Skip to main content
Grupo

Agentblazer Community Group

Welcome, Agentblazer! Are you ready to connect with AI leaders and enthusiasts from around the world? You've come to the right place. The Agentblazer Community is the premier destination to skill up on AI, discuss best practices, gain access to Agentforce product experts, showcase your expertise, and supercharge your career.

Hi Team I was practising https://trailhead.salesforce.com/content/learn/projects/customize-a-service-agent-with-prompts-flows-and-actions/update-a-flow-agent-action-and-topic?trail_id=become-an-agentblazer-innovator-2026

this trailhead and I saw get sessions method being called 

 

actions:

Get_Sessions: @actions.Get_Sessions

with experienceId = ...

with startDate = @variables.converted_date

        although in trailhead it was not mentioned to try but i wanted to see how to pass  vairables to actions .I observed that irrespective of variable value available or not it always passes null 

in other actions slot filling mechanism is used .  

Get_Experience_Details: @actions.Get_Experience_Details

with experienceName = ...

    As per observations it seems to call any actions we have to use slot filling only variables cannot be used.  

Community please correct me if i am doing wrong

2 respuestas
  1. Hoy, 14:58

    @Ankit Jain I would not read this as "variables cannot be passed to actions." They can be used, but the timing matters.

    In Agent Script there are two different patterns. If the action is exposed to the LLM as a reasoning action, slot filling can populate the action inputs. That is the with experienceName = ... style. The LLM decides the value at runtime.

    If you pass with startDate = @variables.converted_date, then converted_date must already have a real value before that action runs. If the variable was not explicitly set earlier, or if it is only mentioned in instructions but not actually assigned, the action will receive null. Salesforce’s docs say variables are used to store state and can be used as inputs to actions, but when actions are run deterministically in instructions, inputs and outputs must be manually assigned because the action runs before reasoning takes place. 

    So for your case, either let the action slot-fill the date directly:

    with startDate = ...

    or set converted_date first through a prior action/logic step, then call Get_Sessions using that variable.

    Also check that converted_date is defined as the correct type, likely date, and that the value is in the expected format. Salesforce’s Agent Script variable reference says variables are referenced as @variables.variable_name in script, and as {!@variables.variable_name} inside reasoning instructions. 

    Slot filling works when the LLM is deciding the input. But variables also work, as long as they are populated before the action is called.

0/9000
1 comentario
0/9000
0/9000

The Agentblazer Collection 🗞️ 

 

What is it? 

The Agentblazer Collection is a weekly post shared in the community with resources from Salesforce related to Agentforce, community-contributed resources, and other exciting items ✨ 

 

Featured Resources this Week:

 

Agentblazers, hope your day is going well 😊

2 comentarios
0/9000

Can someone help me when trying to ask questions related to Knowledge object , Agentforce doesn't do anything even if Knowledge grounded data is setup with fields accessible and user also has access to all Knowledge articles?

Are there any pre requisites to do? 

11 respuestas
  1. 31 ene 2025, 12:52

    Following the quick and easy setup of Service Agent and its 'Answer Questions with Knowledge Action' doesnt tell you that you first need to have Data Cloud enabled. 

     

    Even once data cloud was enabled,  it wasnt working for us, the agents were not consuming the knowledge articles. 

     

    What we found is that because we created the agents and its required einstein data library BEFORE enabling data cloud,    it didnt create the Search Index. 

     

    We were able to make it work : 

     

    - Make sure Data Cloud is enabled 

    - Create a NEW einstein data library 

    - Go to Apps->Data Cloud->Search Index 

    - You should see a new search index created with a status of In-Progress 

    - While its processing,   go to your agents builder > Data Library and change the einstein data library selection to the new one you created.  

    - Once all agents are no longer linked to the old data library, delete it. 

    - Once the Search Index status changes from in progress to completed/ready,   the 'Answer Questions with Knowledge Action' actions properly linked to this new Einstein data library will start to work. 

     

    Enjoy !

0/9000

If you loved the and want to put the Well-Architected Framework into actual practice, this is something you’ve been waiting for.If you loved the #TDX26AgentblazerChallenge and want to put the Well-Architected Framework into actual practice, this is something you’ve been waiting for. 

