In Salesforce Flow, the Decision element determines which path the automation takes based on the conditions you set. Until now, there was only one option to set these conditions manually using field values and formulas. In the Winter '26 release, Salesforce has taken this a step further with the new "Let AI Determine Conditions" option. This powerful enhancement allows you to provide instructions for AI to dynamically evaluate your automation’s data and decide which outcome to run. In other words, you can now let AI handle the decision-making for you. This brings intelligence and flexibility directly into your Flow logic.
What is "Let AI Determine Conditions" Option in Decision Element?
In the Winter ’26 release, Salesforce added a new level of intelligence to Flow by allowing generative AI to power Decision elements. Instead of building complex condition logic manually, you can now use natural language instructions to tell the AI how to analyze your data and decide which path to take. This capability is especially useful when working with unstructured or descriptive data (e.g. free text, sentiment analysis, descriptive fields). This is where traditional operators like equals, greater than, or contains fall short.
How to Use the "Let AI Determine Conditions" Option in Decision Element
When configuring a Decision element in Flow, you can now choose how the automation determines which outcome to run. You’ll see two options: Manually Set Conditions and Let AI Determine Conditions.
With Manually Set Conditions, you define your own logic based on explicitly defined data. This is the traditional method that Flow builders are familiar with. The new Let AI Determine Conditions option (available with an appropriate AI license) allows you to provide instructions for AI to dynamically evaluate your automation’s data and decide which outcome to run. Simply select this option, enter your instructions, and let Salesforce’s AI determine the most appropriate path for your automation.
Entering Instructions to the Decision Element
When using Let AI Determine Conditions option, you will see two separate instruction areas. Each one serves a different purpose.
Element Level Instructions
This is where you describe the overall decision the element is trying to make.
Think of it as the main objective. It is what you want the AI to figure out based on your automation’s data.
You can include context and reference Flow resources to help AI understand the data it’s analyzing.
AI evaluates these instructions at runtime to determine which outcome fits best.
Outcome Level Instructions
These are specific guidelines for each possible outcome. They help AI understand when to select a particular path. Each outcome instruction acts as a rule or hint that supports the main decision goal defined above.
Example
Here is a Decision element configuration that determines which path the automation takes based on what the Case record is about. Since traditional operators can’t interpret natural language descriptions, we’ll let AI determine the conditions instead.
Here are the element level instructions for this Decision element:
Analyze the case to understand what type of issue it describes. Use the Case Description and Subject fields to decide whether it’s a financial issue, a technical issue, or a general inquiry. Choose the outcome that best represents the main problem described by the customer.
Case Subject: {!Get_Case.Subject}
Case Description: {!Get_Case.Description}
Here is the instruction for the "Financial Problem" outcome:
Choose this outcome if the description mentions payment failures, billing errors, refund requests, or anything related to invoices or account charges.
As you can see, when you run or debug the Flow, AI analyzes the Case record’s details and executes the most relevant outcome.
Limitations of Let AI Determine Conditions Option
The Let AI Determine Conditions feature isn’t available in every environment. It’s available only in Lightning Experience for Enterprise, Performance, and Unlimited Editions that have the Einstein for Sales, Einstein for Service, or Einstein Platform add-on. Additionally, each AI-driven decision consumes Einstein credits, so be sure to monitor your usage in your Digital Wallet to manage costs effectively. Moreover, as of the Winter ’26 release, it’s not available in record-triggered flows.
Leave a Reply