Skip to main content
Group

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.

Salesforce Summer '26 ships 50+ pre-built AI agents for IT service management. 

 

How many of those are you planning to actually deploy? 

 

I've been thinking about the IT Service Domain Pack this week. The concept is strong — agents handling password resets, software requests, and hardware provisioning without Level 1 tickets. That frees up IT staff for work that actually needs human judgement. 

 

But "out of the box" still requires connecting each agent to your systems, mapping your workflows, and doing proper change management. That's not a dealbreaker — it's just the realistic scope. 

 

My question for the community: 

 

For those planning to adopt the IT Service Domain Pack — what's your approach? 

 

→ Pilot with 2-3 agents, validate, then expand? 

→ Go broad from day one and configure in parallel? 

→ Wait and see how early adopters find it? 

 

And is there a specific IT workflow you're hoping one of those 50 agents covers? Curious what the actual demand looks like from practitioners. 

 

#Agentforce #Salesforceadmins

0/9000

The new Ask Agentforce / Coworker seems excellent.  We've set it up for a pilot group in our org and it seems to be really powerful.  Great UI and should be able to hand-off to other Agents you build.  Makes for a much better AI experience on early testing. 

 

However, we're unable to fully set it up and as a Beta product it's tricky to get support so wondering if anyone has any recommendations or

@Joel Primack

if you have any way of passing on contact to the product team. 

 

Standard search funcitonality seems to work really well.   It can access our org metadata to figure out what a pilot user is asking about, retreive matching records. 

 

However, when it needs to either query by owner (what tasks do I have due today?) or needs to hand off to an agent users get an error.   

 

e.g. 'It looks like the current session isn't authorized to access your user details right now — I'm getting an

INVALID_SESSION_ID

error, which means the session may have expired or doesn't have API access enabled.' 

 

I've checked the recommended settings.  I suspect it may be due to us having API Access Controls enabled in our org and there may be an Oauth policy I need to configure but I can't find any documentation to help with this. 

 

I'd love to be able to use this new Coworker experience before we start extending access and rolling our our agents so any help very much appreciated.

#Agentforce 

5 answers
  1. Jun 13, 3:28 PM

    Thanks @Joel Primack

     

     

    I tested with 'lock session to originating domain' disabled and still get the same error.  Summer 26 dropped into our org so I see the updated branding (Coworker rather than Ask Agentforce) but again no effect on the error. 

    Will wait to hear from you or the product team

0/9000

Summer '26 is 3 days out. What's on your pre-launch checklist? 

 

Here's the 6-point checklist I run before every major Salesforce release: 

 

1. Audit active Flows — especially anything touching Leads, Cases, Opportunities 

2. Lock down Autonomous Field Permissions — Agentforce defaults are permissive 

3. Refresh your Sandbox — test Summer '26 features there first 

4. Review Permission Set assignments — new Agentforce features = new permissions 

5. Brief your end users — the UI is changing; a 3-sentence email prevents a wave of tickets 

6. Document your rollback plan — you can't revert the release, but you can disable features fast 

 

How are you preparing for Summer '26? Anything on your checklist I missed? 

 

Drop your prep tips below 👇 

 

#Agentforce #Salesforce Admin

1 comment
  1. Yesterday, 5:25 PM

    That’s a solid checklist. One item I’d add is reviewing integrations and API-dependent automations, since release changes can sometimes create unexpected issues outside of core Flows and permissions. I also like validating key reports and dashboards after sandbox testing to catch any data visibility problems early. Similar to how users ask how much does David Lloyd membership cost, before committing to a gym, end users often want clear expectations before a major platform change, which makes your communication step especially valuable.

