost impossible to eradicate, says Slack.

Noah Weiss, Head of artificial intelligence at Slack

Emails are the cockroaches of the internet and almost impossible to eradicate, says Slack.

That’s a notable statement from the maker of workplace teams collaboration software that some see as best placed to eradicate the hundreds of irrelevant emails that an employee might process each day.

But artificial intelligence and machine learning are making strides in improving internal company communications which is what Slack worries about, says Noah Weiss, the head of Slack’s Search, Learning, & Intelligence group.

As well as offering a team approach to internal communications, it can make sense of the huge pile of data that organisations generate daily. But email from the outside world stays, he says.

“All of a sudden you see a massive reduction in total email, all your internal discussions happens in one place, and your email box is basically for all external communication.

“Slack is not trying to kill email in any way. If anything, email is like the cockroach of the internet. It may never die.”

Overnight Slack announced a highlights feature that places the most relevant top 10 messages you’ve missed at the top of your feed, curated using machine learning. That follows the launch of a new home page for Slack users that also AI and machine learning to present an organisation-wide catch-up view across all the channels.

But it’s success will depend on whether Slack’s AI does indeed deliver information staff really need to know and does not omit important communications. It comes down to trust.

Slack says its AI analyses factors such as the channels you care about the most, which people you interact with the most, the way you interact with them, and your level of authority among your team.

Weiss previously was senior vice president of product management at Foursquare and a lead product manager at Google.

At the moment Slack is upping its game as it tries to fend off a challenge from Microsoft which is adding team messaging to its business software.

Weiss sees the two offerings as fundamentally different. “We’re very glad that Microsoft has decided to add its 35th application to its office suite which is to be a Slack-like tool,” he says.

But Microsoft was about selling a big bundle of software, Slack was about integrating a business’s existing software tools with Slack messaging, he said.

With 6.8 million weekly active users Slack has proved a highly visible venture since Canadian Stewart Butterfield, a co-founder of Flickr, started Slack four years ago.

However some find Slack’s business model and its long term profitability hard to work out.

Weiss’s job is to integrate AI in the messaging platform. He said organisations lived in a state of analytics overload and needed help.

“There’s been this explosion of data. But data is data. It’s not insights. It’s not recommendations. What’s interesting about machine learning is that there’s finally a place to take all that data and make meaning out of it.”

He said getting AI to understand the needs of users was key. “We’re at the critical juncture that Instagram was that may be a year ago, where Twitter was at, what got them to where they are … over time they built up a deeper understanding of each individual’s preferences.”

Slack bots were now active in chat rooms. “A bot can participate in a channel with a group of people and can help put interesting information into a channel.

“You can have a bot that says when a new sales lead comes in, and puts it into the sales channel. Or a customer service request comes in, and people interact in that channel.

“You can connect all the different services that your company uses so that all that information is flowing through Slack instead of you having to switch tools 20 times a day. You can have that all in one place.”

Another example was The New York Times’s Blossom bot. “They have an algorithm that can project internally when a story might go viral. Instead of waiting for an editors’ meeting the next day, to figure out which of these we should enhance, they want to have those conversations to happen in real-time.

“The machine pushes that prediction into Slack and the editors in real time discuss the merits of whether we publish this to our Twitter feed. Should we send an email with this? With the click of a button from Slack they can initiate a message they send to millions of customers.”

He said Slack had proved popular with publications. The New York Times, The Washington Post, The LA Times, The Verge, Quartz, News Corporation (publisher of The Australian) and 21st Century Fox were examples.

“The LA Times uses Slack more as a content management system. All the conversation that happen about a particular piece, instead of being siloed in some CMS, now you are able to search those conversations in the future when you’re writing the next story.”

Given concern about earthquakes in California, the LA Times had developed a bot that linked to the city’s public information system and made the paper aware of earthquakes in Slack in real time.” It alerts directly into a Slack channel that is set up for quakes. Within milliseconds they know about an earthquakes and can break the story first.”

eCommerce company Jet.com gets alerts from Slack in its inventory management system prompted by AI. “It knows when the next shipment is coming in and if there’s insufficient upcoming supply, it will generate an alert.”