A loop with a short-interval wait that actively checks for the element’s existence is the most robust. It balances efficiency (not waiting longer than necessary) with reliability (ensuring the element is present before proceeding). Fixed waits are inefficient and brittle, relying solely on global timeouts can lead to inconsistent failures, and unnecessary navigations are poor practice.

Intermittent failures on write stages, especially with numeric data, are often due to data type conflicts with the target cell’s format. A cell formatted as ‘Text’ will reject a pure number from Blue Prism. The first diagnostic step should be to check and standardize the cell format in the spreadsheet template, often to ‘General’.

Using input and output parameters is the cleanest, most encapsulated, and recommended method for passing data between processes in Blue Prism. It creates a clear contract, avoids hidden dependencies on global state or external resources, and makes the processes more reusable and easier to debug.

A ‘Recover’ stage in the main path executes unconditionally as part of the normal workflow. Placing it right after navigation means the bot will reset the application at the start of every process run, which is incorrect behavior. ‘Recover’ stages should be placed within ‘Recovery’ blocks so they only execute when a specific exception is thrown and caught.

The ‘Attempt Limit’ defines the maximum number of processing attempts (including the first). A limit of 3 allows for the initial try plus two retries. The ‘For Processing’ flag on ‘Get Next Item’ is crucial, as it automatically locks the item (setting it to ‘In Progress’) so other robots cannot pick it up, ensuring single-agent processing.

DateTime.ParseExact() allows you to specify the exact input format (‘dd/mm/yyyy’), which safely handles variations and validates the date. Once you have a valid DateTime object, you can use .ToString(“yyyy-MM-dd”) to output it in the required format. This method is robust against input errors and clearer than manual string manipulation.

When window titles are dynamic, relying on the ‘Title’ attribute is brittle. ‘Active Accessibility’ (AA) mode often exposes underlying, static properties like ‘AutomationId’, ‘Name’, or ‘Class Name’ that are more reliable identifiers for the application window, leading to a more resilient object.

A ‘Recovery’ block is designed for this exact scenario. It allows you to catch a specific exception (e.g., a database timeout), execute a recovery sequence (like reinitializing the connection), and then use ‘Resume’ to retry the exact stage that failed. This is more precise and efficient than a simple retry loop or restarting the entire process.

This is a fundamental design principle. ‘Action’ pages are the primary means of interacting with target applications (using Read, Write, Navigate stages from the Application Modeller). ‘Function’ pages are used for supporting logic within the Business Object, such as data validation, transformation, or complex calculations, where no direct application interaction is required.

A loop with a short-interval wait that actively checks for the element’s existence is the most robust. It balances efficiency (not waiting longer than necessary) with reliability (ensuring the element is present before proceeding). Fixed waits are inefficient and brittle, relying solely on global timeouts can lead to inconsistent failures, and unnecessary navigations are poor practice.0