As a corporate communications platform, Slack stores institutional knowledge, but traditional search tools have trouble finding it. Today, Slack released an AI-powered search engine and the ability to summarize content within conversations to make that information more accessible.
The chief product officer of Slack, Noah Weiss, believes the platform organically accumulates business data that is unorganized. Finding such concealed information is the challenge. “The punchline to all of this is that now this new wave of generative AI capabilities allows us to extract a whole new set of meaning and intelligence out of all that analysis that has been created for years [on our platform],” Weiss told Eltrys.
Slack introduced generative AI at the Salesforce World Tour in New York City in May. With SlackGPT, its own generative AI for Slack content, it was more of a broad call to action.
Today’s announcement details how to apply that. Weiss believes summarizing channel information helps staff catch up after time off or avoid reading extensive threads to grasp the point. When you request a channel summary, Slack’s AI model creates a summary of all the subjects covered, along with references to demonstrate how the model built each portion, which Weiss says was crucial to the feature’s design.
We display extensive context for whatever region you dive into. We thought about openness, trust, showing our work, and letting people dive in to understand more, he added.
Like ChatGPT, the startup lets users ask natural questions using Slack content instead of internet material, so they may ask, ‘What is Project Gizmo?’ Slack AI then provides a response with a source to show where it originated from and whether it can be trusted.
Users may rate each answer as excellent, terrible, or neutral to help the model learn about response quality and system developers evaluate model performance.
He said the model was a blend of major language models but wouldn’t offer more. “We found that they perform differently with different speeds and quality. We worked hard on prompt engineering and fine-tuning Slack data models.
On top of the licensing fee, Slack AI with search and summarization costs extra for corporate subscriptions. Slack did not offer fee figures, but it is now only accessible in English in the U.S. and U.K. but will soon be available in other languages.
AI-powered search and summarization in Slack
