Salesforce has rolled out a rebuilt Slackbot that turns Slack’s longtime assistant from a lightweight notification feature into a broader AI agent for enterprise search, drafting and task execution. According to VentureBeat, the new product is now generally available to Slack customers on Business+ and Enterprise+ plans, with a wider rollout continuing through the end of February and mobile support expected by early March.

The launch matters because Salesforce is trying to reposition Slack as more than a messaging app at a moment when Microsoft Copilot and Google Gemini are increasingly embedded into the software workers already use. By moving the AI layer directly into chat, Salesforce is making a clear argument: the winning workplace assistant may be the one already sitting in the flow of internal conversations, permissions and shared documents, not a separate destination app.

The company’s executives framed the release as a foundational step toward what Salesforce calls an “agentic” enterprise. That language is now common across the industry, but in this case the near-term product appears more concrete than speculative. VentureBeat reported demonstrations showing Slackbot pulling from Slack conversations, files, calendar data, Google Drive content and Salesforce records, then turning that context into summaries, plans and shareable Canvas documents.

What changed in the new Slackbot

The most important part of this launch is architectural, not branding. VentureBeat reported that Salesforce rebuilt Slackbot around a large language model, search infrastructure and connections to third-party data sources, replacing the older rules-based assistant that mainly handled reminders and simple prompts.

That means the new Slackbot is designed to answer questions across enterprise systems, synthesize information and trigger actions without forcing users to leave Slack. In the product examples described to VentureBeat, users could ask it to analyze customer feedback, compare that with a dashboard image, identify relevant accounts in Salesforce, generate a plan in Canvas, and check stakeholder calendar availability. Some features are available now, while meeting booking is still due in the coming weeks, according to Slack executives cited by VentureBeat.

This is a meaningful shift in product scope. Many workplace AI tools still stop at retrieval and summarization. Salesforce is trying to move Slackbot one step further into workflow execution, even if the tool-calling surface is still early. VentureBeat also reported that Slack leadership described Canvas generation as a sign of the larger roadmap, with additional third-party tool calls planned later.

That roadmap matters because the product is being positioned not simply as a chatbot, but as a central interface for other software and, eventually, other agents. Salesforce executives told VentureBeat they see Slackbot becoming a kind of “super agent” inside Slack, coordinating work across internal and external tools.

Why Salesforce chose Claude first

The new Slackbot currently runs on Claude, the model from Anthropic. According to comments cited by VentureBeat from Salesforce co-founder and Slack CTO Parker Harris, one reason was compliance: at the time Slack started building the product, Anthropic was the only provider that could satisfy the FedRAMP Moderate requirements Slack needed for some government-related use cases.

That does not appear to be a permanent single-model strategy. Harris told VentureBeat that Salesforce plans to support additional providers this year and specifically mentioned Google Gemini as a strong candidate for some workloads. He also said OpenAI remains a possibility.

For buyers and builders, that is one of the more important signals in the launch. Salesforce is effectively treating foundation models as interchangeable infrastructure rather than the product itself. That model-agnostic stance aligns with broader enterprise demand for optionality on cost, latency, capability and compliance. It also reduces the risk that Slackbot becomes too dependent on a single vendor’s pricing or policy changes.

Salesforce also used the launch to restate a familiar enterprise AI assurance: customer data is not used to train foundation models. That claim, as reported by VentureBeat, came directly from Harris. The rationale he gave was straightforward: once confidential information is absorbed into a model, access control becomes difficult to guarantee. For enterprise customers, especially regulated ones, that distinction remains central to procurement.

The competitive target is Microsoft and Google

Salesforce is not entering an empty market. The release places Slackbot directly against Microsoft Copilot in Teams and Google Gemini in Workspace. All three vendors are pursuing the same basic thesis: enterprise AI works best when it is embedded into the productivity environment employees already inhabit.

Salesforce’s differentiation claim, according to VentureBeat, is context and proximity. Slackbot lives inside the collaboration graph that already contains channels, direct messages, decisions, shared files and organizational habits. That could be a real advantage if the assistant can use those signals without introducing security problems or drowning users in noisy outputs.

At the same time, the competitive gaps are still visible. VentureBeat reported that Salesforce declined to provide specifics when asked about support for competing CRM systems such as HubSpot or Microsoft Dynamics. That omission matters. If Slackbot becomes most powerful when paired with Salesforce data, then the product may be strongest in accounts already committed to the Salesforce stack, rather than as a neutral workplace layer across mixed environments.

