My summary of Salesforce Agentforce looking into it — A couple of months ago I saw this tweet about how a Salesforce AI agent can do traditional sales leads management by sending emails, answering replies and getting meetings booked in.
AI agents managing sales leads is an example of Salesforce’s new Agentforce platform, through which businesses can build and manage autonomous AI agents.
How does Agentforce explain itself in the first minute? “An Agentforce Agent is a proactive, autonomous application that provides specialised, always-on support to employees or customers. They’re equipped with the necessary business knowledge to execute tasks according to their specific role.” The platform offers out of the box Agentforce Agents for its customers to use: Service Agent, Sales Development (SDR) Agent, Sales Coach, Personal Shopper Agent and Campaign Agent. I’m particularly interested in AI agents can impact customer support, sales and marketing:
Sales Development Representative (SDR) Agent
Agentforce customers can decide which specific bot features to enable (e.g. Einstein Copilot and Einstein Generative AI). The agents can then be customised; you can set a tone of voice or determine the topics that you want the agent to handle (and how you’d like the routing of specific topics to work). You add the files to train Salesforce’s data model on, using RAG to make sure the agent retrieves the right information.
Salesforce already has a strong reputation for its marketing campaign tools. The Campaign Manager intends to build on this reputation by enabling marketers to use natural language prompts to create target segments, campaign briefs and dynamically generated multi-step campaigns.
The flows in the multistep campaign can be updated by using Salesforce’s Einstein co-pilot which will suggest next best actions to the marketer.
Salesforce offers the Prompt Builder for all its generative AI based functionality; providing prompt templates that can be easily created, configured and reused. These prompts can be ‘grounded’ in company CRM data so that the output generated through the prompt is personalised but still reusable.
Did Salesforce Agentforce deliver on my expectations?
Yes. Agentforce provides customer and employee facing AI capability that you would expect from Salesforce CRM. Using the Salesforce product ecosystem, users can create end to end workflows or prompts, fully supported by Salesforce’s Einstein copilot.
Main learning point:
Whilst I was learning about Agentforce, I started thinking more broadly about Agentic AI Systems. Thinking about editable work spaces within the AI application to enable content creation collaboration and modification (with Open AI Canvas and Anthropic Artifacts as good examples).
Or thinking about multiple AI Agents conversing with each other to automate complex tasks or processes. Depending on complexity, a task can be done based on single or multiple LLM calls by an AI Agent. Adept, for instance, is a relevant product in this space which I want to learn more about.
Related links for further learning:
