Dreamforce 2026 Takeaways: A Builder's Perspective

---

Takeaway 1: Builder Central (The New Architectural Mandate)

Summary: Salesforce has officially enabled AI to be at the core of every administration task and development workflow. This is not merely a feature update; it is a fundamental re-architecture where AI agents act as the primary interface for configuration, deployment, and maintenance.

What it means:

1. Configuration-as-Agents: Traditional click-path configurations (like complex Flow Builder or Apex logic) are being replaced by AI-driven agents that generate metadata records, define process definitions, and execute deployments directly.
2. Development Shift: The developer workflow is shifting from “writing code to perform actions” to “defining agent behaviors.” This means developers will focus more on defining prompt engineering, system prompts, and guardrails for agents rather than writing boilerplate Apex or LWC code for CRUD operations.
3. Unified Development Environment: The platform is moving toward a single, AI-centric IDE where all tools (CLI, UI, MCP) communicate through an agentic layer.

Actionable Insight: Future builds must leverage AI Central for metadata customization. Administrators should expect to delegate more tasks to AI agents for system maintenance and troubleshooting.

---

Takeaway 2: HXL (Hyper-Extensible Language Protocol)

Summary: Salesforce introduced HXL as the unified protocol for cross-interface communication between Salesforce Clouds, Slack, Agentforce Agents, and external AI models (such as Claude). This ensures consistent user experience (UX) regardless of the interface used to interact with Salesforce data.

What is HXL?

• Protocol: A lightweight language that allows agents and systems to communicate seamlessly across different platforms (Salesforce, Slack, Agentforce, etc.).
• Benefits:
  • Consistency: Users get a consistent experience whether they interact with Salesforce via LLMs, chatbots, or traditional UI.
  • Interoperability: External AI models can now access Salesforce data without requiring complex integrations or API calls.
  • Scalability: As new platforms emerge, HXL allows them to integrate seamlessly into the Salesforce ecosystem.

What it means:

1. Unified Customer Experience: Customers can interact with your organization via any channel (chat, voice, mobile) and get a consistent experience powered by AI agents using HXL.
2. AI Agent Orchestration: Multiple AI agents (e.g., an Agentforce agent for sales, another for support) can collaborate using HXL to resolve complex customer inquiries without human intervention.
3. Developer Enablement: Developers can build custom integrations using HXL to connect their internal systems with Salesforce data streams.

Actionable Insight: Developers should start building HXL-based integrations to ensure their solutions remain compatible with the evolving agentic ecosystem.

---

Takeaway 3: AIforce (Headless Toolkit & Agentforce Coworker)

Summary: AIforce is not just a new product—it’s a headless toolkit designed to empower developers and administrators to build custom agents for specific use cases without needing to understand the underlying infrastructure. Agentforce Coworker, as a companion tool, helps manage agent behaviors, permissions, and workflows.

What is AIforce?

• Headless Architecture: AIforce exposes APIs and SDKs that allow developers to create custom agents using standard programming languages (Node.js, Python, etc.).
• Agentforce Coworker: A dedicated tool for managing agent lifecycles, including:
  • Agent provisioning
  • Permission sets
  • Workflow definitions

What it means:

1. Customization Without Complexity: Developers can build highly specialized agents (e.g., a sales discovery bot, a compliance auditor) without needing deep Salesforce expertise.
2. Agent Orchestration: AIforce allows for orchestrating multiple agents across different domains (sales, service, marketing) to deliver seamless customer experiences.
3. Scalability: Agents built with AIforce can scale horizontally as demand increases, leveraging cloud-native infrastructure.

Actionable Insight: Teams should start prototyping with AIforce SDKs and building custom agents for niche business processes. Start by defining the agent’s persona, tools, and workflows using the Coworker tool.

---

Takeaway 4: Agentic Identity Management (Licensing & Ecosystem Impact)

Summary: Agentic identity management is redefining how Salesforce handles licensing, access control, and integration security. This change will impact Enterprise Search platforms such as Glean, Google Vertex AI Search etc.,, Agentic orchestration platforms such as Workato, UIPath, Camunda etc.,, and MCP (Model Context Protocol) ecosystems.

