Salesforce has launched a rebuilt Slackbot that can search enterprise information, draft documents and carry out tasks for employees, turning Slack’s longtime notification helper into an AI agent. The product is generally available to Business+ and Enterprise+ customers, according to VentureBeat AI, marking Salesforce’s clearest attempt yet to make Slack the operating layer for workplace automation.

The move places Salesforce directly against Microsoft Copilot in Teams and Microsoft 365, and Google Gemini across Workspace. It also comes as Salesforce tries to show that AI will expand the value of its software rather than weaken the company’s position in enterprise applications.

What changed inside Slackbot

The previous Slackbot handled relatively simple, algorithmic jobs such as reminders, channel suggestions and notifications. The new system is built around a large language model, a search engine and connectors to enterprise data sources, according to Salesforce executives quoted by VentureBeat.

At launch, Slackbot can work across Slack conversations, Salesforce records, Google Drive files and calendar information. It can also create a Canvas, Slack’s collaborative document format, allowing users to turn an answer into a shared workspace for review and further editing.

A product demonstration described by VentureBeat showed Slackbot analyzing customer feedback, examining an uploaded dashboard image, identifying Salesforce accounts that could be suitable for an early-access program, and assembling the findings into a Canvas. It could then check stakeholder availability on calendars. The ability to check availability is available at launch, while automatically booking meetings was described as a near-term addition.

That workflow matters because it moves Slackbot beyond retrieval. A conventional workplace assistant might summarize a conversation or locate a file. Salesforce is presenting this version as a system that can gather evidence, create an output and begin coordinating the next step without forcing an employee to switch among several applications.

Slack retained the Slackbot name despite the architectural overhaul. Parker Harris, Salesforce co-founder and Slack’s chief technology officer, described it as a fundamentally new product rather than a gradual upgrade to the original tool.

Evidence and claims remain largely Salesforce-reported

The strongest adoption and productivity figures cited in the report come from Salesforce’s own internal testing or customer testimonials, rather than independent research. Salesforce said it had made the new Slackbot available to its approximately 80,000 employees during a test period.

According to the company, two-thirds of employees tried it, 80% of those users continued using it regularly, and satisfaction reached 96%. Salesforce also said employees reported saving between two and 20 hours per week. Those figures indicate strong internal engagement, but they have not been independently verified in the available evidence and do not establish comparable results across other organizations.

Salesforce executives also pointed to organic usage patterns. The company said employees created a shared Canvas containing more than 250 prompts and that social sharing accounted for 73% of internal adoption, based on research by Salesforce’s user-experience team. These details suggest that prompt sharing and peer learning may be important to deployment, but they remain vendor-reported signals.

Pilot customers offered similar accounts. Beast Industries, the company associated with YouTube creator MrBeast, said its security review was relatively quick because Slackbot respects each user’s existing permissions. One employee reportedly saved at least 90 minutes per day, while an executive at Engine estimated that the tool saved about 30 minutes daily by reducing context switching. Such testimonials illustrate possible use cases, but they are not controlled productivity studies.

Salesforce said it does not train its AI models on customer data. Harris argued that using confidential conversations as training data would make it difficult to preserve individual access controls. The more immediate security question for buyers is therefore how retrieval, permissions, connectors and generated actions are governed in each deployment.

Claude powers the launch as Salesforce keeps its options open

The new Slackbot currently runs on Anthropic’s Claude. Harris said compliance requirements influenced that choice because Slack’s commercial service operates under FedRAMP Moderate certification for U.S. federal customers, and Anthropic was the only provider able to meet the relevant requirement when development began.

Salesforce plans to add other model providers. Harris specifically cited Google Gemini, saying Salesforce expects to use it for some workloads, while OpenAI remains a possibility. This points to a model-agnostic strategy in which Slackbot’s value rests less on exclusive access to one model and more on its enterprise search, permissions and workflow integrations.

That approach also reflects Salesforce’s broader view that large language models are becoming interchangeable infrastructure. For customers, model choice could eventually affect cost, latency, quality and compliance without requiring a complete redesign of the Slack experience.

Salesforce is also describing Slackbot as a future “super agent” that coordinates with other agents and tools. Slack already hosts or supports agents from companies including Anthropic, OpenAI, Google and Vercel. Harris said Slack could become a Model Context Protocol client, allowing Slackbot to use tools across a wider software ecosystem.

The company is tempering that vision. Harris acknowledged that enterprise systems are still primarily operating in a single-agent phase and said Salesforce would focus on useful coordination rather than promoting scenarios involving hundreds or thousands of agents working together.

Why the launch matters to enterprise AI buyers

Slackbot’s competitive advantage, according to Slack product executives, is proximity to work already happening in Slack. Users do not need to open a separate application or configure a new knowledge base before asking questions about conversations, documents or decisions already visible to them.

That convenience is also the product’s central dependency. Slackbot’s answers and actions will only be as reliable as the information available through approved connectors and the permissions attached to it. Enterprises evaluating the tool will need to test whether access controls survive retrieval and document generation, whether citations or source trails are sufficient for review, and how the system handles contradictory or outdated information.

Pricing may be straightforward at the product level: Salesforce says Slackbot is included with Business+ and Enterprise+ plans at no additional charge. The broader cost picture could be less simple. VentureBeat connected the rollout to concerns about Salesforce’s pricing for API access, which could affect companies that move Salesforce data into systems such as Snowflake or connect it to external AI tools.

Those concerns are based partly on comments from Fivetran CEO George Fraser and represent a market risk rather than a demonstrated Slackbot charge. Still, they matter to CIOs because an assistant embedded in Slack may reduce application switching while increasing dependence on Salesforce-controlled data pathways.

For builders and product teams, the launch highlights a practical design pattern: place AI where conversations, permissions and work artifacts already coexist, then add tool calls incrementally. For Salesforce, the challenge will be proving that this embedded experience is reliable enough for decisions and actions, not merely convenient for summaries.

What to watch next

The first signal will be rollout completion for eligible customers and the arrival of mobile support, which Salesforce said would follow the initial release. Buyers should also watch for the promised meeting-booking capability, image-generation support and additional model providers such as Gemini.

More important will be evidence on repeat usage outside Salesforce. Independent customer studies, clearer measurement of time savings and reporting on error rates would help distinguish durable workflow value from early-adopter enthusiasm.

The development of third-party tool calls will show whether Slackbot can become a useful coordination layer without creating a difficult-to-audit chain of automated actions. Integrations with competing CRM systems, including HubSpot and Microsoft Dynamics, are another unresolved competitive question.

Finally, enterprise customers will need to track changes in API access, data-sharing policies and governance controls. Slackbot’s appeal depends on broad context, but its acceptability will depend on whether that context remains permission-aware, auditable and cost-predictable.

Creati.ai perspective

Salesforce is not winning this contest by offering another general chatbot. It is betting that Slack’s accumulated conversations, documents and work relationships can provide the context that standalone AI tools lack. That is a credible product strategy, but the reported adoption figures are still Salesforce’s own evidence, and the most ambitious agent-coordination features remain on the roadmap.

The decisive test will be whether Slackbot can safely complete small, verifiable tasks across enterprise systems while preserving user permissions and giving teams enough visibility into how results were produced. If it can, Slack becomes more than a communication hub; if it cannot, Microsoft Copilot and Google Gemini will retain an important advantage through their broader suite-level integrations.