Salesforce: New Agentforce version especially for developers

Contents

The software company Salesforce presented a new version of its AI agent platform Agentforce at its TDX conference in San Francisco. Agentforce 2dx is aimed specifically at software developers. It includes new functions for embedding AI agents in existing applications and, above all, a comprehensive support package for developers. This includes a range of low-code and pro-code tools to configure, test and deploy Agentforce faster and more securely.

In combination with the advanced analytics, the performance of AI agents can now also be debugged, monitored and optimized using real-time data. Administrators and developers can try out all these tools for free in the new Agentforce Developer Edition. This includes 10 GB of storage space, access to the Data Cloud and 150 LLM generations per hour.

Embedding with REST API

Agentforce 2dx offers the following new features and modules: Agentforce API now allows Agentforce to be embedded in existing backend processes of third-party systems via REST API. With the new callable Agentforce actions, Agentforce can be integrated into the Salesforce process environment such as Flow and Apex. Agentforce can be used both as part of Flow and programmatically in Apex.

Salesforce has significantly expanded the Mulesoft integration platform to better support Agentforce. The Topic Center now allows developers to create topics and actions in natural language. Mulesoft's new API catalog simplifies the development of AI agents by centralizing access to various APIs from Mulesoft, Salesforce and Heroku within Agentforce.

The Agent Builder has been expanded to include AI functions that primarily support troubleshooting. Another new feature is that developers can create, update and test Agentforce configurations using CLI and VS Code. All aspects of an Agentforce configuration can be created, updated and tested using Pro Code tools. An extended Test Center is available for this purpose. It offers new analysis tools and allows teams to better understand Agentforce integrity and performance.

Marketplace for AI agents

In addition to Agentforce 2dx, an expansion of the Salesforce marketplace AppExchange to include the so-called AgentExchange was also announced. Ready-to-use agents are to be provided here by partners and the community. The platform starts with more than 200 partners and hundreds of ready-made actions. Initial partners include Docusign, Box and Workday. Salesforce hopes that AgentExchange can build on the success of AppExchange, where over 13 million apps are now available.

An extensive cooperation with Google was also announced. Gemini will be connected to Agentforce to provide agents with images, audio and video. To this end, Salesforce is connecting the Service Cloud with the Google Customer Engagement Suite. According to Salesforce, this enables better, AI-supported contact center functions such as real-time language translation, intelligent handovers from one agent to another, personalized agent recommendations and AI-driven conversation insights across all channels.

Agentforce's apps, Data Cloud and Customer 360, run on the Google Cloud infrastructure. Users can also use Agentforce Grounding with Google Search via Vertex AI, providing up-to-date data, news, events and credible quotes. For example, an Agentforce-created agent in supply chain management and logistics could track shipments and account for potential disruptions using real-time data from Google Search, including weather conditions, port congestion and geopolitical events.

AI is transforming software development

Salesforce sees the new agent AI as a paradigm shift in the IT landscape. "The traditional notion of data, logic and user interface is no longer helpful," said Salesforce manager Adam Evans at the announcement of Agentforce 2.5. "Data is moving from strictly structured databases to unstructured data with vector databases and semantic understanding. And logic is moving from hard-coded if-then rules with declarative logic to more dynamic, non-deterministic, more natural language-based methods," is his rationale for the new IT structures.

Salesforce manager Alice Steinglass therefore also expects a fundamental change in the entire software development process. "The focus is shifting from coding to achieving business goals," she predicts. She refers to a Salesforce survey, according to which 92% of developers believe that "AI can significantly increase their productivity and performance". This is probably also the view at Salesforce – At the most recent quarterly results, CEO Marc Benioff announced that he would not be hiring any developers this year due to the productivity gains achieved by AI tools.

(axk)