Salesforce Dreamforce 2026 Launches AIforce: 7 Billion Agentic Work Units, Long-Horizon Runtime, 7 Job-Ready Agents — Enterprise AI Agent Ecosystem Fully Upgraded

Introduction

At Dreamforce 2026, held September 15–16, Salesforce unveiled a major upgrade to its enterprise AI agent strategy. The centerpiece announcements included the new AIforce interface layer, a significant expansion of the Agentforce platform, and a suite of "job-ready" specialized agents. At the time of the announcement, the Agentforce platform had already served more than 30,000 customers and processed 7 billion Agentic Work Units (AWUs) — with 3.2 billion of those AWUs completed in Q2 2026 alone — signaling that enterprise AI agent adoption is accelerating rapidly.

AIforce: A New Interface Layer for Enterprise AI Agents

AIforce is the core architectural innovation Salesforce introduced at Dreamforce 2026, positioned as an "interface-independent agentic computing" layer. Its design philosophy enables enterprise employees and external AI agents to interact with Salesforce's business logic, data, and workflows through natural language — without needing to log directly into the Salesforce application.

Three Core Components of AIforce

1. Claudeforce

Claudeforce is the result of a deep collaboration between Salesforce and Anthropic. Through a prebuilt Model Context Protocol (MCP) server, it integrates Salesforce's sales, service, and development capabilities directly into the Claude AI assistant. Enterprise users can query CRM data, update customer records, and trigger business processes directly within the Claude interface — no application switching required. For Asia-Pacific enterprises already using Anthropic Claude, this means accessing Salesforce's full business capabilities within a familiar AI assistant interface.

2. Slackforce

Slackforce transforms the Slack workspace into an AI-driven productivity hub, featuring three sub-capabilities:

• Slackforce Surfaces: Real-time CRM data presented directly within Slack, allowing sales and service teams to view customer information without leaving Slack
• Slackbot: Intelligent business assistant functionality capable of answering business questions and executing CRM operations
• Slack Code: Collaborative AI development support, allowing developers to build and test Salesforce integrations directly within Slack

3. Agentforce Coworker

An AI teammate embedded within the standard Salesforce Lightning interface, capable of assisting users in real time within the existing permissions and business rules framework. Unlike external integrations, Agentforce Coworker operates entirely within the Salesforce ecosystem, ensuring the highest level of data security.

Zero Data Retention Commitment

A key security feature of AIforce is its Zero Data Retention policy: business data is used only to fulfill specific requests and is not retained by third-party model providers or used for training. All AI-driven actions execute within Salesforce's existing trust boundary, ensuring compliance with enterprise security and regulatory requirements. This commitment is particularly important for Asia-Pacific enterprises operating under strict data protection regulations (such as those subject to PDPA, APPI, or the Privacy Act).

Major Expansion of the Agentforce Platform

The 7 Billion AWU Milestone

Salesforce defines an Agentic Work Unit (AWU) as a single, discrete task successfully completed by an AI agent. The 7 billion AWU total reflects the platform's scale and the fundamental shift from simple Q&A to agents autonomously handling business processes. Notably, 3.2 billion AWUs were completed in Q2 2026 alone, indicating accelerating rather than linear adoption growth.

Long-Horizon Runtime

The most significant technical upgrade to Agentforce is the introduction of the Long-Horizon Runtime, which enables agents to maintain context and pursue goals over days or even weeks — rather than being limited to single-turn interactions. This runtime integrates three key capabilities:

• Persistent Memory: Retaining business context across sessions, allowing agents to remember previous conversations and decisions
• Durable Execution: Maintaining state consistency across long-running tasks, even when tasks span multiple workdays
• Dynamic Steering: Dynamically adjusting agent behavior as business objectives evolve, rather than mechanically executing predefined steps

The introduction of the Long-Horizon Runtime enables Agentforce to handle truly complex business processes, such as: multi-day procurement approval workflows, customer service escalations requiring multiple rounds of communication, and compliance reporting tasks requiring continuous monitoring and adjustment.

7 Job-Ready Specialized Agents

Salesforce launched a series of "job-ready" agents designed for specific business functions — these agents work out of the box and can be deployed without extensive customization:

Supporting Technologies: Koa, Guardian, and MuleSoft Agent Fabric

Koa: A CRM reasoning model built on NVIDIA Nemotron, trained exclusively on synthetic data to handle complex, multi-task business outcomes. Koa's synthetic data training approach ensures the model doesn't leak customer data while maintaining high-quality business reasoning capabilities.

Salesforce Guardian: A security suite for managing agent identity, detecting rogue agents, and providing data lifecycle protection. In multi-agent environments, Guardian ensures each agent can only access the data and functions it is authorized for.

MuleSoft Agent Fabric: An agent discovery and governance tool spanning AWS, Google, Microsoft, and other providers — addressing interoperability challenges in multi-agent enterprise environments. For Asia-Pacific enterprises already using multiple cloud platforms, Agent Fabric provides a unified agent management view.

Real-World Deployment Cases

Real deployment cases shared at Dreamforce 2026 demonstrate that enterprise AI agents have moved from proof-of-concept to scaled production:

Live Nation: Used Agentforce to handle tens of thousands of fan inquiries with high resolution rates, significantly reducing the workload on human customer service agents.

Siemens: Deployed Agentforce in industrial sales and service scenarios, redefining industrial AI workflows — particularly in complex B2B sales processes.

Policybazaar: AI agents now manage 60% of customer interactions — a figure with important reference value for the insurance industry.

Asia-Pacific Perspective

For enterprises in the Asia-Pacific region, the AIforce and Agentforce upgrades carry special significance. Salesforce has a large enterprise customer base across APAC, particularly in financial services, retail, and manufacturing.

Financial Services: The Long-Horizon Runtime enables agents to handle complex loan approval, insurance claims, and compliance reporting processes — workflows that typically span multiple workdays and involve multiple departments.

Retail: The Hunter and Piper agents are particularly suitable for APAC e-commerce and omnichannel retail enterprises, automating lead generation and customer engagement.

Manufacturing: Marshall's supply chain and procurement capabilities have direct application value for manufacturing enterprises with complex supply chains across Asia-Pacific.

The launch of Claudeforce also provides APAC enterprises using Anthropic Claude with direct Salesforce data integration capabilities, without additional integration development work — significantly lowering the technical barrier to AI agent deployment.

Market Impact and Competitive Landscape

The AIforce launch marks Salesforce's strategic transformation from a traditional CRM software company to an "agentic enterprise operating system." This positioning directly challenges Microsoft (Copilot 365), ServiceNow, and SAP in the enterprise AI agent market.

The 7 billion AWU figure also provides an important industry benchmark: enterprise AI agents are no longer experimental technology — they are processing real business tasks at quantifiable scale. This figure will become an important reference point for other enterprise AI platforms.

Conclusion

Salesforce's Dreamforce 2026 announcements clearly outline the future of enterprise AI agents: from single assistants to multi-agent collaborative ecosystems, from single-turn interactions to long-horizon task execution spanning days and weeks, from closed CRM systems to open, interface-independent computing layers. For Asia-Pacific enterprises evaluating their enterprise AI agent strategies, Salesforce's AIforce and Agentforce ecosystem offers a mature, secure, and scalable option — particularly for enterprises already using Salesforce CRM, this upgrade represents significant added value to their existing investment.