0/9000

we’re building a knowledge agent for a customer. When I use the OOB Agent action (Answers questions with Knowledge), it works well. The Prompt (Answers questions with Knowledge) uses the GPT 4 Omni model, but per our requirement, we need to use the Anthropic model. I created new version of prompt to use Anthropic, and it works in prompt simulation, but I get error (Sorry I could not find it) every time when using the same Agent action with the new prompt version. I also tried creating a new custom prompt, but it asks for a Retriever Id as input, even though the prompt already uses the Dynamic Retriever action. Have you encountered this issue before?

1 answer
  1. Jun 11, 5:13 PM

    Hi @Ragini Shukla

     

    Thank you for your reply. I have already tried using the Retriever Id as an input parameter in the custom prompt and created a new agent action to utilize this prompt. However, in this setup, I don't see the purpose of the Data Library, and if the Data Library isn't provided, the source article doesn't appear. Please advise.

0/9000

[Day 2 The Summer '26 governance lever nobody's talking about — controlling autonomous field updates ]

 

Following up on yesterday's post about Multi-Agent Orchestration (and the governance worry a few of you echoed in the comments). 

 

Summer '26 actually ships a direct answer to it: "Control Which Fields Get Autonomous Updates." 

 

The mechanism is the Process Field Update Suggestions Flow. You can modify it to define exactly which fields the Sales Management agent is allowed to touch in your org — so you can let it update next steps and activity dates while keeping it away from Amount, Stage, or anything feeding your forecast. 

 

What I find interesting is that this is arguably the most consequential admin feature in the release, yet it's getting almost none of the attention the headline AI features are. 

 

A couple of things I'm thinking through, and would love this group's input on: 

 

• Are you planning to start restrictive (whitelist a few low-risk fields) and widen over time, or open it up broadly and monitor? 

• How are you handling fields that multiple teams depend on — off-limits to the agent entirely in your org? 

• Has anyone tested the Process Field Update Suggestions Flow in a sandbox yet? Curious how granular the control actually is. 

 

My instinct is to start narrow and earn trust, but I'd genuinely like to hear how others are approaching the automation-vs-control balance here. 

 

Day 2 of my daily series — thanks to everyone who jumped in on Day 1. 

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 ✨  

  

Community Contributed Resources:

For a resource be considered for a future post of The Agentblazer Collection, please share content with us by using the workflow

 

Featured Resources this Week:

 

Have a great day!

0/9000

So I want to use prompt builder to read a text selected Open AI GPT 4 model which uses 1000 tokens to produce result. How will it be billed either by license or usage by using flex credit. 

 

#Agentforce #Salesforce @PromptBuilders

1 answer
  1. Jun 11, 9:09 AM

    Salesforce Prompt Builder is billed based on a combination of licensing and usage (consumption via credits), depending on how it is implemented.

    Key Points

    • Base Access (License-Based) 

      Prompt Builder is included as part of

      Einstein / Agentforce / Data Cloud related licenses. So you need the appropriate license to use the feature.
    • Model Usage (Consumption-Based) 

      When you use external or large language models (such as OpenAI GPT‑4), billing is based on

      usage via Einstein Request Credits or Flex Credits

      This is typically measured using:

      • Number of tokens processed (input + output)
      • Frequency of prompt execution
    • Your Scenario (1000 tokens GPT‑4)
      • The request will consume credits based on tokens processed
      • Both input tokens (your text) and output tokens (generated response) are counted
      • Charges are deducted from your credit allocation, not per user license

    Conclusion

    Prompt Builder requires a license to access the feature, but the actual cost of running prompts (especially with GPT‑4 or external models) is usage-based and charged via credits depending on token consumption

0/9000
Tom Bassett (Senior Solution Architect at Vera Solutions) Forum Ambassador

A few weeks ago at True to the Core (TTTC) during TDX, I asked the Salesforce Team to clarify the pricing for Prompt Builder. 

 

Salesforce Foundations is a free offering that includes Agent Conversations, Data Cloud, and features from Marketing and Commerce. However, despite what Trailhead and Salesforce Pricing pages suggest, it does not include Prompt Builder. 

