Salesforce launched a developer edition of Agentforce and a marketplace for pre-built partner agents and actions.
The launch comes as Salesforce follows up with its next phases of its Agentforce rollout. The focus on developers, announced at Salesforce's TDX conference, is designed to build an ecosystem around Agentforce.
For Salesforce, it's critical that Agentforce courts developers as well as administrators. Salesforce is betting that generative AI as well as AI agents revamp how user access software.
- Salesforce landing Agentforce deals, but Q4 and outlook mixed
- Salesforce, Google Cloud expand partnership: Here's what it means
- Behind the Scenes: The Force Behind Agentforce
- Salesforce launches Agentforce 2.0 as it ramps its release cadence
- Salesforce Dreamforce 2024: Takeaways on agentic AI, platform, end of copilot era
The company has unified its various clouds and integrated them with Data Cloud, which enables Agentforce in many ways.
Agentforce Developer Edition, which will provide Agentforce, Data Cloud and Salesforce APIs in a free developer environment, will include features that enable enterprise builders to manage the life cycle of an AI agent including, planning, building, configuring and deploying them.
Other tools will include a metadata explorer and change tracker to assist with building agents.
Agentforce Developer Edition will also include tools for monitoring costs, analyzing performance and testing to ensure agent deployments are efficient.
Key items include:
- Batch testing of agent quality using AI-assisted test case generation.
- Zero copy data integration.
- Open ecosystem and model choices.
- Tools to convert bots into agents.
Salesforce will demo the steps in the Agentforce developer lifecycle. Those steps and features include:
- Plan with a bot to agent migration tool and developer edition for Agentforce and Data Cloud.
- Build agents with AI assistance, AppDev bar and configure Agentforce APIs.
- Test via Testing Center.
- Release and deploy via DevOps Center.
- Monitor costs with Digital Wallet, which tracks consumption, Agent Analytics and Agentforce Interaction Explorer.
The company at its developer conference also outlined Agentforce 2dx, a new version of Agentforce that adds low-code tools to deploy agents.
Agentforce 2dx, available in April, includes:
- Agentforce API to integrate back-end processes, systems and applications.
- Agentforce Invocable Actions, which enable Agentforce to be embedded into Salesforce business logic.
- MuleSoft for Agentforce, which includes a topic center to use natural language to create Agentforce topics into MuleSoft APIs. Salesforce is using the MuleSoft API Catalog to simplify AI agent development. A MuleSoft Agentforce connector gives developers the ability to integrate AI agents into existing workflows and integrations.
- Agentforce Steps in Slack Workflow Builder to create no-code automation within Slack.
- Agentforce Employee Template.
- Agentforce Surfaces, which leverages AI agents across digital engagement channels, Agentforce Cards that put Lightning web components in agent actions, and Tableau Semantics to create data structures and semantic models.
Salesforce also launched AgentExchange, which will allow developers to integrate new actions into agents. Enterprises will also be able to discover Agentforce agents from partners. AgentExchange builds on Salesforce's long-running AppExchange.
AgentExchange will use Agentforce to instantly find apps and agents using natural language and recommend partner actions and topics in Agent Builder.
According to Salesforce, AgentExchange includes more than 200 partners including Google Cloud, DocuSign, Workday and Box.
Components available on AgentExchange include:
- Actions that can expand the job agents built with Agentforce can do with new integrations.
- Prompt templates including pre-written, reusable prompts.
- Topics to focus and refine agent behavior.
- Templates for agents across multiple topics and actions with metadata and instructions.
