One of the persistent frustrations of enterprise software is that your tools don’t talk to each other. You pull a CRM record, switch to a doc, reference a meeting transcript in another tab, and then copy the summary into a chat thread. The context lives in your head because the systems can’t hold it together.

On April 22, at Google Cloud Next in Las Vegas, Salesforce and Google Cloud announced an expanded partnership designed to change that. The new integrations allow AI agents from both platforms to execute end-to-end workflows across Agentforce, Gemini Enterprise, Slack, and Google Workspace without moving data between systems or forcing employees to leave the tools they already use.

What Was Actually Announced

The partnership ships in stages, with several features available now and more rolling out through the rest of 2026.

Agentforce Sales in Gemini Enterprise is available in Open Beta now on the Gemini Enterprise Marketplace. Sales agents built in Agentforce can engage leads, create meeting briefs, surface deal risks, and update CRM records while a rep is working inside Gemini Enterprise. The agent does the CRM work in the background while the rep stays in their workspace.

Gemini Enterprise in Slack is live in Private Preview on the Slack Marketplace. Gemini can now search across Slack threads, Google Meet transcripts, and connected apps simultaneously, pulling together context that previously required jumping between tools.

Gemini-Powered Reasoning for Agentforce lands in May 2026. Agentforce’s Atlas Reasoning Engine will natively support Gemini models, enabling agents to process text, images, and video together rather than treating them as separate inputs. The intent is agents that can reference a video recording from last quarter’s all-hands alongside written deal history to give a sales rep an accurate summary.

Zero Copy with Google Lakehouse is coming in late 2026. Agentforce will be able to read data from Google Lakehouse directly, without copying or moving it. For businesses with strict data governance requirements, this is meaningful: the agent operates on live data without creating a copy that sits outside your security perimeter.

Why the Fragmented Context Problem Matters

The feature list is less interesting than the problem it’s solving.

Enterprise AI agents have been impressive in demos and limited in production for one main reason: they don’t have enough context about what’s actually happening in the business. A customer service agent that can only see the ticketing system misses the email thread from two weeks ago. A sales agent that only sees Salesforce data misses the Google Doc where the account team recorded their notes from the last call.

The Salesforce-Google Cloud integration is a direct attempt to give agents a broader view without requiring businesses to consolidate everything into a single platform. Instead of moving all your data to one system, the agents work across systems and hold the context themselves.

This is meaningfully different from the point integrations that have existed for years. Previous integrations let humans copy data between platforms. These integrations let agents reason across platforms and act without a human in the loop.

What This Means for Business

For business leaders deciding how to build their AI stack, this announcement reinforces a pattern worth paying attention to: the major enterprise platforms are betting that agents will run across their whole technology stack, not just within a single product.

That has practical implications for how you plan. If your CRM, communication tools, and productivity suite are all building toward cross-platform agent coordination, the value of investing in clean data and structured workflows increases significantly. Agents are only as useful as the data they can reach and reason over.

The companies showing up in Google Cloud’s customer showcases this week, including Verizon, Toyota, and Walmart, are not running pilots. They are using agents in production across their core business workflows. The gap between those companies and ones still evaluating whether to start is widening.

For businesses that want to start building with AI agents now, rather than wait for the 2027 enterprise suite roadmaps, the opportunity is in deploying purpose-built agents against specific workflows. Voice AI for customer intake, agents for internal knowledge retrieval, automated report generation. These deliver measurable results faster than waiting for the full platform picture to settle.

If you want the playbook other teams are using with Claude and Codex right now, grab the free Working With Claude field guide. Download it here.