Dreamforce 2025 Day One: Welcome to the Agentic Enterprise

In this blog post, we’ll bring you all the biggest announcements, key moments, and behind-the-scenes insights from Day One of Dreamforce 2025 — touching on what Salesforce’s new vision for the Agentic Enterprise really means for DevOps and the wider Salesforce ecosystem.

Salesforce opened Dreamforce 2025 in San Francisco with a clear message: the Agentic Enterprise is now central to its strategy. The first day of the conference focused on how autonomous agents, data context, and data governance are converging to shape the next phase of enterprise AI.

Marc Benioff’s keynote emphasized that Salesforce’s goal is no longer to create assistive copilots but to build autonomous systems that act with context and control.

Dreamforce main keynote overview

In the Dreamforce Main Keynote, Marc Benioff described the company’s direction as “AI that, when built with trust, elevates people.” This statement framed Salesforce’s new positioning: AI agents that operate independently but within defined guardrails.

The company introduced Agentforce 360, the latest evolution of its agent platform, which integrates agent-building tools, governance frameworks, and observability capabilities. It forms part of Salesforce’s broader shift from CRM to what Benioff called an enterprise orchestration platform.

Day One announcements also highlighted deeper integrations between Agentforce, Data 360 (the renamed Data Cloud), and Slack, showing how the ecosystem is being built to support end-to-end agentic workflows.

Agentforce 360

Agentforce 360 is designed to unify the tools needed to build, deploy, and manage autonomous agents across Salesforce’s clouds. The platform includes:

• Agentforce Builder, which converts natural language into deployable applications using Agent Script.
• Agentforce Voice, providing conversational interaction with agents.
• Agentforce Vibes, which translates context into executable code.
• Command Center and Testing Center, new features since Dreamforce 2024 for governance and observability. These components will allow users to design agents that perform business processes automatically, with built-in oversight and auditability. Salesforce said this represents the fourth major release of its agent platform in a year.

Demonstrations featured customers including Williams-Sonoma, Pandora, FedEx, and Dell.

• Williams-Sonoma’s “Olive” agent recreated personalized in-store experiences for online shoppers.
• Pandora’s “Gemma” agent provided an interactive shopping experience entirely within an agentic chat.
• FedEx used Data 360 Governance to create and enforce compliance policies using natural language.
• Dell demonstrated how Agentforce Supply Chain can analyze documentation to optimize supplier onboarding.

Data 360 and context engineering

Salesforce introduced Data 360 as the next generation of its Data Cloud, combining structured and unstructured data sources to give agents contextual awareness. According to Salesforce, this will enable agents to interpret documents, extract relevant details, and perform reasoning tasks directly within enterprise environments.

This shift aligns with what Salesforce executives referred to as context engineering — developing systems that are programmed through data rather than code. The platform aims to allow enterprises to harmonize disparate data and make it available for automation and analytics.

Governance, testing, and observability

Governance was a recurring theme throughout the keynote. Salesforce detailed new capabilities that provide visibility into how agents make decisions. The Testing Center now records each step of agent reasoning and execution, while the Command Center gives IT teams control over policy, permissions, and audit logs.

These features are designed to make agent behavior transparent, addressing common enterprise concerns about accountability and security.

Slack as the Agentic OS

Salesforce positioned Slack as the “Agentic Operating System” — the interface layer where employees and agents collaborate. New integrations will allow agents to execute tasks and deliver updates directly within Slack channels.

Analysts from theCUBE described this as a step toward conversational orchestration, where users can trigger or monitor workflows without leaving the chat interface. The approach also provides a mechanism for human oversight within automated systems.

Adoption and industry context

Benioff acknowledged that customer adoption is still catching up with the pace of innovation, stating that “the speed of innovation is far exceeding the speed of customer adoption.”

This reflects a broader industry trend: enterprises are experimenting with AI agents but remain cautious about scaling them. Analysts described this as the “execution phase” of enterprise AI — where focus is shifting from prototypes to production-ready systems.

Implications for DevOps teams

For DevOps professionals working within the Salesforce ecosystem, the announcements point to a growing need for DevOps for Agentforce — applying established delivery, testing, and observability practices to intelligent systems. This includes:

• Treating agents as deployable software components with version control and rollback.
• Implementing continuous testing to validate behavior and prevent drift.
• Extending observability tooling to monitor agent reasoning and performance.
• Embedding governance policies directly into CI/CD pipelines.

These principles will be critical as enterprises begin integrating autonomous agents into production environments.

Summary

The first day of Dreamforce showcased Salesforce’s vision for the agentic future — no surprises there. But discussions also acknowledged challenges around data quality, compliance, and change management.

Salesforce’s trajectory — from CRM software to an enterprise orchestration platform powered by AI agents — was the defining narrative of the day.

Stay tuned for our coverage of highlights and announcements on Day Two of Dreamforce!

Keep learning with DevOps Launchpad

Dreamforce made one thing clear: the future of Salesforce DevOps is intelligent, observable, and automated. Whether you’re looking to sharpen your DevOps fundamentals or explore the new world of Agentforce, DevOps Launchpad has free, expert-led courses to help you level up. Check out our full list of courses and certifications here.