* Salesforce Einstein *

I have included in my Custom Agent instructions to output the response (list of quote lines based on a search query) in a table format. It used to work perfectly until the Spring 25 release. No matter how I tweak the instructions, it's not working anymore. It outputs in JSON format (even though I had it in my instructions to never output in JSON format by default). When I follow up with a second prompt to output it in a table format, it does do it, but I need it to output it in table format by default without the user having to prompt it to.     I checked out this video (https://www.youtube.com/watch?v=oxHNzaCmEfE) and it also shows the user prompting it to output in table format after the initial response.     Anyone have a similar scenario and is able to get the agent to output in table format by default? 

Hi, 

I was hoping to better understand the score_correlation on the SalesAIScoreModelFactor object, which seems to determine feature importance of Einstein opportunity scoring models. Based on the 

documentation provided, is my understanding correct that this score is a correlation between a single factor's values (e.g. the opportunity amount) and the predicted opportunity score on the training or testing set used by Einstein? If my understanding is incorrect, could you please clarify how the correlation is made? 

@* Salesforce Einstein * 

Hi, 

I'm using EAC via Gmail and I am aware in the chrome app there is a feed of activity...

My question is, is it possible to have that feed in Salesforce like as a user notification?

My instinct is to either build a flow or use a 3rd party tool but I find it hard to believe SF doesn't have such a simple feature so asking here...

Hi! 

Starting to work with Agentforce in a sandbox, but I when trying to create a new agent, I only see one type: 

Shouldn't there be several types of agents to choose from? 

What am I missing? 

In the sandbox I am systemadmin and have the permission sets Customize Application, Agentforce Default Admin and Access Agentforce Default Agent ... 

/ Dan Sletten

Hello! 

 Seeking your valuable input.

I have enabled EAC for the events capture in Salesforce and Google calendar. The sync is working fine with desktop version of calendar and SF events. 

 However the events created in Google Calendar mobile app is not showing up on the SF mobile app events calendar. Am I missing something here? Please suggest

Thanks!!

⭐️ NEW: Chat with Agentforce Service Agent in More Languages (GA) ⭐️ 

In addition to English and Japanese, French, German, Italian, Portuguese, and Spanish are now generally available for Agentforce Service Agent in certain locales. 

➡️ See: Release Note and Help 

Hello! 

We tried activating EAC Metrics in our Production Org and discovered we could not. 

We opened a case with Salesforce Support and they explained that EAC metrics only works with Orgs with 1.5 million records or less.  Our Org has way more record than that. 

SF Support shared an IdeaExchange to request SF increase the record amount and in that post they indicated you could request SF support enable the ability to run EAC Metrics with more than the 1.5 million records but my fear is that it would greatly impact performance. 

I looked back in research that included Salesforce documentation, partner videos and this limitation is never mentioned in any of them. 

Did I miss where it is explicitly called out in formal Salesforce documentation?  Is there a central point with ALL EAC information?  Such as one presentation on EAC it was show that the EAC to Office 365 can be updated to the Graph API but I have not heard anything since. 

Thank you!

⭐️ NEW: Use More Languages with Prompt Builder ⭐️ 

In addition to the languages fully supported by the Einstein Trust Layer, you can use other languages supported by Salesforce. However, the Trust Layer features might not be available for the added languages. 

➡️ See: Release Note and Help 

Who is going to manage Agents? Is HR or IT going to be AR (Agent Resources) 

If we think of agents as digital labor, they go through the same steps as humans 

• Train (build agents, clean data, update knowledge) 

• Onboard (deploy into production and integrate with human labor processes) 

• Coach (update if improvements are identified or for regulatory non-conformance) 

• Monitor (audit and governance of changes) 

• Salary (pay agent platform which requires tracking) 

• Promote (add new skills to broaden the power of the agent) 

• Demote (remove skills from the agent if they are not performing) 

• Off sick (take down if the agent goes rogue) 

• Sack (supersede by newer agent) 

This blog dives into more detail 

⭐️ NEW: Einstein Service Replies Trailhead Module ⭐️ 

Learn how to enhance service efficiency with Einstein Service Replies. 

➡️ Earn a Badge: Einstein Service Replies