Introduction: TDX 2025 – the clear start of the agentic era

Salesforce TrailblazerDX (TDX) 2025 delivered game-changing announcements that all Salesforce professionals should know. Held between March and June 2025 in San Francisco, Tokyo, Bengaluru, and London, TDX 2025 was strongly focused on AI and the new Agentforce platform. 

The event’s messaging made it clear that Salesforce is committed to a major focus on artificial intelligence. The main highlight was Agentforce 2dx, Salesforce’s next-gen AI agent platform, which is now deeply integrated across the ecosystem. 

From admins and developers to architects, everyone’s toolkit is evolving. This article highlights the most important TDX 2025 announcements, starting with the biggest news, and explains the top five updates for admins, developers, and architects.

Image source: Salesforce channel on YouTube

TrailblazerDX 2025 overview: AI-first innovation

TrailblazerDX 2025, held in San Francisco, was a very important event showcasing Salesforce’s deep commitment to AI, particularly through its Agentforce platform. The conference focused on building and deploying intelligent agents.   

1. Conference focus: the AI agent era

The central theme was the “AI agent era,” highlighting AI as core to the Salesforce platform. This highlighted Agentforce 2dx’s evolution to proactive, autonomous agents revolutionizing workflows.  

Salesforce reports that 96% of developers believe AI agents will positively reshape their work.

 

Image source: Salesforce

2. Agentforce 2dx: proactive AI at scale

Agentforce 2dx, a major highlight, can proactively identify issues and initiate communication without user prompts, transforming AI into a smart system that manages workflow. Its effectiveness relies on Data Cloud as a foundational data source.

3. AgentExchange: a new marketplace for AI solutions

Salesforce introduced AgentExchange, a new trusted marketplace for Agentforce, mirroring AppExchange. It allows quick discovery, deployment, and customization of partner- and community-built AI templates and actions.   

Image source: AgentExchange

4. Broader platform innovations

Beyond Agentforce, TDX 2025 featured enhancements to Lightning Web Components, ApexGuru, and Salesforce Flow.   

5. Global reach of TrailblazerDX

TDX is expanding globally with events in Tokyo, Bengaluru (TDX 2026 returning), and London, making learning opportunities more accessible worldwide.   

Your journey to becoming an Agentblazer: upskilling for the future

Salesforce is actively driving a community-wide transformation, trying to give one million “Agentblazers” the necessary AI skills by the end of 2025. 

This initiative recognizes the important statement by Leah McGowen-Hare, Salesforce SVP, that “Tomorrow’s jobs belong to today’s learners“.

As business growth relies on people and not enough workers are appearing, AI is crucial for increasing what people can do, making it essential to learn new skills.   

To make this easier, Salesforce introduced structured learning and recognition:

• Agentblazer statuses: A new recognition system with three levels: Champion, Innovator, and Legend, earned by completing Trailhead modules and passing the Agentforce Specialist certification.   

Image source: Trailhead

• Agentforce specialist certification: This updated certification validates expertise in Agentforce, serving as a key credential.   

Trailhead resources and the free Agentforce Developer Edition provide practical learning environments. The formal introduction of these statuses and certification shows a clear expert area, making these AI skills very important for advancing your career within the Salesforce ecosystem.   

And if you’re pursuing the Salesforce Agentforce certification, consider resources like the Agentforce Specialist Study Guide to structure your learning.

Top 5 updates for Salesforce admins

For Salesforce administrators, TDX 2025 brought a mix of productivity boosters and signals of how the admin role is evolving in the AI era. Here are five key updates admins need to know.

1. Agentforce for Setup – your new AI assistant in Setup

Perhaps the most exciting admin feature is Agentforce for Setup, a built-in AI helper for the Salesforce Setup area. This Agent can handle common admin queries and tasks via natural language. 

• Initial capabilities: It can answer common task questions and assist with creating custom report types, explaining Connected Apps, and helping to understand user permissions.  
• Future roadmap: Planned features include generating formulas from natural language prompts, creating objects and fields, guiding navigation to specific pages, and assisting with authentication setups.  
• Bundling: Agent for Setup is bundled with the Agentforce platform, meaning if Agentforce for Sales is purchased, Agent for Setup is automatically included.  

Importantly, Agentforce for Setup is designed to save admins time on routine work, not replace them.

2. Free Developer Edition orgs with Data Cloud and Agentforce

A significant announcement for administrators was the availability of a new, free Salesforce Developer Edition that includes both Agentforce and Data Cloud. This directly addresses the community’s desire for hands-on access to Agentforce.

• Robust environment: It offers 10 GB of Data Cloud storage and 150 LLM generations per hour, providing an effective environment for experimentation and learning.  
• Practical experience: This provides a practical environment for hands-on experimentation with unifying customer data and building segments, allowing admins to understand the foundational role of data in AI agent performance.

Image source: Salesforce channel on YouTube

3. AI Assistance in Agent Builder: Accelerating Agent Creation

AI Assistance in Agent Builder allows administrators to describe their desired agent function in natural language. Agent Builder will then generate the necessary instructions and actions, significantly accelerating agent creation for low-code builders.  

