Intro

Flow Orchestrator takes a wide range of rich, sophisticated, and complex business process use cases that require code today and makes them easy to create with clicks. It makes it possible to automate multi user processes so I try to verify if Flow Orchestrator replaces the standard approval process.

Use case

A SaaS company provides various products with their customers. They give free trials for their products to their prospects and manage this process using a custom object called “Free Trial”, which contains Start Date of trial, End Date, Products, and related Opportunity.

The Sales VP would like all Free Trial requests to be routed controlled by an Approval Route table which is a custom object and it to be routed with serial and parallel mode.

Functionality

• Send an email to approvers
• Lock & unlock a record
• Three Approval Steps
  • Serial mode : same as standard approval process
  • Parallel mode
    • Majority Wins
    • CEO Overrides where a particular approver can override the majority
• Approval History 
• Approval Route
• Custom Notification for only parallel mode

Demo Video

Flow

AI Approver Screen

This is a screen flow where an approver decide to approve or reject and return result parameter (approved is 1 otherwise 0)

AI Requester Screen

This screen flow receives is_approved (boolean)  and recordId (string) as input parameters and update a Free Trial record’ status based on is_approved value. In case it is rejected, Apex action for unlocking the record is executed. Finally it shows whether the approval process is accepted or rejected to a requester. This implementation looks weird to you because the status of the record is updated after the requester executes this screen flow. Yes, it does. The current Flow orchestrator supports only screen flow. We can not hook up an event where three approvers complete their decisions for parallel mode (send approval request to approvers simultaneously) Do not worry about it. Flow orchestrator will support Autolaunched Flow soon. Once it comes true, we can get rid of updating and unlock record logics from this screen flow.

AI Submit Approval Process Action

Custom action of Free Trial record calls this screen flow. After requester completes to submit for approval process, Submit_Approval__c of the target Free Trial record is set to true. This triggers AI Approval Process Orchestration.

AI Parallel Approval Decision Flow

One of my favorite features of Flow orchestrator is that we can customize how a step or a stage will be triggered or considered complete. We can invoke an evaluation flow to evaluate the entry condition of a stage or step. This flow is an evaluation flow for parallel mode. An evaluation flow must be Autolaunched Flow and returns isOrchestrationConditionMet parameter (boolean). Since autolaunched flow can query objects and make API calls via Apex , execute logic, and so on, you can make complicated decision logic.

The flow receives parallel decision mode (Majority wins or CER overrides), role names for three approvers, three approver’s decisions, and “from approved stage” ( boolean value, from approved stage or rejected stage) as input parameters and returns isOrchestrationConditionMet boolean value. The current Flow orchestrator does not support evaluation flow within the Decision element so I need “from approved stage” input parameter so that only proper step is triggered. I will jot about this at AI Approval Process Orchestration section below.

AI Approval Process Orchestration

It is Record-Triggered Orchestration where object is Free Trial and condition is Submit_Approval__c Equals true. It orchestrates AI Parallel Approval Decision Flow, AI Submit Approval Process Action, AI Requester Screen, and AI Approver Screen.

Serial mode

Parallel mode

As I mentioned necessity of “from approved stage” input parameter at AI Parallel Approval Decision Flow section. I want to execute either a Work step named “Approved” or “Reject”. For instance, AI Parallel Approval Decision Flow returns true at Work step named Approved and do false at Work step named Rejected under approved Free Trial record. 

Not implemented

• Recall approval process
• Assign Queue
  • This can be implemented using Select Id from Group where type=’Queue’ and Name=’Queue Name’ and Select UserOrGroupId From GroupMember where GroupId =:Id  but complicated
• Custom Notification for serial mode
  • Supporting Autolaunched Flow will fix easily

Conclusion

I have confidence in replacing the approval process with Flow Orchestrator because I believe that new functionalities related to the approval process will be available at GA. My implementation of the approval process is a little bit complicated because of lacking standard functions such as supporting Autolaunched Flow, recall approval process (back to previous stage), standard actions (lock & unlock record, get queue member) but I am very excited to continue to watch new features. 

You want to try?

