Salesforce publicly promoted a product that competes with its own AI stack. That is not a strategy contradiction. It is a financial position made visible.
When Anthropic launched Claude Tag, a persistent AI agent invocable via @Claude inside Slack channels, Salesforce helped amplify the announcement on social media. The Information reported that this confused employees at Salesforce, which owns Slack: Claude Tag runs parallel to Slackbot and to Agentforce, Salesforce’s own agentic AI platform. Slack customers now face a three-way choice between those products inside a single workspace. Salesforce Ben, an independent Salesforce publication, described this as potentially “too much choice” for enterprise buyers trying to standardize on one AI interface.
The confusion inside Salesforce is understandable. The company spent $27.7 billion acquiring Slack in 2021 and has since built Slackbot into a capable agentic system. In March, Salesforce added more than 30 new AI capabilities to Slackbot, covering meeting transcription, desktop activity monitoring, task execution through third-party tools, and lightweight CRM functions. Agentforce, Salesforce’s autonomous agent platform, reached $800 million in annual recurring revenue as of the most recent earnings call, up 169 percent year-on-year with 29,000 deals closed. Both products now share workspace real estate with a third AI agent that Salesforce had no hand in building.
The financial exposure explains the behavior. Salesforce expects to spend roughly $300 million on Anthropic tokens this year and holds approximately a 1 percent stake in Anthropic, now valued at $380 billion. That means Salesforce is both a major revenue source for Anthropic and a direct equity holder. Promoting Claude Tag is, in part, protecting that position. CEO Marc Benioff has characterized Slack as “the interface to AI” and framed the platform as model-agnostic, a positioning that lets Salesforce collect rent from the AI ecosystem rather than bet entirely on its own models winning.
The practical problem with model-agnostic positioning is that it requires the platform to stay neutral between products that compete for the same user behavior. Claude Tag does what Agentforce does: it accepts tasks, breaks them into steps, executes them inside the workspace, and builds up institutional knowledge over time. Anthropic is the first LLM provider that is fully contained within the Salesforce trust boundary, meaning data stays inside the ecosystem and is not used for model training. That is a meaningful technical differentiator and a reason for deep integration. It is also a reason why the line between partner and competitor has become impossible to draw cleanly from the inside.
Anthropic plans to expand Claude Tag to Microsoft Teams, email, and other productivity platforms in coming weeks. When that happens, Claude Tag becomes a cross-platform agent that does not need Slack to reach users. For Salesforce, that creates a second-order problem: the partner it is funding at $300 million per year is building the infrastructure to operate independently of the platform Salesforce owns.
Teams evaluating enterprise AI consolidation should note that Salesforce’s own product strategy is now internally contested. Agentforce’s revenue trajectory is real, but so is the structural ambiguity Claude Tag introduces. Any buyer negotiating a Salesforce contract that bundles Agentforce should ask explicitly how feature overlap with Claude Tag will be resolved, and whether Salesforce’s commitment to Agentforce development changes if Claude Tag captures the same workflow share.
Reported by The Next Web on June 28, 2026, based on reporting originally from The Information.
