Salesforce introduces AI Summarist: a conversational AI tool that helps users manage their information consumption based on their work preferences

Slack is an efficient productivity tool used by professionals and teams around the world to stay in touch and stay up to date with the latest work-related information asynchronously. But as the number (and types) of conversations in Slack that a user follows increases, their overall information consumption also increases. This can cause users to experience increasing cognitive load, which can be distracting and hinder deep work.

A user’s cognitive load can often increase when they actively participate in multiple Slack channels throughout the day while running their business.

To help users efficiently deal with the constant stream of content updates from various channels, DMs, and threads on Slack, Salesforce has developed new work using state-of-the-art conversational AI. By reading summaries to catch up on news, users of their service dubbed AI Summarist can tailor their news consumption and (indirectly) manage their notifications on Slack channels.

The advantages of AI Summarist are mentioned below:

• Improved insights: Users can receive automated summaries from channels less directly relevant to their day-to-day work while staying informed about the channels that matter most.
• Personalization of information consumption: Users can customize how and where they get their business information. They can customize how often and when they receive summaries using AI Summarist.
• Users can discover new (and more) channels to follow by reading previews of recently discovered or suggested channels.
• If a user has been away from work for a few days, they can quickly catch up on messages by reading channel summaries, from which they can choose to drill down (read the full content) if they deem it necessary.
• Try to choose the main ideas: when an AI summary is tasked with summarizing a large volume of discussions, it uses reactions and responses to determine the most important topics.

Impromptu end-to-end operation of the synthesis process. The following procedures are performed when a summary request is made using AI Summarist:

1. Messages can be sent via the API: calls are made to the Slack API to retrieve all conversation threads for the requested period.
2. Conversation in a less tangled way: Artificial intelligence summarization software can select individual conversations from a large group of communications.
3. Artificial intelligence is the assistant: a summary of discussions is generated using an AI algorithm.
4. Synopsis transmitted: a private summary, priority for the requesting user, is delivered once the time has elapsed.

As AI Summarist understands that the contents of a user’s Slack messages are private and confidential, it takes great care when handling user data and reading private communications. Neither the raw chat logs nor the summary version are stored. Everything is done as needed whenever the user makes a request. The fact that AI Summarist does not log any of the data it summarizes gives customers peace of mind when using it for private and sensitive topics on Slack.

Users can provide critical feedback on the created summary, and AI Summarist only retains metadata about functions invoked by users. AI Summarist’s dialogue synthesis capabilities can be improved by learning more about gaps in user expectations, which can be learned from this feedback. Users are urged to exercise caution when providing comments that could be construed as personally identifiable information.

Check Salesforce Article. All credit for this research goes to the researchers on this project. Also don’t forget to register. our Reddit page and discord channelwhere we share the latest AI research news, cool AI projects, and more.