 

Salesforce Developer Group, Dubai, UAE  is bringing a brand-new series for the Trailblazer Community: De-Fragging the Franken-Org!

 

Helping you Make your Legacy Salesforce Org Agentforce Ready. 

 

Register now to secure your seat and enter to win exclusive Well Architected Prizes.  

 

Registration Link : https://forms.gle/P1Sip2URYGSdaGK27

 

@Salesforce Developer Group, Dubai, UAE @WFD Partner Cohort: TDX2630DaysChallenge @Agentblazer Community Group @* Salesforce Developers * @Architect Trailblazers 

2 comentarios
  1. 28 may, 00:07

    Hi @Niki Vankerk

     

    There are weekly sessions, Sunday 11:00 AM -1:00 PM GMT+4, in your case PDT will be 12:00 AM. You may not be able to catch live-session but you may access the recordings and participate 

     

    Game 1 rewards participants who take the series content and apply it to a real Salesforce org, demonstrating measurable improvement in their org's Agentforce Readiness Score across the 8 weeks. 

     

    Game 2 rewards the participant who shows up most consistently, engages most genuinely with the community, and contributes most to the series' reach and impact across all 8 weeks. 

     

    Game 3 rewards knowledge and consistent learning — open to absolutely everyone, regardless of org access, LinkedIn following, or technical level.

0/9000

hello trailblazers, 

 I am encountering a issue where my Custom Lightning Type (CLT) successfully overrides the UI for a standard

Employee Agent but fails to override the UI for a Service Agent running within the Enhanced Chat v2

environment. 

I followed the official documentation:

Custom Lightning Types for Generative AI (Agentforce)

.  

 

When the action is run by the

Service Agent (tested using the "Test Enhanced Web Chat" link), the output always renders as plain text (or the default Agentforce UI) and does not load my custom LWC

.  

 

What specific permissions, meta tags, or deployment steps might the Service Agent require that are

different

from the Employee Agent setup?  

 

Thank you in advance for any insights! 

3 respuestas
  1. 27 may, 15:57

    @Suresh Bandi

      

    I faced the same issue with Custom Lightning Types in Service Agent. 

     

    The problem was not with the renderer configuration. The root cause was missing Apex class access for the Service Agent runtime user. Once the required Apex access/permission set was assigned, the custom renderer started working correctly. 

     

    A few key points to check:

    •  Service Agent user must have access to all Apex classes involved in the CLT flow 
    •  Required permission sets should be assigned

     

    Also, an important observation:

    •  You may not see the proper rendered UI inside the Service Agent builder/preview 
    •  The Custom Lightning Type renders correctly in the actual Embedded Service Chatbot deployment, especially in Embedded Service Chatbot v2

     

    So always validate the behavior in the real deployed chatbot experience after publish, not only in the builder preview.

0/9000

Agentforce Demo Day is now on demand. 🎬

 

Real builders. Real agents. Real results — see how Indeed and SharkNinja are shipping AI agents in production today built on Agentforce. 

 

Watch the demos to learn how to:

  • Build and deploy agents directly in your IDE — no Salesforce UI required
  • Orchestrate and control agents with precision using Agent Script
  • Work with the unstructured data you actually have

 

👉 Watch Agentforce Demo Day

0/9000

Trailhead Tuesday! 

 

Learn how to build and configure custom agents using pro-code developer tools. 🚀 

 

From this project, you'll learn:

  • How to set up your environment with Agentforce DX tools.
  • How to author your custom agent using Agent Script.
  • How to apply agentic AI practices to real-world scenarios.

 

Get hands-on with Agentforce developer tools today!

0/9000

🚀 Shape the Future of Agentblazer Content (and enter to win!) 

 

The world of AI is moving fast, and we want to ensure Trailhead is keeping pace with your needs. 

 

Whether you’re just starting your AI journey or already building autonomous agents, we want to hear about your experience with Agentblazer Statuses.

Take our quick 10-minute survey to directly influence future badges and trails.

 

🎟️ Bonus:

 Complete the survey by May 15 for a chance to win 1 of 5 Salesforce Certification Vouchers!   

**See official rules in linked survey.

 

#Trailhead  #Salesforce Admin  #Salesforce Developer  #Certifications  #Salesforce

31 comentarios
0/9000