AI helps organisations assess team morale in Slack

Tokyo-based company AIR has developed an AI tool that allows managers to monitor team morale based on public Slack conversations.
Written by Tas Bindi, Contributor
(Image: Supplied)

A bad relationship with a direct manager is among the top three causes for employees leaving their companies, according to AIR, a subsidiary of Japanese HR giant En-Japan.

However, for managers, particularly those in charge of large or remote teams, keeping up with the emotions of every employee is no easy feat.

AIR, which develops HR technology solutions, is looking to address this problem by using a combination of artificial intelligence and natural language-processing technology to monitor team morale.

"From our own experience and our friends', we felt there was a disconnect between managers and team members," Frederic Peyrot, COO at AIR, told ZDNet. "Conducting weekly one-on-ones when you have 20 or more people in your team is impractical. Therefore, we designed Vibe as a tool to help managers follow the morale of their team, see what causes the team vibe to rise, but also be notified when the morale drops.

"Many tools (such as Culture Amp and Officevibe) have been trying to address that problem, but they often rely on surveys which, since they don't provide real-time information, prevent managers from acting in a timely manner."

AIR's new product Vibe is able to analyse Slack conversations in real time and detect the mood of the team, breaking down the morale into five emotions: Happiness, irritation, disappointment, disapproval, and stress.

Dots are displayed along a morale graph whenever the team mood has suddenly dropped or surged. By clicking on the dots, managers get an understanding of what caused the change by highlighting the main topic related to it. For example, it would display a message such as: "Slow internet had a negative impact on your team morale with an increase of irritation one hour ago."

Vibe, which was funded by En-Japan, also shows which team members were discussing the problem, allowing managers to go back to the conversation for further context.

Peyrot explained that the tool does not monitor private channels or direct messages -- only conversations in public channels that users provide access to.

Once access is provided, everything that is visible to the manager is also visible to team members, Peyrot said.

"Since we only process already public information, we do not reveal any information that was private in the first place. We often make the analogy that it would be the same as having a super user analysing all public Slack discussions," Peyrot said.

"In the end, a question managers often ask us is whether they should disclose it with their team members. We tell them that it is better to be transparent, as the tool would be better accepted by team members if they can use it themselves."

Vibe currently operates on a freemium model, with two paid options: $50 a month and $120 a month. Peyrot admitted, however, that the pricing might change to a "cost per user per month" structure with no channel limitations.

Peyrot said the company will soon release Vibe's "reports" feature for enterprise members, which will distill key information about a company's morale on a monthly basis. Peyrot said the company hopes it will take over engagement surveys that companies currently conduct, which can cost $50,000.

Vibe is being used by more than 500 companies globally, 80 percent of which are based in the US, with users also coming from Australia, India, and the United Kingdom. Given the tool's algorithm only works with the English language, Peyrot said the AIR team will focus on English-speaking countries.

While Slack has a strong enterprise customer base with 5 million daily active users as of January, Peyrot said AIR is considering opening Vibe to other platforms beyond Slack, such as Yammer and Facebook, to grasp a wider market.

Earlier this year, Slack itself revealed that it was looking to improve its core communications platform through artificial intelligence and machine-learning technology. During the launch of Enterprise Grid, the company also revealed that it would be introducing new search, learning, and intelligence features powered by machine learning that will allow Slack to behave more like a digital assistant.

These features include "universal search", which gives users instant answers to questions powered by the collective knowledge of their organisation; "channel highlights", which visually highlights important messages within and across channels for users looking to quickly catch up on what they've missed; and "daily briefings", which provides a bird's eye view of the top 10 most important messages to read in the morning or at the end of the day.

It's unclear, however, whether Slack is mulling a foray into sentiment analysis.

Editorial standards