Prerequisite

• Unlimited, Enterprise, Developer Editions have Flow orchestrator enabled

Package Installation 

1. Check https://youtu.be/DhnkMsQwImA at youtube for install package
2. Go to https://sforce.co/3qEh40g 
   1. Enter your credentials for your org
   2. Select “Install for All Users” at Install AI Flow Orchestration Approval Process screen and hit Install button

Post Installation

1. Setup → Process Automation → Process Automation Settings
   1. Check “Allows Apex code to set and remove approval process locks. Enable record locking and unlocking in” checkbox
   2. Hit Save button
2. Setup → Users → Profiles
   1. Search “Demo Salesforce Platform” profile
   2. Hit edit link
   3. Check the both “Run Flows” and “Send Custom Notifications” checkboxes
   4. Hit Save button
3. Setup → User Interface → Path Settings
   1. Hit “Enable” button
4. Create three “Salesforce Platform” user license users with “Demo Salesforce Platform” profile
   1. By Hand
      1. Role must be specified and one of three must be CEO role
      2. 

2. By Apex code
   1. Open developer console
   2. Debug→ Open Execute Anonymous Window
   3. Aki.createUsers(‘YOUR_REAL_EMAIL_ADDRESS’); and Hit Execute button
   4. You will receives password reset emails for three users at YOUR_REAL_EMAIL_ADDRESS.
   5. 

5. Select “Approval App” at App Launcher
6. Select “Approval Route” tab
   1. Hit New button
   2. Create a Approval Route record
      1. Approval Route Name : High Priority
      2. Approver1 : one of users created at post installation step
      3. Approver2 : one of users created at post installation step
      4. Approver3 : System Admin user

Demo Steps

Requester

1. Login in as the user which is not in the Approval Route record you defined
2. Select Approval App at App Launcher
3. Select Free Trial Tab
   1. Hit New button
   2. Fulfill fields
      1. Opportunity in the form is a custom object because  “Salesforce Platform” user license user can not access standard Opportunity
   3. Hit Save button
4. Hit Approval Process Action
   1. Fulfill all fields because I do not implement proper error handling ;(
5. Check if the record is locked

Approver

1. Login in as the user based on the selected Approval route record
2. Select Approval App at App Launcher
3. Select Free Trial Tab
4. Select the record
5. Enter your decision at the Work Guide

# The custom notification is not implemented for serial mode and you need to refresh a Free Trial detail page to change status after requester confirms at Work Guide

Uninstall Package

Since the uninstall package contains flow orchestrator, you need to delete some components manually first.

Please check https://youtu.be/mwIVAZb9pVA at youtube for uninstall the package

This new Datatable release includes two new features and a few minor bug fixes.

Because the outputEditedRows attribute is not Null even if no rows were edited nor included in the output collection variable, I’ve added a new output attribute numberOfRowsEdited that can be used to check if any rows were edited. I’ve also updated the component to take advantage of the new Automatic Output Variables feature so outputs from one Datatable can now be specified as inputs to another Datatable in the same Flow without have to manually assign the first Datatable’s outputs to a variable.

Updates:

• New Output Attribute for the Number of Rows Edited (Because even when no rows are edited, the OutputEditedRows attribute is not null)
• The Datatable CPE now supports Automatic Output Variables in the Flow Builder

Bug Fixes:

• Orgs with multi-currency enabled can now add currency rollup and currency formula fields to the datatable
• The edit picklist dropdown will overflow the displayed table if necessary (Only if the Table Height attribute is not set)
• The dropdown picklist values when editing will show the picklist labels instead of the picklist API names (The selected edit prior to selecting Save will show as the API name)

Documentation Updates:

The Troubleshooting & Restrictions section of the documentation page has been updated as well.

• The datatable will only display the first 1000 records in the input collection. A collection passed in with more than 1000 records could cause an Apex CPU time limit exceeded error.
• When filtering rows in a datatable, any previously selected rows that are not part of the resulting filtered rows will be deselected.
• Some data types cannot yet be edited in a datable (lookup, location, time, encrypted, rich text, long text area).
• In order for a lookup or master/detail field to display as a link, it must be “reparentable”
• In order to have the Datatable display a Lookup field as a link to the record rather than just the recordId value in the field, the running User must have Edit access to the object whose records are being displayed in the Datatable.  For example, if you have a Datatable for Contact records and include the AccountId(Lookup) field, the running User needs to have Edit access for the Contact object in order for the AccountId field to show as a clickable link with the Account’s Name.  Without Edit access to the Contact record, the Datatable will display, as unclickable, the Account’s recordId value. For information on how you can temporarily add then remove Edit access in a Flow, read the referenced Help Article and/or complete the referenced Trailhead.
  Help Article: Create a Flow That Can Activate or Deactivate a Session-Based Permission Set
  Trailhead: Session-Based Permission Sets and Security

