The Modern AI Agent Stack

 

RAG :  Pulls in outside knowledge before answering.

Context : Everything the model can see right now.

Memory: What it remembers across turns and sessions.

Agent : An AI that acts, not just answers.

Agentic AI : System that plan, decide and self -correct.

Tools : External functions the agent can call.

MCP : One open standard to connect tools and data.

Skills : Reusable know - how loaded on demand.

Hooks : Triggers that run logic at set points.

Subagents : Helpers the main agent can delegate work to.

Orchestration : Coordinates the whole multi-step workflow.

Eval : Measures whether the system actually works.

Winter ’27 Apex Heap Limit

 Salesforce is increasing Apex heap limits, giving developers more flexibility when working with larger datasets and complex processing.

 Synchronous Apex: 6 MB → 10 MB

Asynchronous Apex: 12 MB → 25 MB

Salesforce Elastic Limit for Async Apex

 Salesforce introduced Elastic Limits for Async Apex (Beta).

Here’s what it does:

→ Your org can now handle up to 500,000 async jobs per day
→ Jobs after 250k will still run, but at a slower speed
→ No sudden failures when the limit is crossed
→ You can now see live usage directly on the Apex Jobs page
→ Developers can also monitor limits using System.OrgLimits.getMap()
→ Available for Enterprise, Performance, Unlimited, and Developer editions

How to enable it?

→ Go to Setup​
→ Search for Apex Settings​
→ Enable “Use elastic limits for asynchronous Apex jobs (Beta)”

That’s it.

Once enabled, the Apex Jobs page will show:

→ Current 24-hour usage
→ Daily limit
→ Elastic limit
→ Whether throttling is active

No more guessing.
No more surprise failures in the middle of the night.

This is a really useful update for orgs handling large-scale automation and integrations.

Cleaner Apex Code with Multiline Strings and Using String.template() -

 The new Multiline Strings and String.template() functionality in Salesforce Summer ’26 is a major quality-of-life improvement for Apex developers.

• Create multiline strings using triple single quotes (''')
• Perform named variable interpolation using String.template()
• Avoid repetitive string concatenation
• Replace index-based placeholders from String.format()

Ex :

String formatted = '''

{

    "Account": "${accountName}",

    "Last Updated": "${date}"

}

'''.template(new Map<String, Object>{

    'accountName' => 'My Account',

    'date' => DateTime.now()

});

System.debug(formatted);

Agentforce reasoning engine

 Most Salesforce developers don't know that Agentforce has a sophisticated reasoning engine that decides which action to call, and understanding how it works can greatly improve your automation workflows.

The reasoning engine consists of three main components: the planner, the action selector, and the response synthesizer. The planner determines the overall strategy for the conversation, the action selector chooses the specific action to take, and the response synthesizer generates the response to the user. 

In a real-world scenario, this can be useful for automating customer support workflows, where the Agentforce reasoning engine can determine the best course of action based on the customer's input. For example, you can configure the action selector to call a specific Apex method, such as "handleCustomerComplaint", by setting the "actionName" parameter to "handleCustomerComplaint".

Salesforce Revenue Cloud Advanced / Agentforce Revenue Management Tips

 1.Workflow Services :

It’s a handy salesforce feature that lets you track and manage automated processes running behind the scenes - like Data Processing Engine (DPE) runs, Batch jobs, Billing Scheduler, data syncs or even decision table refreshes.you’ll find it in setup under workflow services.

2.AppUsageAssignment :

AppUsageAssignment provides application context for a record - meaning salesforce can decide  what actions are allowed and which related objects should be created based on that context.

3.Decision Table refreshes

1. Record-Triggered Flow - Refresh a Decision Table automatically when a related record changes.
2. Screen Flow (Manually Refresh) - Display Decision Tables in a Data Table component and let admin users select which ones to refresh.
3. Scheduled Flow - Run Periodic refreshes at defined intervals (daily, weekly etc)

Pro tip :

 Leverage the Incremental Refresh option in the Apex action to improve performance - especially helpful for larger tables.

