Action and also a workaround through which we can have multiple suggested text options in the approval action so let's get started .
For the approvals connector in power automate we now have a new action called start and wait for an approval of text with this new action we get the ability to create an approval.
Task that we can assign to users or groups provide details about the approval and even link to an item but what's new here is the ability to.
Create suggested text that the approver can approve or modify for the suggested text I have put Microsoft flow as an example if I was to test this flow.
I receive a new approval task that I can approve or eject and here apart from approval overview and details we have a new section here called text.
This can be modified by the approver before accepting or rejecting the request this is a mandatory field that the approver would have to fill in.
I already had a default value in there the approver can modify that text and complete the task if we explore the flow run for this.
Specific action we have all the standard details that we get from a completed approval task but apart from that we also get accepted text.
That's the text that the approver has accepted which could be modified by the approver directly in the approval task itself this new action has been introduced.
Posts Related:
- Talk Power BI LIVE Q&A (Subscribe & Join) January 29, 2021
- Birthday Invitation Cards Templates Free Download In PSD Files Urdu Hindi Photoshop Tutorial
- Learn How To Create This Never Seen Before Drag & Drop Task Scheduler In Excel Today Part 1
Keeping scenarios around - ID Card Make
Responsible AI in mind scenarios where you would like to leverage AI Builders new text generation model that is based upon the Azure open.AI service which allows us to create text summarize information and a lot more with GPT AI generated content can be factually incorrect inappropriate or biased.
And keeping that in mind it is extremely important to introduce human oversight for AI generated content and this new action is perfect for such scenarios.
Create text with GPT action comes with some amazing templates that we can take advantage of extracting information from text the sample prompt here includes.
Extracting container numbers and registration numbers from some sample text and if we test this AI interprets The Prompt performs the.
Necessary action and provides a response that gets the container numbers and registration numbers from this unstructured text now let's take a real world example.
My organization receives shipment information via emails it includes information about order numbers the status of the shipment the estimated delivery date and the tracking.
Number associated with the order and today these key information points from the email is manually captured and input in Excel now I would like to automate this and.
Take advantage of AI so I'll go ahead and copy sample email and here I will paste that email.
I can prompt AI
To get order numbers order status estimated delivery date and tracking.Number if I click test AI starts providing me all the details I would like this output to be in a CSV format.
So I'll change the prompt and click test it provides the details in CSV format we can also change the format of the data for example the date updated the prompt I'll click test.
Ai's thoughts extracting the details so let's create an automated Cloud flow that triggers when a new email arrives in my mailbox I will give my flow a name click create.
For the float trigger I'll go to show Advanced options the emails that I receive the subject always includes Contours or shipment.
So I would like to trigger the flow for emails where the subject includes contoso shipment I'll add a new step in my flow and call the action create text with GPT.
The prompt I already have my prompt ready I will copy and paste and the text that is hard coded.
I will remove that and here I would like to place the body of the email now the body of the email will be returned in HTML format.
But I want to pass plain text so I will add a new step and use the HTML to text content conversion action the content would be the body of the.
Email and now for create text with GPT in the prompt in between start of text and end of text tokens.
I will place the plain text content so this action will output an AI generated response that can have mistakes so I would like to add human oversight.
Here basically I would like to Leverage the start and weight for an approval of text action here which gives the approver the opportunity to review the AI generated content and at the same.
Time also modify it if needed the title I will put the subject of the email suggested text.
Would be text which is generated text from the create text with gbt action I will assign this to my approver in this case I'll pick Reza and in the details for this approval.
I will pick the body of the email in plain text that will allow the approver to see the actual content of the email and also the AI generated content.
Now once the approval decision is taken I need to add a condition to check the outcome of the approval so if the approver.
Says approve only then I would like to create a CSV file I will add an action here I can create the file in OneDrive or SharePoint.
For this demo I will pick OneDrive for business the path in the root for OneDrive for business I have a folder created called shipment.
Information so I will select this the name of the file I'll pick the subject of the email dot CSV and the dynamic content here is accepted.
Text that's the text that the approver has accepted so I will select this I will click save.
My flow is saved and listening to any new email that arrives in my mailbox that includes the subject so I have just received an email that has that specific subject.
Once the float triggers it will call the AI action to generate the CSV output and now I am waiting for the approval action.
Now this action can be taken from the approval Center or I can directly respond to it from my email.
I have the details of the email body plus I have this option where I can modify the text that has been output from the GPT action I can make changes if need be.
And I can either approve or reject in this case I'll go ahead and approve it and submit my response once the response is received.
Flow will check the response if it's approved then it will go and create a file in my OneDrive and if I go to the shipment information folder in my OneDrive here is the CSV.
File that has all the data that is extracted I have a flow that triggers when a new item is created in a SharePoint list this list tracks sales discount requests.
And to track the details of the approval I added a simple Choice column called approval status that has three values and I added an approval commands column which is of.
Type multi-line of text and both these columns are hidden from the form experience so that when users make the request they don't see the options to update those.