• With the update to selecting a record type for picklist field edits, the running User must have Read Access to the object being displayed in the Datatable. If they do not, any Record Type Id attribute will be ignored and all Picklist values will be available when editing a Picklist field.
• When editing Percent fields you must enter the true decimal value. Example: .25 for 25%
  • 2 fewer decimal places will be available when editing than what has been defined for the percent field
• If you are selecting an Object that contains Geolocation fields (ie: Contact), you may experience a Salesforce Bug where the Flow will stop just before the Datatable should be displayed or the Flow will display an error when clicking Next and show another copy of the same datatable. If this happens, unselect “Automatically store all fields” in your Get Records and choose your fields manually.
• If you are editing, then updating records that contain a non-custom integer field (Account: NumberOfEmployees) you may experience a Salesforce Bug and see this error:
  • This error occurred when the flow tried to update records: Argument must be a big decimal java.lang.Long.
• If you see this error when trying to Configure Columns, try going to Setup > Security > Session Settings and unchecking “Enable clickjack protection for customer Visualforce pages with headers disabled“. Also, in your Setup > User settings, make sure both Debug Mode and Development Mode are Unchecked.

This new version of the converter (the original is here) has been updated to convert more kinds of Process Builder workflows to Flows, leveraging features added to Flow in the last 2 releases. It also now converts Workflow Rules.

Important Notes

1. ConvertToFlow creates a clone of your process and deploys it as a Record-Triggered Autolaunched Flow. This flow shows up in the Flow List like any other flow. ConvertToFlow does not modify or deactivate the process builder process or workflow rule that it’s cloning, so when you activate the new flow, keep in mind that if you don’t deactivate the original, you’ll have two identical processes firing on the same event. We did this for safety, but you really need to keep this in mind. If there’s demand, we can create an option to delete or auto-deactivate the original process so you never have two of them.
2. The newly created flow has an api Name that has ‘converted_’ prepended to it, and it has the same label as the original process:

Looking at a Typical Conversion

Let’s take a look at what a conversion looks like. Here’s a simple PB process:

Here’s what the converted version looks like:

Let’s look at some of the improvements on the Process Builder side. Since the original version was released, Flow has added a bunch of matching features.

ConvertToFlow converts processes that have the “Only Specified Changes” checkbox checked:

The setting of the checkbox is represented in Flow Builder here:

ConvertToFlow converts processes that are using related record references

Here’s an example of a process that has a reference to a related record:

Here’s the converted version:

ConvertToFlow converts processes that use Scheduled Actions

If the process has no more than 1 scheduled action per decision ‘row’, ConvertToFlow can convert it. Here’s an example of a process with 2 scheduled paths:

Here’s what it looks like after conversion:

Converting Workflow Rules

This update to ConvertToFlow adds the ability to create Flows from Workflow Rules. Most WFR are supported, although Flow doesn’t support Outbound Messages yet and is not expected to support Recursion. Here’s a video showing some conversions:

Support for Process Builder Processes that Launch Flows

The latest update of Convert To Flow adds support for Process Builder processes that launch flows, converting them to Triggered Flows that use a Subflow.

NOTE: this requires a Winter ’22 Org.

The Newly Generated Flow Will Use Before-Save Triggers If Possible

If the source process or rule does a straightforward field update, a high-performance Before-Save trigger will be used. If the source process does something that’s not allowed Before the Save, an After-Save trigger will be used.

Important Notes

1. ConvertToFlow creates a clone of your processes and rules and deploys it as a Record-Triggered Autolaunched Flow. This flow shows up in the Flow List like any other flow. ConvertToFlow does not modify or deactivate the process builder process or workflow rule that it’s cloning, so when you activate the new flow, keep in mind that if you don’t deactivate the original, you’ll have two identical processes firing on the same event. We did this for safety, but you really need to keep this in mind. If there’s demand, we can create an option to delete or auto-deactivate the original process so you never have two of them.
2. The newly created flow has an api Name that has ‘converted_’ prepended to it, and it has the same label as the original process:

Save Order Considerations

Some Flows generated from Processes and Workflow Rules may behave slightly differently because of differences in the save order. See “Save-Order Considerations”, below. Currently ConvertToFlow always creates After-Save Triggers. Note the different points in the save order used by Process Builder, Workflow Rules, and After-Save Triggers in the list below.

