Loading…

What we learned at TDX 2025

Some high-level takeaways, with more to come.

Article hero image
Credit: Alexandra Francis

Last week, we were on-site in San Francisco at Trailblazer DX, the developer conference of cloud colossus Salesforce. TDX 2025 was all about the future of AI agents and agentic AI’s potential to reimagine software development and the developer experience.

If you’re new to the topic or could use a refresher, AI agents are AI systems that use machine learning (ML) and natural language processing (NPL) to understand and respond to users without human oversight or intervention. They are self-learning, capable of teaching themselves to improve.

The promise of AI agents is that they free organizations to focus on the what (as in, what do we need to accomplish? what are our business requirements?) and let the AI figure out the how (as in, how do we make it happen?).

Here, we’ll share some high-level takeaways from the event, with more to come.

Salesforce is all in on agentic AI

Introduced in 2024, Agentforce is Salesforce's platform for building, customizing, and deploying autonomous AI agents. At TDX 2025, Salesforce announced Agentforce 2dx, the latest version, which the company says “expands beyond the reactive, user-initiated world of chat interfaces and enables proactive AI agents to work behind the scenes, without constant human oversight, to unlock new customer and employee workflows of any kind.”

A bevy of new tools in Agentforce 2dx will let developers and admins embed autonomous agents into their workflows to figure out those all-important hows. There are new pro-code tools as well as low-code and no-code tools that further lower the barrier to entry for those without a pro-coding background.

Also announced was AgentExchange. Along the same lines as Salesforce’s AppExchange, which gave companies an on-ramp to the cloud economy when it was launched 20 years ago, AgentExchange is a community marketplace where Agentforce users can find and market ready-to-use AI agents and templates. The idea is that these fully-functional agents will make it faster, easier, and cheaper for organizations to build and deploy agentic AI, much the way AppExchange opened up cloud computing. Big names like Google Cloud, DocuSign, Workday, and Box are partnering with Salesforce to build AI agents via AgentExchange.

Developers can build AI agents for free

Good news for developers who want to get their hands dirty: Agentforce is included along with Data Cloud in the Salesforce Developer Edition, a free Salesforce environment where developers can experiment with building AI agents that interact with Salesforce users, end customers, or act autonomously. For the first time, the company announced at TDX, devs “can explore Salesforce’s latest AI and data technology in a dedicated environment that doesn’t expire as long as you keep using it.” Call it motivation to check in every 45 days.

The developer edition lets users build, customize, and test AI agents for free, trying their hand at using Agentforce and Data Cloud to bring agentic reasoning into the development workflow. Tools like Agent Builder with AI assistance help development teams get their feet wet (apologies for a second appendage metaphor, reader). Devs can also create, manage, and test agents via Salesforce CLI and VSCode.

Looking for more technical resources? Find them in the Agentforce Developer Center and Data Cloud Developer Center.

In future posts and a podcast conversation, we’ll unpack more of what we learned about agentic AI, how it’s already transforming the developer experience, and the central role data quality plays in the success or failure of your AI projects.

Were you at TDX last week? Do you have thoughts about agentic AI? (We know you do.) Let us know in the comments!

Add to the discussion

Login with your stackoverflow.com account to take part in the discussion.