• Simplified setup: This feature simplifies the setup process by automatically generating logic, topics, and actions based on a clear description of the agent’s role.  
• Feedback on instructions: It also offers feedback on topic instructions and descriptions, further accelerating agent creation and scaling the building of multiple agents.  

4. DX Inspector: Enhanced Metadata Management

DX Inspector is a new management tool that allows administrators to view the metadata of any Agent Builder project and make small adjustments to applications.  

• Unified view: It provides a unified view of all modifications across the org in its Changes tab.  
• Hierarchical view: The Metadata Explorer within DX Inspector offers a hierarchical view of agent metadata, including agents, topics, and actions, for deeper insights and streamlined management. This helps admins track changes and ensure consistency across their AI implementations.  

5. AgentExchange: The Marketplace for AI Agents

Following the huge success of AppExchange, Salesforce launched AgentExchange, a new trusted marketplace specifically for Agentforce.

• Discover and deploy: This platform allows customers to discover, deploy, and customize partner- and community-built Agentforce templates, skills, and actions with a single click. It launched with over 200 partners and hundreds of prebuilt assets, significantly accelerating AI agent adoption.  
• Component types: AgentExchange supports various component types, including Actions (ready-to-use integrations), Prompt Templates (reusable prompts), Topics (grouped actions and instructions), and Agent Templates (pre-built AI workflows).  

Top 5 updates for Salesforce developers

For developers, TDX 2025 brought many new tools and features. A lot of these are about making it simpler to create, check, and add to AI agents using code.

1. Agentforce Developer Edition orgs (with free Data Cloud)

We’ve already mentioned the new Developer Edition version above, but this innovation is also of great importance for developers. It provides a personal dev environment pre-loaded with Agentforce and Data Cloud. Practically, this means you can write and test Apex that calls AI agents or build Flows with Agentforce actions, all in a free org. 

• Generous limits: It offers 10 GB of Data Cloud storage and 150 LLM generations per hour, providing a robust environment for experimentation and learning.  
• Scratch Org support: Scratch orgs now also support Agentforce and Prompt Builder, allowing for free tinkering.

2. Invoking Agentforce from Flow, Apex, or via API

Agentforce can now be invoked programmatically from Flow, Apex, or via API, enabling headless operations where agents work in the background without needing a chat interface. This provides developers with full control over when an agent takes action.   

• Agentforce API: Now generally available, this API uses the server-sent events (SSE) HTTP standard, ideal for long-running agents that show ongoing results, providing real-time feedback on progress. A synchronous mode is also available for simpler agent-to-agent API usage.   
• Invocable actions: Agentforce Invocable Actions are generally available, allowing agents to be triggered within Flow Builder or Apex.   
• Trigger with Change Data: Agentforce can now be automatically invoked whenever a record changes, enabling proactive automation.   

3. Agentforce for Developers: AI-Powered Coding Assistance

Agentforce for Developers is a VS Code plugin that provides a “dev assistant” with features designed to offload tedious tasks and free up developers for more strategic work.   

• Code generation: It offers features like code autocomplete, Apex test class generation, Lightning Web Component (LWC) test case generation, and TODO code generation capabilities.   
• ApexGuru insights: ApexGuru, powered by xGen, provides actionable insights to identify performance bottlenecks in Apex code. It offers specific code recommendations, identifies critical expensive methods, and highlights unused classes to reduce technical debt. ApexGuru is currently generally available for Unlimited Edition and Signature customers in Scale Center and is planned to be generally available for Enterprise Edition by the end of 2025.   
• Code Analyzer updates: Integration with Code Analyzer 5 (currently in beta, planned for GA in Winter ’26) will enable finding and fixing common bad practices earlier, directly within VS Code, promoting cleaner and more secure code.   

4. “Action All The Things”: Expanding Agent Actions

Agent actions are expanding beyond Apex or Flow to include anything that can publish an OpenAPI Spec document, such as Apex REST APIs, Heroku APIs, MuleSoft APIs, and other third-party external services.   

• Orchestration layer: This means the AI agent itself is becoming a central point for orchestrating complex, multi-system processes, acting as a “hidden” intelligent automation layer that can initiate and complete tasks across the enterprise.   
• Platform events: Agents will also soon be able to publish and subscribe to platform events.   
• VS Code plugin: A new VS Code plugin is also available to generate the OpenAPI Spec for Apex REST.   

5. Lightning Out 2.0: Secure and Performant External UI

This feature allows embedding Salesforce UI in other web applications more securely and efficiently.   

• Enhanced security: It brings enhanced security through iframes and Content Security Policy.   
• Improved performance: It offers improved performance with asynchronous resource loading and Content Delivery Network (CDN) usage.   
• Aesthetic alignment: It ensures compatibility with both Salesforce Lightning Design System (SLDS) and SLDS 2 for better aesthetic alignment with host applications.   