As can be seen, the sensitive functionality is Escalation Rules and Entitlement Rules. Entitlement Rules execute after Process Builder and Workflow Rules but before Record-Triggered Flows that execute After Save. Escalation Rules execute after Workflow Rules but before Record-Triggerd Flows that execute After Save.

Not Yet Supported

ConvertToFlow does not:

• generate Before-Save Triggers.
• convert PB Processes that have more than one set of Scheduled Actions on a single row
• convert Workflow Rules that use Outbound Messages or Recursion.
• respect the Notify Assignee checkbox on the New Task action in Workflow Rules

Troubleshooting

Remote Site Settings issues

If you see this kind of error:

This page has an error. You might just need to refresh it. [LWC component’s @wire target property or method threw an error during value provisioning. Original error: [error returning from _getWFRDataApexIO Exception: Unauthorized endpoint, please check Setup->Security->Remote site settings. endpoint = https://saas-force-2107-dev-ed–c.visualforce.com/services/Soap/m/42.0%5D%5D Failing descriptor: {markup://c:flowPickerFSC}

Just as it says, you need to create a Remote Site Setting that points to your own org. In the example above, you would go to Remote Site Settings in Setup and create this:

Deployment Never Completes Issue

If retrieval works, but deployment of the converted flow seems to hang, it may be because of this: Your Apex Tests are all run when the metadata deploys even though Convert To Flow is deploying just a single piece of Flow metadata and no code. We recently changed Deploy to do this because with the previous setting that bypassed test runs, you couldn’t do Convert To Flow on production orgs (which don’t allow test runs to be bypassed for this situation).

What makes this confusing is that the test failure error isn’t getting delivered to Convert To Flow.

To see if this problem is caused by test failures, you can check your deployment status, which might look like this:

In this case, an unrelated class called DataManager, which is already installed on the org, has some broken test classes that are preventing Convert To Flow from successfully deploying.

You can also verify whether or not you have a broken Apex Class by clicking the Compile all classes link found here:

If you have _any_ compile failures, you will get this Deployment Never Completes error.

To get past this, you either need to fix the test class code or delete the class in question.

https://login.salesforce.com/packaging/installPackage.apexp?p0=04t5G000003rUwnQAE

Install

Step 1

This component requires that you first install or upgrade the Flow Base Packs

• FlowActionsBasePack Version 3.0.0 or later
• FlowScreenComponentsBasePack Version 3.0 or later

Step 2

Install (Production) V3.0.2. 6-8-21

Install (Sandbox) V3.0.2.0 6-8-22

Troubleshoot Installation Issues

Old Versions

Install (Production) V1.2.3 9-22-21

Install (Sandbox) V1.2.3 9-22-21

Install V1.2.2 9-19-21 removed a demo flow that was calling

Install V1.2 9-10-21 Added Before-Save and Subflow Support IMPORTANT: this version requires a Winter ’22 org to successfully convert Process Builder Processes that themselves launch flows.

Install ConvertToFlow V1.1.0.0 7-14 Added WFR Conversion.

V 1.0.4.0 Unlocked 6/23 First package

Source

view source

Here are some minor changes that were not documented officially.

Number and Currency Components

The Number and Currency components were upgraded to new LWC technology. This results in the following new behavior:

• When not in focus, the value is rounded or zero-padded according to the scale. This affects value presentation only. It has no impact on the actual value: when the field acquires focus again the actual, unrounded number will be displayed. It is also the actual number, which gets sent to the backend.
• The input allows for localized digits: e.g. US – 123456789; SA – ١٢٣٤٥٦٧٨٩
• The input allows
  • for entry of an arbitrary number of digits and a few other symbols (e,m,t,k,b) upto 255.
  • to increment or decrement current value using the up and down arrow keys (+/- 1)
  • for the use of the binary form by prefixing a sequence of 0s and 1s with 0b (link)
  • for the use of the scientific notation
• Big numbers
  • When in focus can get converted to scientific notation
    • 1e+38
  • When out of focus get expanded
    • 100,000,000,000,000,000,000,000,000,000,000,000,000.00000 instead of 1e+38.