Intelligent Character Recognition (ICR) mode is designed for reading text from documents where the text’s position is consistent but the exact formatting varies. It uses anchor text and relative positioning to find the target data, making it more resilient to changes in font, size, or minor layout shifts compared to Static Region mode, which relies on fixed coordinates.

Intermittent ‘Element not found’ errors are often caused by dynamic attributes (like IDs or classes that change). The first practical step is to re-spy the element and identify more stable attributes for matching, such as a unique name, title, or a combination of attributes that consistently identifies the element, rather than relying on volatile properties.

The most robust and efficient approach is to build resilience into the business logic. By checking for a known timeout message or state after key interactions, the process can identify the failure context (it’s in a logged-out state) and execute a targeted recovery (re-login and resume from the appropriate point), rather than restarting the entire process or using inefficient fixed waits.

Blue Prism’s DateAdd function expects its first parameter to be a Date data type. If [Date] is spied from an application as text (e.g., “01/01/2024”), it must be converted using a function like DateParse(“[Date]”) before it can be used in date arithmetic. The correct expression would be: FormatDate(DateAdd(DateParse(“[Date]”), “30”), “dd/mm/yyyy”).

The core principle for maintainability is modularization. Grouping related logic into Sub-Processes or calling centralized Actions from a Business Object reduces duplication, isolates logic for easier testing and debugging, and makes the main process page much simpler to read and understand. Comments are helpful but don’t change the structural complexity.

Configuration values that vary by environment should be externalized. The ‘Data’ tab in Process or Object Studio (for values used by that item) or, better yet, System – Environment Variables (for values shared across multiple processes) are the correct locations. This allows the same process blueprint to be deployed to different environments by simply changing the external configuration, not the logic itself.

An empty cell is a predictable business condition, not an exceptional system error. The correct approach is to proactively check for it using business logic (a Decision stage). This is clearer, more efficient, and aligns with the principle of using Recoveries for unexpected system failures (like Excel not being open), not for expected data variations.

Blue Prism provides dedicated spying modes for different technologies. For a Java application, the ‘Java’ mode is specifically designed to interface with Java Access Bridge and will often provide more reliable and deeper insight into the application’s elements than the generic ‘AA’ mode. This should always be the first alternative to try for Java apps.

Placing the Log stage inside the Recover block ensures the audit log is written only when the exception is caught for that particular iteration. The data relevant to the failed item is still in scope. Logging at other points would either create too much noise (logging successes) or lose the specific item’s context (logging at the end).

Intelligent Character Recognition (ICR) mode is designed for reading text from documents where the text’s position is consistent but the exact formatting varies. It uses anchor text and relative positioning to find the target data, making it more resilient to changes in font, size, or minor layout shifts compared to Static Region mode, which relies on fixed coordinates.0