Favorite
Write a comment...
0/9000
Featured group
Slack + Agentforce
Write a comment...
0/9000
So, after I passed the Agentforce Exam, I am interested to practice building AI agents for different uses. I like to share this practice with other professionals. If you are:
- Subject Matter professional
- AI/ Tech Builder
- Marketing expert
- Sales expert
Let's team up and build a golden future. DM me if you are an entrepreneurial-minded professional. No obligation or fees needed, just your ambition to learn.
Hi everyone,
I’m currently experiencing an issue with the General FAQ topic in Agentforce and wanted to check if anyone has encountered something similar.
When I execute the prompt directly from Home page, I receive the following response:
"I couldn't find any information regarding knowledge article registration in the available resources. Would you like me to escalate this issue to a live agent for further assistance?"
However, when I run the exact same prompt from the Agentforce Builder, it works correctly and returns the expected information from the Knowledge Base.
A few additional details:
- Same prompt used in both cases
- Tested using an Admin profile in both scenarios
- No configuration changes between tests
Has anyone seen this behavior before ?
Any guidance would be greatly appreciated. Thanks in advance!
7 answers
Hi @Pedro Reyes
-
Points to check -
1. Permissions assigned to Agent User
2. Data cloud objects access
We have a valid concern from the support team that today 50% of their vol is web, 25% chat and 25% phone - this will change to 75% chat (with web going away), so whilst will increase deflection opportunity, will also increase live chat instances. Just checking if we can offer the option to create a case for those that want to 'log and leave'. has anyone ever built to offer log a case option for people waiting in chat queue? Or offer live support and\or log a case option?
1 answer
Yes, this is a well-supported use case in Salesforce Service Cloud and Agentforce. Here are the recommended approaches:
Option 1: Agentforce + "Log and Leave" via Einstein Bot/Service Cloud Messaging
- In your Agentforce (or Einstein Bot for legacy) configuration, add a "Create Case" action that fires when a customer selects "Log and Leave" or when wait time exceeds a threshold.
- Configure this as a Flow Action: the flow collects the case details (name, issue description) from the chat transcript and creates a Case record with the conversation data attached.
- After case creation, send a confirmation message with the Case number and estimated response time.
Option 2: Omni-Channel + Queue Overflow with Case Deflection
- In Omni-Channel, set a Queue Overflow policy: when wait time exceeds X minutes, trigger a message offering to "log a case instead".
- Use a Messaging Pre-Chat Form or Agentforce topic to handle the case creation flow for customers who choose to leave.
Option 3: Embedded Service / Snap-ins "No Agent Available" flow
- Configure the Offline Support or "No Agent Available" settings in Embedded Chat to show a case creation form when all agents are busy.
All three approaches are possible today. Agentforce Option 1 is the most modern and aligns with your current investment in AI deflection.
In a newly created sanbox, I want to edit the agent but when I click to enter the agent configuration I get this screen that does not allow me to contuniate, do you have any idea how I can solve it?
Hi everyone 👋
My Mentor asked me to build an AI agent for a service-based company website, but I’m a bit confused about where to start.
I’d really appreciate guidance on:
- How to design and build one
- Whether it can be built using free tools or free tiers in salesforce
- What other possible approaches or alternatives I should consider
- Does ai agent in salesforce can be made without any licenses?
If you have examples, tools, learning resources, or real-world use cases, that would help a lot.
1 answer
Hi @Koppisetti Charan Sai, Please check that in your org Agentforce option is available, if available then only you can implement agentforce and after implement agentforce, then you need to make embedded services which will provide the code and then that you need to implement in the third party website.