Document AI

Document AI Automation Setup - Quick Guide

This guide provides a detailed overview of configuring Salesforce Flow Automations with CloudFiles Document AI Flow Actions. You'll learn how to process files using AI and run Natural Language Queries on the processed files to perform further automations. These automations may include creating or updating Salesforce Object records, categorizing files, and executing other simple or complex actions using extracted information.

By leveraging CloudFiles Document AI Flow Actions with Salesforce Flow elements, you can create powerful, automated workflows to streamline document processing and data extraction within Salesforce.

What to Expect

By the end of this tutorial, you can expect to -

  1. Gain an overview of setting up CloudFiles Document AI Flow Actions within Salesforce Flows to process files using AI. This applies to both Salesforce Files and External Storage Files (if using CloudFiles Document Management & Sharing)
  2. Understand how to configure CloudFiles Document AI Flow Actions for post-processing automation. This includes running Natural Language Queries on processed files to extract information and automate actions such as Salesforce record creation, record updates, and other simple or complex workflows using the extracted data.

Pre-requisites

Ensure the following before starting:

  1. CloudFiles Installed in Salesforce: Available on Salesforce AppExchange.Check out Installation and Configuration and to know more.
  2. Active Subscription or Trial: Ensure access to CloudFiles Document AI
  3. CloudFiles Integration: Ensure correct setup if using Document Management & Sharing or Document Generation

Processing Files using CloudFiles Document AI

To use file contents for automation within Salesforce, the files must first be processed using CloudFiles Document AI. Think of it as how employees manually review and extract relevant information from files before taking action, such as updating records or categorizing documents. Similarly, CloudFiles Document AI processes files to make them ready for querying.

The Process Document using AI flow action is essential for processing files and can be leveraged in various ways based on your use case. Some examples include:

  • Processing Order-related Files: Extract details from a Purchase Order uploaded as a Salesforce File on an Order record to update corresponding Salesforce fields.
  • Processing External Storage Files: If using using Document Management & Sharing, files uploaded to an Account's external storage folder can be processed to update Salesforce records, modify external metadata (e.g., SharePoint columns), or trigger further actions like moving files between locations.

Regardless of the use case, every file must be processed at least once. Upon processing, a Document Processed event is generated, providing a unique CloudFiles Processed Document ID—this ID is crucial for subsequent post-processing actions like querying the document for specific data using actions like Query Document

Below are some common scenarios where CloudFiles Document AI can be used. While this list highlights key applications, the functionality can be extended to handle complex document automation workflows:

  • Extracting Invoice or Purchase Orders Data to update Salesforce Opportunity or Order records.
  • Classifying KYC Documents (e.g., Passport, Driving License) and updating Contact records.
  • Reading Contract Details and auto-filling Salesforce fields for legal or compliance workflows.
  • Enhancing External Document Management by integrating extracted insights with metadata for SharePoint, Google Drive, or other storage solutions.

Processing Salesforce Files (Content Documents)

Any Salesforce File (ContentDocument) can be processed using the Process Document using AI action.

All you need is to provide the following input parameters:

  1. Context – A string to specify the context in which the file is being processed (e.g., a Salesforce Object RecordId).
  2. FileId – The Salesforce File’s ContentDocument ID.
  3. Library – Set to "salesforce" for Salesforce Files.
  4. Instructions (Optional) – Any special AI processing instructions for more accurate results.

Example : Processing Salesforce Files (Content Documents) uploaded/attached onto Salesforce Object records.

Refer to Tutorial: Using AI to Process Salesforce Files and Update Records

Any instance of a Salesforce File (a Content Document), as and when uploaded or attached can be identified using the publishing of a CloudFiles Platform event of typeSalesforce File Attached. This event's details can than be used to identify the file/s and be processed using the Process Document using AI flow action.

The details of the setup of such flow automation are given below -

1 - Trigger a flow on Salesforce file Upload/Attachment

  • Create a Record-Triggered Flow on CloudFiles Event Object cldfs__CloudFilesEvent__c
  • Set conditions:
    • Type = salesforce-file-attached
    • Object Type = Salesforce Object's API Name.
  • Include a Run Asynchronously path, in which all further executions shall be done.
Document image


2 - Retrieve Event Details

Add a add a Get Event Details flow action to extract the triggering Salesforce files attached event details.

Document image


3 - Loop through all the Salesforce Files Attached

Loop over the the Attached Salesforce files's i.e. Content Document Records "Content Document IDs", output by the "Get Event Details"action added in step 2.