Top 5 updates for Salesforce architects

Salesforce architects play a critical role in designing scalable, secure, and integrated solutions. TDX 2025 provided architects with crucial insights and tools to leverage AI agents effectively across complex enterprise environments.

1. Agentforce 2dx: enterprise-grade agentic automation

Agentforce 2dx represents a big step forward in Salesforce’s AI capabilities, evolving from reactive, chat-based AI to proactive, autonomous agents designed for scalable, enterprise-level solutions. Architects are very important for making sure these solutions are useful across Salesforce and other company systems.  

• Proactive capabilities: This new version moves beyond merely answering questions; it can identify potential issues, offer solutions, and even initiate communication without direct user prompts.  
• Deterministic control: It balances deterministic control with the non-deterministic aspects of generative AI outputs, enabling actions across any system and bringing agentic experiences where needed.  
• Specialized agents: Agentforce 2dx expands beyond a general AI assistant to include multiple agent types tailored for specific business roles and use cases, such as Agentforce SDR, Sales Coach, Service Agent, and Agent for Setup.  

Image source: Salesforce

2. MuleSoft for Agentforce: bridging data silos for AI

MuleSoft for Agentforce simplifies connecting Agentforce to various applications and services, allowing users to create topics and actions from MuleSoft APIs using natural language. This is very important for giving AI agents a strong base of complete, up-to-date company data.  

• Key support: Forrester’s analysis notes that MuleSoft is becoming a “support pillar” for a broader software development platform, rather than just an extra part of CRM.  
• Unlocking data: MuleSoft’s special ability to get data from systems that are difficult to access or located in the company’s own buildings is essential for AI effectiveness, as it acts as the crucial link connecting the AI’s intelligence to operational systems and data across the organization.  
• REST-to-GraphQL: MuleSoft can expose a GraphQL endpoint by acting as a proxy in front of REST APIs, allowing flexible data queries while leveraging existing REST investments.  

3. Data Cloud: The Unifying Foundation for Intelligent Agents

Salesforce’s strategy for AI agents is fundamentally built upon its Data Cloud, which provides a “deeply unified platform” for applications, data, AI models, and agents. 

• Single source of truth: This unified data and metadata model is crucial for eliminating data silos and creating a single source of truth across various departments.  
• Data Cloud governance: Data Cloud Governance will enable monitoring of data Agentforce accesses and ensure compliance across structured and unstructured data. It allows for rules to automatically tag and detect sensitive information and provides fine-grained access controls.  

4. AgentExchange: growing an open AI ecosystem

The launch of AgentExchange and the emphasis on the Agentforce API for seamless external integration, coupled with the “Action All The Things” initiative to ingest any OpenAPI Spec, clearly signals Salesforce’s strong dedication to an open ecosystem.  

• Trusted marketplace: AgentExchange is a new trusted marketplace specifically for Agentforce, allowing customers to discover, deploy, and customize partner- and community-built Agentforce templates, skills, and actions with a single click.  
• ISV Packaging: It supports ISV Packaging for Topics & Actions, enabling developers to monetize their agentic components.  
• External integration: The Agentforce API allows access to Agentforce solutions across the enterprise and allows connections to outside company systems to build experiences that link together.  

5. Comprehensive agentic development lifecycle: build, test, deploy, observe

Salesforce introduced a full suite of tools to manage the entire development lifecycle of agentic applications, making sure they deliver strong, big-company-ready solutions. This directly addresses concerns about AI unpredictability by providing mechanisms for thorough testing, metadata management, and controlled deployment.  

Image source: Salesforce channel on YouTube

• Testing Center: An upgraded user interface with support for batch testing enables the generation of hundreds of automated, AI-powered tests to check Agentforce’s performance and whether it follows instructions before deployment.  
• DevOps Center integration: Enables both low- and pro-code developers to deploy their agents smoothly from development to production environments.  
• Agentforce Interaction Explorer: Will provide a comprehensive view of all agent interactions, performance metrics, and a historical record of conversations. It will help diagnose underperforming topics and offer detailed analytics on uptime, error rate, and escalation for reporting on return on investment (ROI).  

Final thoughts: preparing for the agentic era

TrailblazerDX 2025 was a big turning point, showing that Salesforce is fully committed to an “AI-first” approach and the “AI agent era”. This change means professionals need to become problem solvers who always think about AI first. They should be able to create and handle smart automated systems that make human abilities even better.

Image source: Salesforce channel on YouTube

Key announcements collectively demonstrate together show Salesforce wants to put smart automation into every part of the enterprise. 

Salesforce is positioning Agentforce as the main way to access the huge $6 trillion digital labor market opportunity, increasing human capacity to address labor shortages and drive economic growth. AI is now the fundamental core of the Salesforce ecosystem, not an optional add-on. 

For every Salesforce professional, the message is clear: embrace this transformation. The Agentblazer program and Agentforce specialist certification provide a clear path to learn new skills and stay ahead in this revolution.

To further your expertise and prepare for the future of AI on Salesforce, consider exploring the Agentforce Specialist Study Guide from Focus on Force.