 

Since raising this at TTTC, I’ve spoken with multiple Salesforce teams, and even they can’t agree—some say Prompt Builder is included, others say it isn’t. In practice, it’s not available in a Salesforce Foundations org, as the required Permission Sets (Prompt Template Manager, Prompt Template User) and Permission Set Licenses (Einstein Prompt Templates) are not provisioned. 

 

Currently, it appears Salesforce intends to require an Einstein for Service, Einstein for Sales, or Einstein for Platform license — effectively making Prompt Builder subject to both user-based and consumption-based licensing. 

 

I appreciate that Salesforce teams have been working on this behind the scenes, and I’ve noticed a shift in messaging. Since TDX, Agentforce’s response on Salesforce Help has changed from ❌ “Prompt Builder is not explicitly mentioned as part of Foundations” to ⚠️ “Prompt Builder is available with Salesforce Foundations, but on a limited basis.”

 

This lack of clarity is a problem for the nonprofit clients I work with, as they rely on the Power of Us Program for courtesy licenses in the first place. 

 

I’d like the Salesforce team to: 

 

1️⃣ Clarify whether Prompt Builder is included in Salesforce Foundations, as the official materials currently contradict each other. 

 

 2️⃣ If Prompt Builder requires per-user licensing (even temporarily), consider extending the Power of Us Program to provide courtesy Prompt Builder licenses alongside the standard courtesy Salesforce licenses. 

 

 3️⃣ Correct all marketing and documentation to clearly reflect whether Prompt Builder is included with Foundations (or not). 

 

At TTTC, it was suggested that Prompt Builder was coming to foundations, but I didn’t get a clear time commitment to the above, so I’m posting here. 

 

Before I raised this at TTTC I verified this was not currently available in two separate orgs both with Salesforce Foundations enabled. 

 

If you need a recap you can watch the recording of this question from 11:27 on Salesforce+

 

#TTTC #True To The Core #Prompt Builder #Power Of Us HUB #AI #Prompts #TDX25

 

@Prompt Builder @* Salesforce Platform * @Trailblazer Community Cove @* Customer Success * @Nonprofit Hub 

17 answers
  1. Mar 17, 2:30 PM

    Thanks @Abby Trout we dont currently have Agentforce or Data Cloud license. I was assuming the prompt Builder being GA would mean it appears in the org without any other products being enabled. But the more I am reading about the licensing it may be that Prompt Builder may come with Salesforce Foundations but we will have to check with our AE on this. Thanks for your help.

0/9000

Something I want you to do before Sunday. 

 

  • Go to your Salesforce org.
  • Pick any major object — Account, Contact, Opportunity.
  • Open a record.

Count the custom fields on the page layout and ask yourself: how many of those fields have actual data in them? 

 

Not "could have data." Actually, have data. Populated. Used. 

 

This Sunday we clean that up. Systematically. 

Salesforce Code Analyzer. Safe Deletion Protocol. Unlocked Packages. 

 

Session 2 — Sunday 14 June — 11 AM Dubai.

Speaking at the @Salesforce Developer Group, Dubai, UAE

 

 [Click Here to RSVP]

 

@Agentblazer Community Group @* Salesforce Developers * #NoClicksJustVibes #DeFraggingTheFrankenOrg

0/9000

🌎 Salesforce is expanding its AI model cards to include environmental impact metrics — helping customers better understand the estimated energy consumption and carbon emissions associated with AI models across their lifecycle. 

 

Model cards have long served as "nutrition labels" for AI, providing transparency into how models work, their intended uses, performance, and guardrails. With the addition of sustainability data, we're taking another step toward making AI systems more transparent and accountable. 

 

The new Environmental Impact section is now available for select Salesforce models and uses the AI Energy Score methodology to help standardize reporting across the industry. 

 

📖 Read the post: Expanding AI Model Cards with Environmental Metrics

0/9000