22.2 Data Fabric Activities and the Platform Connector
Key Takeaways
Data Fabric records are handled with Create, Get by Id, Query, Update, and Delete Entity Record activities, plus file field activities.
Classic Data Service activities include batch Create, Update, and Delete Multiple Entity Records.
The UiPath Data Fabric platform connector (renamed May 27, 2026) offers connector-based activities and Record Created and Record Updated triggers.
Filter and page in Query Entity Records rather than filtering all records in the workflow.
Robots need a Data Fabric role such as Data Writer or Data Reader, and folder access for folder-level entities.
22.2 Data Fabric Activities and the Platform Connector
Core Concept: Automations work with Data Fabric records through activities: classic Data Service activities in Studio projects and connector-based activities of the UiPath Data Fabric connector, a platform connector in Integration Service. The exam description names the "Data Service platform connector". On May 27, 2026, the connector was renamed the UiPath Data Fabric connector.
Record Activities
| Activity | Use |
|---|---|
| Create Entity Record | Add a new record |
| Get Entity Record by Id | Read one record by its Id |
| Query Entity Records | Read records that match filter conditions, with sorting and paging |
| Query Multiple Entity Records | Query records from several entities (connector-based activities) |
| Update Entity Record | Change fields of an existing record |
| Delete Entity Record | Remove a record |
| Upload File to Record Field / Download File from Record Field / Delete File from Record Field | Work with File fields |
| Create/Update/Delete Multiple Entity Records (classic) | Batch operations, for example loading many rows from a CSV |
The connector also provides triggers, Record Created and Record Updated, which start automations when data changes.
Classic Versus Connector-Based Activities
| Classic Data Service activities | Data Fabric connector activities | |
|---|---|---|
| Package | The Data Service activity package, used with entities imported into the project | The UiPath Data Fabric platform connector in Integration Service |
| Triggers | Not applicable | Record Created and Record Updated |
| Batch operations | Create, Update, and Delete Multiple Entity Records | Query Multiple Entity Records |
Both work with the same entities and records. Choose the connector activities when you need Data Fabric triggers or are building connector-based automations, and the classic activities for batch operations on imported entities.
Querying Records Well
Query Entity Records takes filter conditions on fields, such as Status equals "New" and Amount greater than 10000, plus sorting and paging settings.
- Filter in the query, not afterwards in a loop, to reduce data transfer.
- Page through large results instead of requesting everything at once.
- Select only needed fields when the activity allows it.
- Use relationships: query invoices by supplier through the relationship field.
File Fields
- Upload File to Record Field attaches a file, such as the invoice PDF, to a record.
- Download File from Record Field saves it locally for processing.
- Delete File from Record Field removes it.
For long-running processes that may resume on another machine, keep the file in the record and download it again after resuming, instead of relying on a local copy.
Permissions for Robots
The robot account or connection used by the automation needs a Data Fabric role that allows the operations, for example Data Writer for creating and updating records, or Data Reader for queries. For folder-level entities, the robot must also have access to that folder.
Record Triggers
Record Created and Record Updated triggers start automations when data changes. For example:
- A new Supplier record triggers a background check process.
- An Invoice whose Status changes to "Approved" triggers the payment process.
Design triggered processes to be idempotent: read the current record by Id at the start and check whether the work is still needed.
Worked Example: Loading and Processing Claims
- A dispatcher reads a daily CSV into a DataTable and uses Create Multiple Entity Records to load claims in batches.
- Each claim's Id goes to a queue.
- The performer uses Get Entity Record by Id, processes the claim, uploads the decision letter with Upload File to Record Field, and sets Status with Update Entity Record.
- A Record Updated trigger on Status "Decided" starts a notification process through the Data Fabric connector.
- A weekly report uses Query Entity Records with filters on Status and decision date.
Exam-Style Scenarios
Scenario 1: A dispatcher must load 20,000 rows from a CSV into an entity overnight. Use the batch activities, such as Create Multiple Entity Records, in chunks, rather than one Create Entity Record call per row.
Scenario 2: An approval app and a performer must work on the same invoice data. Store the invoice as a record, pass the record Id through the queue, and let both read and update the same record.
Scenario 3: A notification must go out when a supplier record is created. Use the Record Created trigger of the Data Fabric connector, and read the current record by Id at the start of the process.
Scenario 4: A performer fails with a permission error on Update Entity Record. Check that the robot's account has a role with write access, such as Data Writer, and access to the entity's folder.
Common Traps
- Looping over all records and filtering in the workflow instead of filtering in the query.
- Forgetting to give the robot account a Data Fabric role.
- Relying on local file copies across a long-running wait.
- Not updating imported entities in Studio after the schema changes.
An automation must start whenever an Invoice record's status is changed in Data Fabric. What should be used?
A time trigger that queries all invoices every minute.
The Record Updated trigger of the UiPath Data Fabric connector.
A queue trigger on the Invoice entity.
A webhook on the Orchestrator Jobs page.
A performer needs only New invoices over 10,000 from an entity with 200,000 records. What is the best practice?
Query all records and filter them in a For Each loop.
Download the entity as CSV first.
Use Query Entity Records with filter conditions on Status and Amount, and page through the results.
Use Get Entity Record by Id for every Id.
A long-running process must reuse the invoice PDF after resuming, possibly on another machine. What is the safest approach?
Keep the PDF in the record's File field and download it again after resuming.
Keep the local file path in a variable across the wait.
Store the PDF bytes in a queue item's Specific Data.
Save the PDF to the desktop of the first machine.
Sections you finish are checked off in the contents.