Salesforce has rebuilt Slackbot into an AI agent that can search workplace data, draft documents and carry out tasks for employees, giving the company a more direct answer to Microsoft and Google in workplace AI.

The new Slackbot is generally available to customers on Slack’s Business+ and Enterprise+ plans. It is designed to work across Slack conversations, Salesforce records, Google Drive files and calendar information, turning the collaboration platform into a proposed operating layer for enterprise work. The launch matters because Salesforce is trying to make Slack central to the shift from chat-based assistants to AI agents that can act inside business systems.

A new architecture behind the familiar Slackbot name

The product retains the Slackbot name, but Salesforce executives described the current version as a complete technical replacement for the older tool, which handled reminders, notifications and other basic tasks. The new system is built around a large language model, enterprise search and connections to external data sources.

Slack Chief Technology Officer Parker Harris told VentureBeat that the redesigned product can draw on information users are already authorized to access. Salesforce says it does not train its models on customer data. Harris argued that retaining customer conversations in a model’s training data would make it difficult to preserve the permission boundaries that apply to the original information.

The initial model provider is Anthropic Claude. Harris said compliance requirements were a major factor, noting that Slack’s commercial service operates under FedRAMP Moderate certification for U.S. government customers. He also said Salesforce plans to add other model providers, including Google Gemini, and has not ruled out OpenAI.

That approach reflects Salesforce’s broader position that foundation models are becoming interchangeable infrastructure rather than the main source of product differentiation. In this case, Salesforce is betting that access to business context, permissions and workflow tools will matter more than exclusive control of a particular model.

What the demonstrations show—and what remains unproven

In a demonstration described by VentureBeat, Slackbot analyzed customer feedback, examined a usage dashboard image, compared qualitative and quantitative findings, and searched Salesforce for accounts that might qualify for an early-access program. It then assembled the results in Slack Canvas and checked stakeholder calendar availability for a review.

The workflow illustrates the product’s intended role: not simply answering a question, but moving from research to a shared document and then toward coordination. Slack Chief Product Officer Rob Seaman said Canvas creation represents an early example of Slackbot making an internal tool call. Additional third-party tool calls are planned.

The strongest adoption and productivity figures in the report come from Salesforce’s own internal testing and participating customers, rather than an independent study. Salesforce said it had made the product available to its roughly 80,000 employees, that two-thirds had tried it, and that 80% of those users continued using it regularly. The company also reported a 96% internal satisfaction rate and claimed employees saved between two and 20 hours per week.

Pilot customers offered similarly positive accounts. Beast Industries employees reportedly described saving at least 90 minutes a day, while an executive at Engine estimated a 30-minute daily reduction in context switching. These examples are useful signals of interest, but they remain vendor-selected testimonials and should not be treated as representative productivity evidence across Slack’s customer base.

Slackbot enters a crowded enterprise AI market

Slackbot is competing with Microsoft Copilot, which is embedded in Teams and Microsoft 365, and Google Gemini, which is integrated across Workspace. Salesforce executives argue that Slack has an advantage because employees already conduct conversations, share documents and make decisions there. The company’s pitch is that an assistant grounded in those existing workflows requires less setup than a separate AI application.

Slack is also becoming a distribution channel for other AI agents. Anthropic has previewed Claude Code for Slack, while OpenAI, Google and Vercel have built agents for the platform. Seaman said many new applications being deployed to Slack are agents, reinforcing Salesforce’s ambition to make the platform a common interface between employees and specialized software.

Harris described Slackbot as a potential “super agent” that coordinates with other agents and tools. Salesforce expects Slack to become a Model Context Protocol client, allowing Slackbot to use capabilities distributed across the software ecosystem. At the same time, Harris cautioned that broad multi-agent coordination is still immature and said Salesforce would focus on practical customer outcomes rather than demonstrations involving large numbers of autonomous agents.

The pricing is intended to reduce initial adoption friction: Salesforce says Slackbot carries no separate charge for eligible Business+ and Enterprise+ customers. However, buyers will still need to assess the wider cost of connecting Salesforce data to external systems. VentureBeat cited concerns from Fivetran CEO George Fraser that changes to Salesforce API pricing could increase costs for data replication and third-party AI access, potentially pushing customers toward Salesforce’s own data and agent products. Salesforce has characterized its pricing approach as consistent with industry practice.

Why the launch matters for builders and CIOs

For product teams, Slackbot points to a practical design pattern for enterprise AI: place the agent inside an established work surface, preserve existing permissions and let the system move between search, analysis and action. That can be more valuable than a standalone chatbot, but only if retrieval is accurate and tool calls are tightly controlled.

Permission-aware access will be especially important for companies with sensitive customer, personnel and financial data. Slack’s claim that users see only information they already have permission to view may simplify security review, as Beast Industries reported. It does not remove the need for customers to validate indexing, data connectors, audit trails and the consequences of an agent taking action based on incomplete or ambiguous context.

The current product also shows that agentic workflows are arriving incrementally. Calendar reading and availability checks are available, but direct meeting booking was expected several weeks after launch. Image generation was not supported at launch. Integration with competing CRM systems such as HubSpot and Microsoft Dynamics remained unspecified.

For Salesforce, Slackbot is more than a feature release. It is a test of whether Slack’s accumulated workplace context can become a durable competitive advantage as models become widely available. For enterprises, the key question is whether that context produces reliable decisions and measurable time savings, rather than simply making it easier to ask questions in chat.

What to watch next

The next signals will be Slackbot’s rollout beyond internal users and pilot customers, including independent measurements of task completion time, error rates and user retention. Customers should also watch for the promised mobile release, meeting-booking capabilities and additional model providers.

Tool access will be another important test. Salesforce’s plans for third-party calls and Model Context Protocol support will show whether Slackbot can safely coordinate with external systems or remains primarily a search and drafting assistant. Buyers should seek clarity on logging, approval steps, connector coverage and pricing for Salesforce data access before expanding deployments.

Creati.ai perspective

Salesforce is making a credible strategic move by combining Slack’s conversational context with enterprise permissions and access to business records. The product’s advantage, if it holds, will come less from Anthropic Claude itself than from how reliably Slackbot can retrieve the right information and complete bounded workflows without forcing users to switch applications.

The evidence is still early and largely Salesforce-reported. The launch establishes Slackbot as a serious workplace AI contender, but sustained value will depend on independent productivity results, transparent governance and whether customers can connect the systems they actually use at a predictable cost.