Inserting and/or replacing machine learning models in a pipeline for robotic process automation workflows
A reconfigurable workbench pipeline for robotic process automation (RPA) workflows is disclosed. Different workbench pipelines may be built for different users. For instance, a global workflow (e.g., a receipt extractor) may be built and used initially, but this workflow may not work optimally or at all for a certain user or a certain task. A machine learning (ML) model may be employed, potentially with a human-in-the-loop, to specialize the global workflow for a given task.
1. A computer-implemented method executed by a computing system for providing a reconfigurable workbench pipeline for robotic process automation (RPA) workflows using machine learning (ML), comprising:
executing a global workflow, by an RPA robot executing on the computing system, wherein the global workflow is an initial workflow for a task and the global workflow comprises a pipeline comprising a plurality of ML models;
determining that the pipeline of the global workflow is not working for a scenario, by the RPA robot, another RPA robot, or another software application of the computing system, by sequentially checking an input and an output of the plurality of ML models of the pipeline of the global workflow;
inserting at least one new ML model into the pipeline, replacing at least one ML model in the pipeline, or both, by the computing system;
generating a local pipeline of a local workflow of the computing system, by the computing system; and
executing the local workflow, by the RPA robot, wherein
the generated local pipeline comprises the at least one inserted new ML model, the at least one replaced ML model, or both,
the local workflow comprises a user interface (UI) automation activity, and
the UI automation activity interacts with the UI of a display of the computing system using one or more drivers to perform part of the task.
2. The computer-implemented method of claim 1 , further comprising:
notifying an RPA developer that the pipeline is not functioning correctly; and
receiving guidance from the RPA developer regarding how to correct the pipeline.
3. The computer-implemented method of claim 2 , wherein the notification comprises sending a communication to the RPA developer that comprises the local workflow, one or more pipeline ML models that are not functioning properly, log data indicating which replacement ML models and/or modifications to existing ML models were attempted, or any combination thereof.
4. The computer-implemented method of claim 2 , wherein the guidance from the RPA developer comprises a repaired version of the pipeline, the local workflow, one or more new pipeline ML models, or any combination thereof.
5. The computer-implemented method of claim 1 , further comprising:
analyzing an input of a first ML model of the pipeline; and
when the input of the first ML model is incorrect, analyzing a data source of the input of the first ML model.
6. The computer-implemented method of claim 1 , wherein the one or more drivers comprise an operating system (OS) driver, a browser driver, a virtual machine (VM) driver, an enterprise application driver, or any combination thereof.
7. A computer-implemented method executed by a computing system for providing a reconfigurable workbench pipeline for robotic process automation (RPA) workflows using machine learning (ML), comprising:
determining that a pipeline of a global workflow is not working for a scenario, by an RPA robot, another RPA robot, or another software application of the computing system, by sequentially checking an input and an output of a plurality of ML models of the pipeline of the global workflow;
inserting at least one new ML model into the pipeline, replacing at least one ML model in the pipeline, or both, by the computing system;
generating a local pipeline of a local workflow of the computing system, by the computing system; and
executing the local workflow, by the RPA robot, wherein
the generated local pipeline comprises the at least one inserted new ML model, the at least one replaced ML model, or both,
the local workflow comprises a user interface (UI) automation activity, and
the UI automation activity interacts with the UI of a display of the computing system using one or more drivers to perform part of the task.
8. The computer-implemented method of claim 7 , further comprising:
notifying an RPA developer that the pipeline is not functioning correctly, wherein
the notification comprises sending a communication to the RPA developer that comprises the local workflow, one or more pipeline ML models that are not functioning properly, log data indicating which replacement ML models and/or modifications to existing ML models were attempted, or any combination thereof.
9. The computer-implemented method of claim 8 , further comprising:
receiving guidance from the RPA developer regarding how to correct the pipeline, wherein
the guidance from the RPA developer comprises a repaired version of the pipeline, the local workflow, one or more new pipeline ML models, or any combination thereof.
10. The computer-implemented method of claim 8 , further comprising:
analyzing an input of a first ML model of the pipeline; and
when the input of the first ML model is incorrect, analyzing a data source of the input of the first ML model.
11. The computer-implemented method of claim 7 , wherein the one or more drivers comprise an operating system (OS) driver, a browser driver, a virtual machine (VM) driver, an enterprise application driver, or any combination thereof.
12. A computer-implemented method executed by a computing system for providing a reconfigurable workbench pipeline for robotic process automation (RPA) workflows using machine learning (ML), comprising:
inserting at least one new ML model into the pipeline, replacing at least one ML model in the pipeline, or both, by the computing system;
generating a local pipeline of a local workflow of the computing system, by the computing system; and
executing the local workflow, by an RPA robot, wherein
the generated local pipeline comprises the at least one inserted new ML model, the at least one replaced ML model, or both,
the local workflow comprises a user interface (UI) automation activity, and
the UI automation activity interacts with the UI of a display of the computing system using one or more drivers to perform part of the task.
13. The computer-implemented method of claim 12 , further comprising:
determining that the pipeline of the global workflow is not working correctly for the scenario.
14. The computer-implemented method of claim 12 , further comprising:
notifying an RPA developer that the pipeline is not functioning correctly, wherein
the notification comprises sending a communication to the RPA developer that comprises the local workflow, one or more pipeline ML models that are not functioning properly, log data indicating which replacement ML models and/or modifications to existing ML models were attempted, or any combination thereof.
15. The computer-implemented method of claim 12 , wherein the one or more drivers comprise an operating system (OS) driver, a browser driver, a virtual machine (VM) driver, an enterprise application driver, or any combination thereof.