Salesforce announces Slack AI with unread message summaries and more

A month after launching a new experience for Sales reps and a redesign that drew both love and hate from its global user base, the messaging app Slack is moving forward with its mission of making collaborative work easier for enterprises.

Today, the Salesforce-owned platform announced Slack AI, a new set of generative AI smarts that will be built right into the platform’s messaging interface, and, ideally, will allow users to save time and be more productive.

The features are set to be demonstrated at Salesforce’s annual Dreamforce conference next week, along with a new Lists capability for better management of assigned work, and an updated workflow builder that will allow even more users to build automations and get things done within the platform. 

“Slack started off as a channel-based messaging platform but we’re really evolving it into an intelligent productivity platform,” Rob Seaman, the SVP of enterprise product at Slack, told VentureBeat.  “There are three key areas of the product we’re focused on:

1. “One is collaboration, which is what everybody knows and loves
2. Two is knowledge, where Slack becomes the most important knowledge repository for your entire company;
3. The third is automation where we’re really trying to let every single human in Slack automate their work and make themselves more productive.”

Not all of these features will be made available right away, the company says.

Making work easier with Slack AI

Back in May, when Salesforce announced Slack GPT, the company promised LLM-powered features in different areas of the product.

Now, with the announcement of Slack AI, the first batch of features is rolling out: Channel recaps, thread summaries, and search answers. 

The first two, as the names suggest, will provide users with AI-generated highlights and summaries for channels and individual conversation threads, giving users a way to quickly get up to speed on what matters the most, without going through all the messages right from the beginning. 

This will be particularly useful in cases when users have been out of the loop. For instance, if a subject matter expert has just been added to a thread related to incident management, they can quickly go through the summary to understand what’s wrong and provide required suggestions for help.

Similarly, if a user has been PTO, they can quickly generate highlights from when they last read a message and cut straight to what’s important. This can be used to extract key themes from feedback channels or draft status reports for project channels.

Seaman noted that the best part about channel recaps is that the feature displays the source along with the highlights, adding trust and transparency into the user experience.

This way, when going through a highlight, users can easily click through and check where the information has been pulled from and go into the details.

Next, with Search Answers, Slack is integrating generative AI into the search experience of the platform, allowing users to simply ask natural language questions to get clear, well-summarized answers (imagine a ChatGPT-like experience).

The feature taps the collective knowledge within Slack, including relevant messages and all the context they hold, to provide the answers within seconds, whether it’s about a project, team process, feature launch or something else.

However, it must be noted that this new experience will not replace the usual search experience. The summary will appear on top of the regular search results that Slack provides, covering relevant messages, files, and channels. 

Slack’s proprietary LLMs power the new Slack AI features, Seaman said while noting that they are all hosted within the company’s own virtual private cloud and nothing from user prompts goes outside the four walls of the VPC.

“The key thing with AI for us is it’s built on the secure foundation of Slack. These native AI capabilities are going to offer the same security and compliance that customers have come to expect from Slack. We also have additional security guarantees that ensure no data is sent to third parties, no data is used in third-party model training and there’s no data leakage across tenants or customers,” he noted. 

But there’s more

Along with the new AI features, Slack is also bringing Lists and an updated workflow builder for enterprise users.

The lists feature loops in work management capabilities into the flow of communication on Slack, allowing users to create lists of active projects (from marketing campaigns to product launches) assign them to relevant parties, and track their progress all the way through completion.

“Teams can manage all types of work with lists – that’s everything from triaging IT requests to reviewing a set of legal approvals to managing a roadmap or cross-functional projects,” ” Seaman explained. “And because it’s built into Slack, teams can collaborate on lists with the same ease and speed that they can on channels. Mentioning teammates is going to send them alerts just like mentioning them in a message does, sharing out to channels is a seamless experience and each item in a list can support a rich set of back and forth with a thread on each item.”

Meanwhile, the improved workflow builder allows teams to create automations without any coding. Using workflow connectors from Google, Asana, Jira and other platforms, they can integrate multiple tools into a single workflow to automate tasks across Slack. Beyond this, it will also come with a new automation hub, which will provide built-in templates to quickly get started and better ways to share, remix and reuse workflows over time. Users can even include Salesforce Flow automations and custom apps, hosted in Slack, into their workflows, the company said.

Availability

While all these features promise a new, improved Slack, they are not ready to ship — yet. 

According to the company, Slack AI and the work management features will be piloted this winter and rolled out sometime in 2024. Meanwhile, the improved automation builder is now available to users – with its hub set to debut later this month.

However, it must be noted that the company is just getting started on the AI front. According to Seaman, the three AI features announced today were found to be delivering the most value in internal testing. The company is also exploring additional use cases, such as generation in Slack Canvas, which might debut at a later stage.

“There’s much much more that we can and will do,” he said.