Document image


4 - Process Each File

  1. Add a Process Document using AI action in the loop to process each file one by one.
  2. Input these 4 parameters
    1. Context = {!Get_Event_Details_Action.SalesforceFiles.ParentId}
    2. FileId = {!For_Every_File_Attcahed}
    3. Library = salesforce
    4. Instructions (Optional) = Optionally, provide any special instructions instructions for specific files processing and what are you willing to query post processing.
Document image


Other Use Cases for Processing Salesforce Files

  1. File Upload via Screen Flow – A file uploader configured in a Screen Flow allows users to upload multiple files and trigger processing. This is useful for cases where files are uploaded manually, and automation is required to extract data for multiple record creation or updates.
  2. Processing Existing Files Attached to Object Records – Any ContentDocumentLink (files already attached to a Salesforce record) can be processed via a button click or an automated trigger. This is useful when users need to select specific files attached to a record and process them to extract data for record updates or further automation.

These examples illustrate a few possibilities. The context in which Salesforce Files are processed depends on your specific use case and the desired automation setup.

Processing External Storage Files

If you are using Document Management & Sharing to manage external storage files, any external file (CloudFiles Resource of "Type" = file) can also be processed using the using Process Document using AI action.

The required input parameters are:

  1. Context – A string to specify the processing context (e.g., a Salesforce Object RecordId).
  2. Library – The name of the external storage library (e.g., sharepoint, google , onedrive etc.).
  3. DriveId – The DriveId of the external file.
  4. FileId – The ResourceID of the external file.
  5. Instructions (Optional) – Additional AI instructions for enhanced processing results.

Example : Processing of External Storage Files uploaded into folders from CloudFiles Widget on Record pages

Refer to Tutorial: Using AI to Process External Storage Files and Update Records

Any file uploaded into connected external storage via the CloudFiles LWC on record pages can be identified and trigger further automation through the publication of a CloudFiles Platform Event of type File Uploaded. This event provides file details, which can then be used to identify the uploaded file and process it directly with CloudFiles Document AI—without storing it in Salesforce—using the Process Document using AI flow action

For instance, if you have set up flows to automatically create and attach folders for object records like Accounts, Cases, Orders, or any other custom object, and have placed the CloudFiles LWC on the record details page, every file uploaded into these folders can be processed automatically.

1 - Trigger Flow on external File Upload via CloudFiles LWC

  • Create a Record-Triggered Flow on CloudFiles Event Object (API Name: cldfs__CloudFilesEvent__c )
  • Set conditions as "Type" field = file-uploaded .
  • Include a Run Asynchronously path, in which all further executions shall be done.
Document image


2 - Retrieve Event Details

In the "Run Asynchronously" path, add a Get Event Details flow action to output the triggering File Uploaded event details.

Document image


3 - Filter Events (Optional)

To optimize resource usage and minimize unnecessary flow executions, you can filter event types and determine whether to process a file based on specific criteria.

For example, to process only files uploaded into an Account record’s attached folder via the CloudFiles LWC on the Account record page:

  • Use the Get Event Details action’s output.
  • Set a Decision Element condition: {!Get_Event_Details_Action.FileUploaded.RecordId} starts with 001 (which represents Account records in Salesforce).
  • Further refine execution by adding conditions based on UserId (to restrict processing to specific users) or other relevant attributes.



Document image


4 - Process the Uploaded File

  1. Add a Process Document using AI action
  2. Input paramters
    1. Context = {!Get_Event_Details_Action.FileUploaded.RecordId}
    2. DriveId = {!Get_Event_Details_Action.FileUploaded.DriveId}
    3. FileId = {!Get_Event_Details_Action.FileUploaded.Id}
    4. Library = {!Get_Event_Details_Action.FileUploaded.Library}
    5. Instructions (Optional) = Optionally, provide any special instructions instructions for specific files processing and what are you willing to query post processing.
Document image




Other Use Cases for Processing External Storage Files

  1. Manual File Selection & Processing – Users can manually select files from a CloudFiles LWC widget and submit them for processing. This can be configured in a Screen Flow or triggered directly from the CloudFiles Widget
  2. Processing on File Attachment to a Salesforce Record – External storage files can be processed automatically whenever a file is attached to a Salesforce record. This can be achieved using the Resource Attached event as a trigger, with the event’s details passed as inputs to the the Process Document using AI action.
  3. Processing on File Upload via Data Room Request – When a user uploads a file via a Data Room File Upload Request, it can be processed automatically. The File Received event can serve as a trigger, providing the necessary details for the Process Document using AI action.