There is also a cost narrative behind the “included at no additional charge” message. Slack executives told VentureBeat that Slackbot comes bundled for Business+ and Enterprise+ customers. But the same report notes wider concerns around Salesforce data access costs, including external criticism that API pricing changes could make it harder or more expensive for enterprises to use outside data tools instead of Salesforce’s own stack. For CIOs, those adjacent economics matter almost as much as the sticker price of the assistant itself.

Evidence, adoption claims and what is still unproven

Most of the strongest early usage signals in this story are vendor-reported and should be read that way. VentureBeat reported Salesforce’s internal testing across 80,000 employees, with the company saying two-thirds of staff had tried the new Slackbot, that 80% of those users stayed active, and that satisfaction reached 96%. Salesforce also said employees reported saving between two and 20 hours per week.

Those numbers suggest strong internal enthusiasm, but they are not independently verified benchmarks. Internal dogfooding can be useful evidence of product readiness, especially for collaboration software, yet it does not always predict customer outcomes across different industries, governance models or knowledge environments.

The same caution applies to customer testimonials. VentureBeat cited pilot users including Beast Industries, Slalom, reMarkable, Xero, Mercari and Engine. Those customers reported time savings ranging from about 30 minutes a day to 90 minutes a day, and one executive said security approval was unusually fast because Slackbot respects existing user-level permissions inside Slack. Those are relevant deployment signals, but they remain anecdotal.

What is confirmed is narrower: Slackbot is available on qualifying paid plans; it uses Claude today; it can search across several enterprise sources; and Salesforce plans additional model providers and more actions over time. What remains unproven is whether employees will sustain usage once the novelty fades, whether answer quality holds up under complex enterprise retrieval, and whether tool-calling can be reliable enough for higher-stakes workflows.

What this means for builders and enterprise buyers

For AI product teams, the launch reinforces a market lesson that has become hard to ignore: distribution and context can matter more than raw model quality. Slack is already where many employees ask questions, escalate issues and make decisions. If Slackbot can translate those moments into retrieval, drafting and lightweight execution, it may gain adoption faster than standalone workplace agents.

For enterprise buyers, the practical questions are less about “agentic AI” branding and more about controls. Can Slackbot stay grounded in permissions? Can it cite enough source context to be trusted? How much administrative setup is really required? And does bundling the assistant into Slack reduce deployment friction enough to overcome caution around enterprise AI?

There is also an ecosystem implication. VentureBeat noted that companies including Anthropic, OpenAI and Vercel have already been building agents for Slack, and that Salesforce sees many new Slack apps taking agent form. If Slackbot becomes the front end for those tools rather than a competitor to all of them, Slack could evolve into a control surface for mixed-agent workflows. But that depends on interoperable standards and consistent user experience, not just more chat commands.

For founders building workplace AI, the warning is clear too. Competing against embedded assistants inside Slack, Microsoft Copilot and Google Gemini will be difficult if the startup product is “ask questions about work” alone. The openings are more likely in domain-specific actions, stronger retrieval quality, governance, or systems that sit behind those assistants rather than try to replace them.

What to watch next

The next signals to watch are product depth and model breadth. Salesforce has said meeting booking should arrive shortly after launch, and support for additional model providers is due this year. If Google Gemini or OpenAI models appear inside Slackbot, that will test how serious Salesforce is about a multi-model architecture.

Enterprise buyers should also watch for clearer statements on external system support, especially beyond Salesforce. Integrations with non-Salesforce records, more explicit admin controls, and stronger auditability will determine whether Slackbot can become a general workplace assistant rather than mainly a Salesforce-centric one.

Finally, usage quality will matter more than launch excitement. Independent customer evidence on time savings, hallucination rates, permission boundaries and workflow completion will be more valuable than internal adoption anecdotes. In workplace AI, novelty spreads quickly inside chat; durable trust takes longer.

Creati.ai perspective

Salesforce’s rebuilt Slackbot is less interesting as a chatbot upgrade than as a distribution move. The company is using Slack to make AI feel native to work rather than attached to it. That is the same strategic ground being contested by Microsoft Copilot and Google Gemini, but Slack has one genuine asset in the fight: rich conversational context tied to real operational work.

The harder part will be proving that context turns into dependable action. Enterprise users do not need another assistant that summarizes meetings and drafts vague copy. They need systems that can retrieve the right data, preserve permissions, and finish tasks without creating more review overhead than they remove. If Salesforce can make Slackbot reliable on those basics, it has a stronger competitive story than the “agentic” label suggests.