Salesforce provides new low-code tools for developing AI agents within its own Agentforce platform. The agents can now also take proactive action.
When Salesforce introduced Agentforce 2 barely two months after launching Agentforce, with the version number being the most visible change, we speculated that edition 532.0 wouldn’t be far behind. Today, less than three months after Agentforce 2, Salesforce takes another step in that direction with the announcement of what it calls Salesforce 2dx.
It’s unclear whether an overzealous AI agent or a team of marketing interns is behind the Agentforce naming. We’ll set aside the confusing version numbering (from no version to version 2 to version 2dx in less than half a year) and look at what’s truly new.
Proactive Actions and Developer Tools
At its TDX developer conference, Salesforce introduces two major improvements to its AI ecosystem. First and foremost, the company announces that agents can now be proactive on the front end. This means that autonomous AI-driven functionality can take action on its own. Previously, human interaction was needed, for example via a chatbot. Now agents can respond independently, for instance to a change in the underlying data.
read also
Salesforce expands Agentforce and needlessly adds ‘2.0’
Next, Salesforce looks to developers. They receive a new set of both low-code tools and classic development environments. With these, they can get to work building, testing, and deploying agents faster.
The tools are available in the new Agentforce Developer Edition. This is a free version of Agentforce where developers can experiment and build prototypes. Data Cloud is also linked to this developer version.
The announcement comes alongside the introduction of AgentExchange, which Salesforce launched at the same conference. AgentExchange makes it easy to retrieve agents (or components thereof) from a catalog and quickly implement them.
Efficiency, Jobs, and Expectations
Salesforce introduces all these improvements with the aim of bringing more efficiency to customers. They should be able to extract value from their data with AI that is easy to implement within the ecosystem they already have. The company claims that it is experiencing savings itself. Although AI experts recently declared in unison that AI would not pose a threat to jobs, Salesforce, driven by AI, just announced that it would no longer hire new engineers.
The company needs to continue pushing Agentforce as a success story. Adoption is indeed slower than (over-enthusiastic) investors anticipated, resulting in negative reactions on the stock market. The expansion of Agentforce functionality is undoubtedly important, though it’s questionable whether the constant updating of Agentforce with new version numbers isn’t contributing to these unrealistic expectations.