These are just a few common use cases, but the processing External Storage Files can be customized based on specific business needs and automation configurations.

Document Processed CloudFiles Event

This event is published once a file is processed by CloudFiles Document AI using Process Document using AI action. This event can further be used as a trigger to initiate downstream processes or workflows containing actions like Query Document or Query Document (Batch).

How To Use This Event ?

You must first setup your platform event triggered flow as shown in the Platform Event Mode Flow Setup. Please use document-processed as the type filter when selecting this event in the decision element in flow builder.

Document image


Output Fields From Get Event Details

Refer to the image below for a list of outputs-



Document image


1 - ProcessedDocumentID

The Processed Document ID is a unique CloudFiles identifier that represents the data of a processed document. The ProcessedDocumentID can be reused as often as needed to reference the document for various CloudFiles Document AI actions.

The ProcessedDocumentID serves as a input parameter for actions like Query Document to query the processed file.

Document image


2 - Context

The context value specified in the input, which helps identify the event’s source or any other intended details.

3 - File

Details of the processed file specified in the input, which helps identify the file post processing.

Document image


Querying Processed Files

The Query Document and Query Document (Batch) flow actions enables to execute Natural Language Prompts on the data of the documents (that have already been processed using Process Document using AI action) identified by ProcessedDocumentID.

Document image




For example, these actions allows you to execute a query (a natural language or English-language prompt) on the data of a processed document. These actions processes the input query or queries and returns the result as a text output.

Imagine you are processing KYC files uploaded to Contact object records. Using these actions, you can:

  1. Classify the uploaded document as a Driving License, Passport, or neither through a simple Classification Query like"Check the type of file. If it's a Driving License, return 1; if it's a Passport, return 2; if neither, return 0."
  2. Using the query result above, the flow can be branched into specific paths for each document type, and execute firther queries to extract relevant information, such as the passport number, from the document. For example, if the classification query returns 2 (Passport), the flow enters the "Is Passport" path and runs queries like:
    1. "Extract the passport number."
    2. "Extract the name on the passport."
    3. "Extract the issue date in the MM-DD-YYYY Format"
    4. "Extract the expiry date in the MM-DD-YYYY Format"
Document image


By using the returned results of these actions within a flow, you can build complex workflows that analyze and extract data from documents efficiently and further update create records or run further automations.

Inputs

Let us discuss the input parameters in brief, in the sections below.

Processed Document Id

Input the Processed Document Id of the file on which you wish to execute the Query.

Query

Input the Query (as a plain text) specifying the prompt you wish to execute on the file identified by the input Processed Document Id.

You can directly pass a string as the query (e.g., "Extract the passport holder's name."). Alternatively, you can pass the query dynamically through a variable, formula etc. allowing to incorporate complex use cases.

Using AI for General Queries

Not all queries need to be executed on a specific document. If you want to run a general AI-based query that does not require document data, you can simply omit the Processed Document ID when executing the query.

This allows CloudFiles AI to process the query independently, focusing only on the text input you provide. This can be useful in scenarios where:

  • You have pre-extracted data from documents and need AI to structure or analyze it further.
  • You need AI to generate text-based responses, summaries, or formatted outputs based on user-defined parameters.

The effectiveness of the CloudFiles: Query Document action heavily depends on the quality of the query or input prompt.

Testing Queries in the Playground:

  1. Navigate to CloudFiles App > Document AI > Playground.
  2. Upload a sample file.
  3. Use the interactive Query side-component to experiment with different prompts against the file.
Document image


This iterative process helps you fine-tune your queries for optimal results before implementing them in your workflows.

Outputs

The output of the CloudFiles: Query Document action is a Text (String) data type containing the result of the query as defined in the query prompt.

How to Use the Outputs?

The extracted text can be utilized in numerous ways, depending on your automation needs:

  1. Creating New Salesforce records
  2. Populating Salesforce fields
  3. Triggering notifications
  4. Automating approval processes
  5. Integrating with other processes or applications. For example, update of external file's metadata based on info in the files.

Since the possibilities are endless, you can customize your workflow to maximize efficiency and automation.

Detailed Tutorials

Check Tutorials on some of the use cases:

  1. Using AI to Process Salesforce Files and Update Records
  2. Using AI to Process External Storage Files and Update Records



PREVIOUS
User Management
NEXT
Using AI to Process Salesforce Files and Update Records
Docs powered by Archbee