Skip to main content

AI Mass Processing with GPTfy

Use GPTfy’s AI Mass Processing to run prompt-based automation across a large set of records — on a schedule or on demand. This guide walks you through configuring schedulers, setting up filters, and executing batch runs.

Step 1: Create an AI Mass Processing Scheduler​

Start by creating a new scheduler that controls when and how your prompts run.

How to create it:

  • From the Cockpit, select the AI Mass Processing tile.
  • Click the New button to begin.
  • In the setup screen, fill in the details:
    • Name: A unique, descriptive name for your scheduler
    • Scheduling Frequency: Choose how often the scheduler should run (e.g., daily, weekly)
    • Schedule Time: Pick the time you want it to trigger

alt text

Note: As you configure these options, GPTfy automatically generates a Cron expression based on your selected frequency and time.

Step 2: Define Prompt Scheduler and Filtering Logic​

Each AI Mass Processing record contains one or more prompt schedulers. These define which records to process and what prompts to run.

To create a prompt scheduler:

  • Click New to add a prompt scheduler under your AI Mass Processing record.
  • Provide the following details:

alt text

Basic Filtering​

  • Where Clause: Apply conditions to narrow down records (e.g., Status = 'Open'). You can use the Query Builder (pencil icon) to create this without writing SOQL.
  • Record Limit: Default is 2 — controls how many records run per batch.

alt text

Cross Object Filters (COFi)​

COFi filters qualify parent records based on conditions in related child records.
Example: To process an Account only if it has a related Contact with Department = 'Banking', define a COFi filter on Contact.

Each COFi filter includes:

  • Sequence: Determines the order filters run (lower numbers run first)
  • Operator: Defines the logical operator (e.g., AND, OR)
  • Qualify Count: Minimum number of conditions that must pass
  • Where Clause: Filtering logic for the child object

alt text

Note: COFi filters act like guardrails — only qualifying parent records where child data matches your criteria.

alt text

Step 3: Run or Schedule the Batch Process​

Once everything is set up, you can either run the process immediately or let it follow the scheduled time.

Available actions:

  • Run Now: Click this to trigger the batch process instantly

  • Schedule: Activates your scheduler based on the configured frequency

  • Unschedule: Cancels the scheduler if you no longer want it running

alt text

alt text

Running a batch will trigger an Apex job, which creates a Scheduled Run Record linked to your prompt scheduler.

Monitor Run Metrics​

Each scheduled run includes detailed metrics on what happened during processing.
To view them, go to the Scheduled Run record’s Details tab.

alt text

Metrics available:

  • Total Records – Total number of records qualified
  • In Progress Records – Records currently being processed
  • Completed Records – Successfully processed records
  • Errored Records – Records that failed (e.g., due to token limit)
  • Disqualified Records – Records blocked by visibility rules or filters
  • Processed Records – Sum of completed and disqualified records

alt text

alt text

Handling Errors with Retry Logic​

Records that fail during processing are marked as Errored.
GPTfy automatically retries errored records based on your configured Retry Count.

Once a retry succeeds, the error status is cleared.

Example: If a record failed due to exceeding token limits, it will be reprocessed during the next attempt — as long as retry attempts remain.