Skip to main content

Hi everyone,

I'm trying to better understand how Agentforce Coworker is supposed to work in organizations that have significant business-specific configurations.

For example, we're using the Nonprofit Cloud (AFNP) data model and some of our fundraising metrics are calculated differently than a standard implementation. When discussing donations, we often need to consider both Gift Transactions and Gift Commitments, not just Gift Transactions alone.

My question is: how can we provide organization-specific instructions, context, or guidance to Agentforce Coworker so that its answers reflect our actual business rules and configurations?

I've been trying to find documentation on this, both through Coworker itself and through Claude, but I keep getting pointed toward Agentforce Studio. However, my understanding is that Agentforce Coworker itself is not directly customizable (or perhaps I'm missing permissions or a setup step?).

So I'm wondering:

  • Is there a supported way to provide custom instructions to Agentforce Coworker?
  • Should this be done through Prompt Builder?
  • Should it be done through Agentforce Studio?
  • If so, how does Coworker know when to apply those instructions versus its default behavior?
  • Is there a way to define organization-wide business context that Coworker can consistently use?

Conceptually, I'm trying to understand whether Agentforce Coworker is intended to be a generic Salesforce assistant, or whether Salesforce envisions it becoming an organization-aware coworker that can reason using company-specific business rules and data models. If not, I suppose that specific agents are the solution, right? 

 

I'd love to hear how others are approaching this.

Thanks! :)

2 件の回答
  1. 9月18日 16:29

    Thanks, this is really helpful and it confirms what I was suspecting :)

    I think I'll go down the Agent route instead.

    One follow-up question: is there anything specific that needs to be configured for Coworker to leverage those agents? For example, will Coworker automatically look at the agents' instructions, descriptions, topics, or keywords and determine when to use them?

    Or does the user need to explicitly invoke a particular agent?

    I'm mainly trying to understand how the routing/selection works in practice so that Coworker can take advantage of the business-specific guidance defined in our agents.

    Thanks again, and have a great weekend!

0/9000