Slack has introduced new AI features powered by Anthropic’s Claude model to automate tasks like document retrieval and name pronunciation, aiming to streamline workflows while maintaining strict security and privacy standards. Early results show high user retention and creative use cases, with the AI acting as a personal assistant that only accesses information available to each user.
Slack has recently introduced new AI capabilities, which are currently being used internally by the company, including by CEO Marc Benioff. The AI features are designed to streamline workflows and save users time by automating tasks such as finding documents and extracting relevant information. For example, one use case described was asking the AI to locate the most recent all-hands document and to provide pronunciation guides for new hires’ names, which the AI accomplished by analyzing presentations and inferring likely pronunciations based on nationality.
The underlying technology for these new AI features is powered by Anthropic’s Claude model. Slack emphasizes that while they are currently using Claude, they remain model-agnostic and are open to exploring other AI models in the future. The integration of Claude allows Slack to quickly bring advanced AI capabilities to enterprise customers, helping them work more efficiently and effectively.
A key focus for Slack in deploying these AI features is maintaining strict security and privacy standards. The company ensures that customer data never leaves their security boundary and is not used to train the AI models. This approach allows organizations to benefit from AI without compromising on data privacy or violating existing security policies. Slack has also invested in sophisticated prompt and context engineering to make the AI a deeply personal assistant for each user.
Security remains paramount for Slack, both from the perspective of individual users and enterprises. The AI acts on behalf of the user, accessing only the information that the user can access, such as direct messages, private channels, and public channels within a Slack instance. Slack reassures customers that their data is protected and that the AI operates within the same security framework as the rest of the platform.
In terms of measuring the impact of these AI features, Slack does not track individual productivity but instead monitors overall usage metrics. These include user penetration, week-over-week retention, and message intensity. The company has observed that retention rates for the AI features are higher than for any previous Slack feature, and message intensity continues to grow. Qualitative feedback also indicates that users are leveraging the AI for creative tasks like brainstorming, highlighting the broad value and time-saving benefits the new capabilities provide.