What’s Changing?

• Licensing Shift: Salesforce is moving from traditional user-based licenses to a model where AI agents are licensed independently of human users.
• Access Control: Agents will require their own identity, permissions, and audit trails. This means that every agent interaction must be logged and audited for compliance.
• Ecosystem Integration: Glean (enterprise search), eMCP (data access layer), and MCP will need to integrate with this new agentic identity model.

What it means:

1. Enterprise Search (Glean): Glean’s AI-driven search and data retrieval must now account for agent permissions and data sensitivity levels. Agents will require scoped access based on user roles and organizational policies.
2. Agentic orchestration (Workato eMCP): eMCP will need to implement robust identity management protocols to allow agents to securely access Salesforce data without exposing sensitive information.
3. MCP: The Model Context Protocol will evolve to support agentic workflows, where agents can dynamically request access to specific data resources based on real-time permissions.

Actionable Insight: Administrators must prepare for a new licensing model that includes agent-specific licenses. Security teams should audit current data access policies and ensure they align with the upcoming agentic identity framework. There is a real cost impact that should be factored in as part of your contract renewal conversations

Related reading: Agentic Identity and the Double-Counting Trap.

---

Takeaway 5: Zero Retention for Agent-Processed Data

Summary: Alongside the agentic push, Salesforce reinforced its zero-retention posture for customer data flowing through its AI layer: data your agents process is not retained afterward or used to train models. For regulated buyers, this is the control that makes the rest of the agentic story deployable.

What it means:

1. Regulated deployments unblocked: Zero retention answers the first security-review question for agents touching customer records.
2. Verify scope per surface: Retention terms can vary across Agentforce, AIforce-built agents, and Data Cloud ingestion — confirm each path contractually.
3. Audit still required: Zero retention covers the model layer, not your own logging. Agent audit trails continue to remain your responsibility.

Actionable Insight: Get zero-retention terms in writing for every agentic surface you deploy, and pair them with the scoped agent identities described above.

---

Takeaway 6: KOA Reasoning Model

Summary: Salesforce introduced KOA, its first in-house CRM reasoning model, as the intelligence underneath Agentforce agents. Read our full review: Salesforce Koa: The First CRM Reasoning Model and What It Means for Agentforce Buyers.

What it means:

1. Vertical reasoning over general models: KOA is built for CRM decision paths, which should mean fewer hedging failures in sales and service workflows.
2. Cost layer: Reasoning models consume more per decision — evaluate KOA usage against the double-counting trap before scaling pilots.
3. Buyer diligence: Pilot with your own data and measure resolution rates, not demo flows.

Actionable Insight: Start KOA pilots on one high-volume workflow, meter token and execution costs from day one, and compare against general-model baselines.

---

Summary Table: Key Takeaways & Actions

Takeaway	Description	Actionable Insight
Builder Central	AI at the core of all admin/dev workflows	Leverage AI for configuration and deployment; focus on agent behaviors
HXL	Unified protocol for cross-platform communication	Build HXL-based integrations for consistent UX
AIforce	Headless toolkit & Agentforce Coworker	Prototype custom agents with SDKs; define personas and workflows
Agentic Identity	Licensing changes impacting Glean, eMCP, MCP	Audit data access policies; prepare for agent-specific licensing
Zero Retention	Customer data not retained or used for training	Get terms in writing per surface; keep your own audit trails
KOA Model	First in-house CRM reasoning model behind Agentforce	Pilot one workflow; meter tokens; see KOA review

---

Conclusion

Dreamforce 2026 marks a pivotal moment in Salesforce’s evolution. The shift to an agentic operating system requires organizations to rethink their architecture, development practices, and security models.

Some suggested next steps:

• Prioritize AI Central: Integrate AI-driven tools into your workflow for faster configuration and deployment.
• Explore HXL: Begin prototyping integrations that leverage the unified communication protocol.
• Prototype with AIforce: Build custom agents for niche business processes using the headless toolkit.
• Plan for Agentic Identity: Audit licensing and access control policies to prepare for agent-specific licenses and permissions.

---

For questions or clarification, feel free to reach out.