Slack is launching a series of broad artificial intelligence features designed to simplify daily tasks and transform the messaging platform into a central hub for enterprise productivity. This move marks Salesforce's direct challenge to Microsoft's dominance in the workplace AI space.

The updates, expected to roll out over the coming months, will include AI writing assistance directly embedded in Slack canvases, contextual message explanations, automated action item identification, and enterprise search capabilities across multiple connected business applications. At the same time, Salesforce has begun restricting external AI companies from accessing Slack data, creating a "walled garden" model that reflects the current trend of platform integration in the industry.

Letter AI Robot

Slack AI: The "Context" Advantage Deep in Workflows

Shalini Agarwal, Vice President of Slack Products at Salesforce, emphasized in an exclusive interview with VentureBeat: "Unlike some AI tools that operate outside of workflows, Slack's AI is integrated into every aspect where work happens—conversations, decisions, and documents. The key difference lies in context, which is presented in Slack as both structured and unstructured data."

This strategic timing highlights the increasingly fierce competition in the $45 billion enterprise collaboration market. Since Salesforce acquired Slack for $27.7 billion in 2021, Microsoft's Teams platform and its Copilot AI assistant have made significant progress. Google is also actively promoting Duet AI within its Workspace applications, with all three tech giants vying for enterprise customers who are increasingly focused on AI-driven productivity improvements.

Automation and Intelligent Assistance: Enhancing Daily Collaboration Efficiency

Slack's new features move away from the traditional AI assistant model where users had to actively seek help, instead displaying relevant information proactively and automatically performing daily tasks within existing workflows.

AI writing assistance will soon be available in Slack canvases, allowing teams to automatically generate project summaries from conversation threads, extract action items from brainstorming meetings, and reformat meeting notes into structured updates. Combined with Slack's existing AI-powered meeting notes functionality, this feature will create an end-to-end document workflow.

Agarwal said, "Artificial intelligence should feel effortless and seamless; you shouldn't have to work hard to use it." She revealed that since Slack launched AI, customers have compiled over 600 million messages, saving a total of 1.1 million hours for all users.

More innovatively, Slack will introduce the contextual message explanation feature. When users hover over unfamiliar terms, abbreviations, or project references, the feature will activate automatically. It leverages organization-specific vocabulary and conversation history stored in Slack, aiming to resolve common issues faced by new employees and cross-team collaboration.

Enterprise Search: The New Battleground for Workplace Data

A core component of Slack AI strategy is the widely available enterprise search feature. This function allows users to query information from connected applications (including Salesforce, Microsoft Teams, Google Drive, Confluence, and Box) from a single interface within Slack.

This feature aims to address long-standing productivity losses in modern workplaces. According to Slack's research, employees spend about 41% of their time on repetitive tasks such as searching for information between disconnected systems. By positioning Slack as a unified search interface for enterprise data, Salesforce is boldly trying to become the primary work center for knowledge workers.

Slack does not establish point-to-point connections between applications but positions itself as a universal translator for workplace information, acknowledging that most organizations' data is scattered across dozens of applications but desperately needing better ways to find and use this information. For IT departments, Slack promises to minimize deployment complexity, with connectors immediately usable once launched.

Data Restrictions: Salesforce's "Walled Garden" Strategy

Although Slack has opened up its search functionality to connected applications for customers, Salesforce has been actively limiting external AI companies' access to Slack data. In May of this year, the company revised its API service terms, prohibiting bulk data exports and explicitly banning the use of Slack data to train large language models.

This move will affect third-party AI search companies like Glean, which previously indexed Slack conversations and other enterprise data sources to provide a unified search experience. Under the new restrictions, these companies can only access Slack data through real-time search APIs, and with strict limitations.

Salesforce is taking a calculated risk. By restricting access to Slack data, the company bets that its own AI capabilities will outperform external alternatives. However, enterprise customers continue to show a preference for choice and flexibility rather than being locked into a single vendor. If competing AI platforms can achieve significantly better results using data from other sources, Salesforce may lose customers to alternative messaging platforms offering more open integrations.

These restrictions highlight the immense value of workplace conversation data. Slack exchanges over 5 billion messages per week, and the platform contains "company history and all information across teams and projects," as Agarwal described. This conversational data provides unique value in the enterprise software space: unstructured, context-rich information about how work is actually done, rather than formal documentation about how work should be done.

Enterprise Security and Early Productivity Results

Salesforce has built its AI features around its "Einstein Trust Layer", emphasizing that customer data will never leave the company's infrastructure or be used to train external AI models. This approach addresses concerns about data sovereignty, which previously hindered AI adoption in regulated industries.

Agarwal said, "Protecting customer data is Slack's top priority. Customer data remains within Slack, and Slack does not share customer data with large language model (LLM) providers, nor does it use customer data to train LLMs."

The platform's AI features inherit Slack's existing enterprise-grade security controls, including support for FedRAMP compliance, encryption key management, and international data residency requirements. Search results automatically respect user permissions within connected applications, preventing unauthorized data leaks.