4.“Is Price Impacting” on Product Classification Attribute :

Is Price Impacting allows us to adjust the price based on selection of the attribute value later on.

What does “ Price Impacting” RCA?

When an attribute is marked as price impacting , it tells Salesforce to adjust the product price automatically based on the selected value of that attribute .

Why using price impacting attributes ?

1. To automate pricing based on product options. To reduce quoting mistakes and manual calculations.
2. To Simplify the sales process - sales reps just choose options, and the system calculates pricing dynamically.

5.Product Won’t Appear During Quoting :

1. Check the Product details

  —> is it active?

  —> Is it assigned to the correct Product Selling Model?

  —> Does it have a valid price book entry for the corresponding product selling model in the right Price Book?

     2. Verify the Product Catalog

—> Is the product is added to a category and is the category added to the right catalog?

—> Has the index has been rebuilt after the product was added to the category and catalog?

    3. Make sure the Product Qualification works

  —> If product qualification/disqualification rules apply, are the product qualification/disqualification records are accurate and active?

—> Do the product qualification/disqualification look up tables have the right filters and conditions?

—> Does the qualification procedure return the right results? You can try simulating the procedure to verify the output.

  4. Check the Set UP

  —> Does the product discovery settings refer to the right context definition and discovery procedure?

—> Have the decision tables been refreshed?

  5. Verify the Quote details

—> Is the price book selected on the quote the right one?

—> Do the qualification criteria during quoting match the qualification records created for the product?

6.Pricing Procedures :

Pricing Procedures are the heart of the Revenue Cloud pricing engine. They define every step that transforms a simple List Price into a final Net Price. Once you understand how they work, you gain real control over your entire Quote-to-Cash flow.

The Pillars of a Pricing Procedure

1. Context Definition: This is your data map. Context Tags pull values from your Salesforce objects into the pricing engine. If your context is wrong or incomplete, your entire procedure will misfire - so this is the first place to get right. 

2. Pricing Elements: These are the individual steps that make up your calculation logic. List Price Element, Volume Discount Element, Formula Element, Rating Lookups, and more.Each element performs one task- calculate something, fetch something, or transform something.

3 The Sequence (The Price Waterfall): This is the execution order of your Pricing Elements. One element’s output becomes the input for the next, so the sequence determines how your price waterfall unfolds. A misplaced element can completely change the final price.

Pro Tip: Use Price Logs to Verify Your Pricing Flow

One of the least known debugging tools in Revenue Cloud is Price Logs. They show exactly how your Pricing Procedure executed - step by step.

How to use them:

1. Go to Setup → Salesforce Pricing Setup and turn Price Logs Capture ON.

2. Open the Pricing Operations Console app.

3. Navigate to Pricing API Executions to view logs for each line item.

You’ll be able to see:

- Inputs passed into each Pricing Element

-Outputs generated

- The exact sequence the procedure followed

Perfect for validating logic, troubleshooting incorrect prices, and refining complex waterfalls.

Product Categories and Attributes

 A ) Category – Groups products inside a catalog and improves product discovery. Supports parent/child hierarchy for deeper organization.

B) Subcategory – A child category used to refine product grouping. Revenue Cloud Advanced supports up to five nested levels.

C) Picklist – A data record in Revenue Cloud Advanced that defines the allowed selectable values for an attribute.

D) Picklist Values – The individual options inside a picklist (for example, 128GB or 256GB), stored as AttributePicklistValue records.

E) Attribute Definition – Defines a product characteristic with properties like data type, required/optional, display type, price-impacting, and more.

 This is the master definition reused across products.

D) Attribute Category – A logical grouping of related attributes (such as Storage Specs or Display Specs) to streamline configuration and reuse.

E) Assign Attribute to Attribute Category – Links attributes to categories using AttributeCategoryAttribute, making them usable in product classifications and product setup.