Salesforce and Google Cloud expand partnership for cross-platform AI agents
The expanded Salesforce and Google Cloud partnership connects Agentforce and Gemini so agents can act across Slack, Workspace, and CRM with less data movement.
Salesforce and Google Cloud have expanded their partnership to let AI agents execute end-to-end workflows across both platforms with deeper context and less data fragmentation. The integrations span Slack, Google Workspace, Agentforce, and Gemini Enterprise, and include capabilities such as Gemini access inside Slack, Agentforce Sales agents inside Gemini Enterprise, and new data connectivity approaches including “zero-copy” access via Google Lakehouse.
This is a market-shaping move for enterprise operations and revenue teams because it targets a core blocker to agent adoption: agents are only as useful as their ability to act across the systems where work and data actually live, without creating new governance and security headaches.
Short on time?
Here’s a quick look at what’s inside:
- What the expanded integrations enable
- Why cross-system context is the hard part of agentic AI
- How this fits the enterprise CRM and cloud competition
- What marketing and revenue teams should do next
ContentGrip
What the expanded integrations enable
The partnership adds several workflow-level integrations designed to reduce context switching and let agents move work across tools:
- Slack and Google Workspace: Users can request outputs (like structured documents or slides) from Slack, with Slackbot pulling relevant context from Slack threads and Google Workspace files.
- Gemini Enterprise in Slack: Gemini becomes accessible in Slack as an assistant that can pull from connectors across apps, such as summarizing a Google Meet transcript alongside a Slack thread.
- Agentforce Sales in Gemini Enterprise: Sales agents can engage leads, create meeting briefs, surface deal risks, and update CRM records from within Gemini Enterprise, aiming to keep work inside the interface where the user is operating.
- Gemini-powered reasoning for Agentforce: Agentforce can natively use Gemini models via the Atlas Reasoning Engine, including multimodal understanding across text, image, and video.
- Zero-copy data access: Agentforce is expected to read data from Google Lakehouse without copying it, reducing data movement and potentially simplifying security posture.
Availability varies by feature, with some capabilities already available, others scheduled across April through late 2026.
Why cross-system context is the hard part of agentic AI
Most “agentic” roadmaps run into the same operational reality: enterprises have multiple systems of record and multiple places where decisions happen. If an agent can draft content but cannot retrieve the right context, apply policy, and complete actions inside the systems that govern work, it becomes a novelty rather than a reliable operational layer.
This partnership is explicitly framed around solving fragmented data and disconnected systems, with a strong emphasis on reducing risky data movement and enabling agents to operate where work happens (Slack and Google Workspace) while still being able to act on CRM and data warehouse context. The inclusion of governance and audit-ready elements, plus the “zero-copy” design, signals that adoption is being approached as a security and data architecture problem as much as an AI interface problem.
Salesforce’s financial scale also matters here: with reported revenue of US$41.525 billion for fiscal year 2026, it has strong incentives to make Agentforce sticky across the customer base by embedding it into daily workflows and aligning with a major cloud and model provider rather than treating AI as a closed ecosystem.
How this fits the enterprise CRM and cloud competition
Salesforce competes in enterprise CRM against Microsoft Dynamics 365, Oracle CX, SAP Customer Experience, and HubSpot, where the battleground is shifting from “best CRM features” to “best workflow and data fabric for AI-assisted work.” A CRM that is isolated from productivity suites and analytics infrastructure becomes a weaker foundation for agent-driven execution.
On the cloud side, the partnership also reinforces a broader pattern: model capability alone is not enough; distribution and embedded workflow surfaces matter. By placing Gemini Enterprise in Slack and Agentforce inside Gemini Enterprise, both companies are effectively competing for where “the agent layer” lives in the enterprise, and they are using integration depth as the differentiator.
The category is highly competitive, and enterprises are likely to favor approaches that reduce integration overhead, limit duplicated data pipelines, and provide consistent governance across channels and departments.
What marketing and revenue teams should do next
For marketing ops, RevOps, and sales leadership, the practical opportunity is not “more AI,” but fewer broken handoffs between systems. A few next steps to make this concrete:
- Map priority workflows end to end: Start with processes that already span Slack, documents, and CRM updates (lead follow-up, campaign-to-pipeline reporting, customer issue escalation).
- Define data access boundaries: If “zero-copy” and connector-based access are part of the design, clarify what data is permitted for which agent roles, and who owns approvals.
- Pilot with audit requirements in mind: Choose workflows where you can measure productivity impact and maintain an audit trail (especially if automating CRM updates or customer-facing messages).
- Prepare governance early: Compliance, retention, and permissioning often slow agent deployments more than model readiness. Treat them as first-class requirements, not phase-two add-ons.
