AI Builder is a Power Automate feature that adds AI capabilities - text recognition, sentiment analysis, document processing - to your Tallyfy automations without writing code. If your flows handle unstructured data like scanned forms, customer feedback in form fields, or images attached to tasks, AI Builder can interpret that content and act on it.
What is AI Builder?
AI Builder gives you pre-built AI models ready to go, plus the ability to train custom models on your own Tallyfy data. You don’t need data science experience.
Where it really shines is unstructured data. Scanned forms that need to launch a process, customer emails that should create a task, feedback text that needs routing based on tone - AI Builder handles all of it.
Model types relevant to Tallyfy
Pre-built models (ready to use): Microsoft trained these, so they work immediately.
Text Recognition (OCR): Pulls text from images and PDFs - great for scanned forms that feed into a Tallyfy process.
Sentiment Analysis: Detects positive or negative tone in text. Useful for analyzing customer feedback from Tallyfy form fields and routing accordingly.
Key Phrase Extraction: Finds main points in text - handy for summarizing lengthy notes from a task.
Language Detection: Identifies which language someone’s writing in. Important if your processes handle multilingual input.
Business Card Reader: Grabs contact info from photos, then fills fields in your Tallyfy process.
Receipt Processing: Extracts expense data from receipts into your approval process in Tallyfy.
Custom models (train with your data): These take more setup but match your specific Tallyfy use case.
Form Processing: Teach it to read your invoices, purchase orders, or any documents that kick off a Tallyfy process. Handles non-standard layouts well.
Object Detection: Spot specific objects in images - think quality control processes where team members upload photos to tasks.
Prediction: Uses historical Tallyfy data to forecast outcomes, like whether a process will meet its SLA based on initial form field values.
Category Classification: Sorts text from Tallyfy tasks or form fields into categories you define. Support tickets logged through Tallyfy get tagged automatically.
Example: sentiment analysis on Tallyfy feedback
Here’s a practical walkthrough. You’ll use AI Builder’s Sentiment Analysis to process customer feedback collected through a Tallyfy task.
Scenario: Your “Customer Feedback Collection” process has a task where customers type feedback into a form field. You want to detect whether they’re happy or upset, then route the response accordingly.
The Tallyfy Power Automate connector has 13 actions and no triggers. To start a flow when a Tallyfy task completes, set up a Tallyfy webhook that calls your flow’s HTTP trigger - the connector itself doesn’t provide trigger events.
Trigger: “When an HTTP request is received” (or a Tallyfy webhook).
Configure a Tallyfy webhook to fire when the “Provide Feedback” task completes, sending the task data (including form field values) to your flow’s HTTP endpoint. See creating your first flow for setup basics.
Action: Tallyfy - “Get a member’s tasks” (Get_User_Tasks).
Use this action to retrieve task details including the feedback text from the relevant form field (e.g., CustomerFeedbackText). Alternatively, parse the webhook payload directly if it includes the form field data you need.
Action: AI Builder - “Detect the language being used in text”.
Sentiment analysis works best when the language is known.
Text: Provide the CustomerFeedbackText value from the previous step.
Action: AI Builder - “Analyze positive or negative sentiment in text”.
Language: Use the Language output from the detection step.
Text: Provide the same CustomerFeedbackText value.
Action: Tallyfy - “Create a task” (Create_Task). Create a follow-up task assigned to a customer service manager with the original feedback and customer details.
If positive (optional):
Send a thank you email, or log the feedback and sentiment score to SharePoint or Excel for trend analysis.
Save and test. Submit feedback through your Tallyfy task and verify that sentiment analysis routes it correctly. See managing and monitoring flows for testing tips.
Custom model example: form processing with Tallyfy
Custom models take more setup but match your exact data. Here’s how Form Processing handles scanned documents that launch Tallyfy processes.
Scenario: You receive purchase orders as scanned PDFs in different formats from various suppliers. Each PO contains a PO Number, Customer Name, Items, and Quantities. You want these to automatically launch your “Order Fulfillment” process in Tallyfy - no manual data entry.
In AI Builder (Power Automate portal):
Create a new Form processing model.
Define the fields to extract (e.g., PO Number, Customer Name, ItemDescription, Quantity).
Upload at least 5 sample PO documents per collection (more if layouts vary).
Tag each sample by drawing boxes around the data you want extracted and mapping them to your fields.
Train, test, and publish the model.
In your Power Automate flow:
Trigger: “When a new file is created” in a SharePoint folder where scanned POs land.
Action: AI Builder - “Extract information from forms”.
Select your published Form Processing model.
Form type:PDF Document (or image).
Form: Provide the file content from the trigger.
The extracted fields (PO Number value, Customer Name value, etc.) become available as dynamic content.
Action: Tallyfy - “Launch a Process” (Create_Run).
Map extracted values to the corresponding kick-off form fields in your Tallyfy template.
Licensing and considerations
Licensing: AI Builder requires separate Power Platform credits - they aren’t included with standard Power Automate licenses. Credits are consumed each time models process data.
Effort: Pre-built models work immediately. Custom models need sample documents from your Tallyfy processes, careful tagging, and training time.
Accuracy: Custom model quality depends on training data. More varied examples produce better real-world results.
What AI Builder adds to Tallyfy workflows
Less manual data entry: Documents and emails get processed automatically - extracted data flows into Tallyfy form fields without copy-paste errors.
Content analysis at scale: Customer comments in Tallyfy tasks, support notes, survey responses - AI Builder pulls out sentiment and key phrases you’d otherwise have to read manually.
Smart routing: AI analyzes incoming data before it reaches Tallyfy and directs it to the right process or team. Urgent issues get fast-tracked automatically.
Predictive actions: Prediction models can flag at-risk processes before deadlines are missed.
Microsoft Power Automate connects Tallyfy with other business apps through actions like creating tasks and launching processes, while Tallyfy handles human-centric workflow management and Power Automate handles system integrations.
Microsoft Power Automate serves as a no-code bridge between Tallyfy and your other business systems—including Office 365 and SharePoint and Dynamics—using a Premium connector with 13 actions and OAuth 2.0 authentication to automate data syncing and trigger Tallyfy processes from external events like emails or CRM updates.
Power Automate connects email services like Outlook 365 with Tallyfy workflows - automatically launching processes from incoming emails, extracting data and attachments, and sending notifications based on task completions.