Dreamforce 2025 Recap: The Rise of Agentforce 360 and Agentic AI in the Enterprise

Dreamforce 2025 felt like a line in the sand. Across Moscone, leaders debated one big shift: how to move from scattered AI pilots to an operating model where human expertise works in concert with autonomous AI agents. Salesforce framed this as the Agentic Enterprise. The message from the main stage was clear. Get your data right, get integrated, set priorities, and put governance at the core.

Below are the highlights Jade Global believes matter most for CIOs, digital leaders, and line of business owners planning the next two to four quarters.

1. From experiments to the Agentic Enterprise

Attendees heard a consistent theme. Do not chase every AI use case at once. Start small, move fast, measure, then scale. Success depends on delivering the right information to the right agent — human or digital — at the right time. Data quality, thoughtful design, and real-world context drive adoption, not just model choice.

A practical framework repeated in multiple sessions:

• Start small and go fast.
• Test, tune, repeat.
• The right data is gold.
• Measure value and impact.
• Embrace new ways of working.

2. A new architecture for AI agents

Salesforce advocates an expanded reference architecture for agentic work. In addition to traditional layers, leaders should plan for four capabilities that enable agents to sense, reason, and act with increasing autonomy.

• Semantic layer for shared meaning and context.
• AI and ML layer as the intelligence hub for internal and external models.
• Agentic layer to build and run agents at scale.
• Enterprise orchestration layer as the control plane for workflows across agents, humans, and systems.

Adopt this in phases. Move from information retrieval agents to domain-specific action agents, to cross-domain orchestration, and ultimately to multi-agent, multi-domain operations.

3. Agentforce 360 becomes the core platform

Salesforce introduced Agentforce 360 as the platform that powers everything the company does. It brings together four ingredients of the Agentic Enterprise:

• Agentforce 360 with a conversational builder, hybrid reasoning, and voice.
• Data 360: The trusted, unified data context layer.
• Customer 360 Apps, where business logic lives for sales, service, marketing, and operations.
• Slack is the conversational interface where humans and agents collaborate.

Important note on naming. Data 360 is the new name for Salesforce Data Cloud. If you documented architectures or roadmaps this year, update references from “Data Cloud” to Data 360 to reflect the current branding and feature set.

Slack’s role expanded significantly. Think of it as an agentic operating system that converts knowledge, data, and actions into a natural language surface, enabling people and agents to work together.

4. Proof points leaders will care about

Dreamforce brought real outcomes to the stage.

• PepsiCo reported efficiency gains of 25-30 percent. Agentforce is already live in more than 1.5 million stores with a target of 5 million by the end of 2026.
• Williams Sonoma demonstrated voice and shopping agents that personalize experiences by connecting to real-time product and customer data.
• FedEx highlighted how unified data and agentic workflows boost conversion and productivity in complex logistics environments.
• Financial services and public sector examples showed how agentic assistants reduce preparation time for advisors and scale citizen services without compromising trust and compliance.

A useful reminder from Andrew Ng resonated with many. Perfect data is not a prerequisite to start. Establish governance and security, then let real applications pull your data architecture forward.

5. What this means for CIOs and enterprise architects

Three practical implications stood out.

• Data 360 first. Treat unified, governed data as the center of the plate. Standardize identity resolution, semantics, and lineage. Your agents will only be as good as their context.
• Design for human plus agent work. Do not aim to replace teams. Aim to elevate them. Define clear handoffs, observability, and controls so adoption sticks.
• Phased orchestration. Sequence value so the business sees results quickly. Start with narrow retrieval or triage, then layer in action, then orchestrate across domains.

6. Revolutionizing Salesforce Development: Agentforce Vibes and Builder

Dreamforce 2025 marked a major leap in Salesforce development with the launch of Agentforce Vibes and Agentforce Builder—AI-powered tools transforming how enterprises build, test, and deploy applications.

Agentforce Vibes

• Introduces “vibe coding” with real-time, context-aware code suggestions in VS Code.
• Automates repetitive coding tasks and detects bugs early.
• Strengthens code security through the Einstein Trust Layer.
• Shifts developers’ focus from manual syntax to intelligent architecture and logic design.
• Boosts productivity, innovation, and code quality.

Agentforce Builder

• Enables the creation of intelligent AI agents through simple conversational design.
• Combines low-code development with powerful AI automation.
• Reduces development friction and shortens delivery cycles.
• Democratizes app building across technical and non-technical roles.

Together

• Usher in an agentic development era where AI amplifies human creativity and collaboration.
• Accelerate delivery of smarter, more secure, and agile enterprise applications.

7. Jade Global’s take: how to move from digital to truly agentic

Enterprises do not operate in a single stack. Most organizations run Salesforce alongside Oracle, NetSuite, ServiceNow, Workday, Snowflake, Boomi, and other systems. The Agentic Enterprise requires these worlds to speak the same language, share trusted data, and coordinate actions.

Jade Global helps customers do exactly that.

• AI-ready data foundations. We align Salesforce Data 360 with your broader data estate, including Snowflake or Databricks, to deliver reliable context without copy sprawl.
• Agentic workflows across platforms. Using our accelerators and enterprise-ready agents, we connect Salesforce with ERP, supply chain, finance, and IT service operations, enabling agents to act where work happens.
• Guardrails and governance. We bake security, observability, and change control into our processes, enabling CIOs and risk teams to scale safely and securely.
• Value first sequencing. We start with a focused use case, quantify impact, and build the orchestration roadmap you can defend to the board.

8. Key takeaways you can act on this quarter

• Update your plans to reflect Data 360 as the renamed and expanded Data Cloud. Align semantics, identity, and governance with enterprise standards.
• Pick one high-friction process where retrieval or triage agents could remove manual steps in weeks, not months.
• Stand up an agent observability and evaluation approach on day one. Measure accuracy, latency, containment, and business impact.
• Treat Slack as a first-class platform for human and agent collaboration. Pilot a workflow where agents brief, suggest next best actions, and execute with approvals.
• Draft a two-quarter roadmap that moves from retrieval to action in one domain, and defines the data contracts needed for cross-domain orchestration next.

Closing thought:

Dreamforce 2025 made one thing unmistakable. The Agentic Enterprise is not a hype cycle. It is a new way to operate, where trusted data, integrated platforms, and well-designed agents give every team always-on intelligence. Companies that start small, measure value, and scale with governance will move faster than those waiting for perfect conditions.