Granted Patent B1

US 12,379,948 · App. 19/006,505 · Granted Aug 5, 2025

Inventors: Victor Vlasceanu (Bucharest, RO); Andrei Neculaesei (Bucharest, RO); Silviu Georgescu (Craiova, RO); Taqi Jaffri (Kirkland, WA); Venkata Syam Prakash Rapaka (Cupertino, CA)

Assignee: UiPath, Inc.

G06F9/453G06F16/3329

Abstract

Claims (90)

1. A computer-implemented method, comprising:

receiving a request to execute an agentic automation process (AOP), by a conductor application;

starting a job for an AOP engine to execute the AOP, by the conductor application;

receiving a request from the AOP engine to execute an automation for a step of the AOP via an artificial intelligence (AI) agent or a robotic process automation (RPA) robot, by the conductor application;

starting a job for the requested AI agent or RPA robot, by the conductor application;

receiving an indication from the AI agent or RPA robot that the execution of the automation was successful or failed, by the conductor application; and

informing the AOP engine that the execution of the automation was successful or failed, by the conductor application, wherein

the conductor application orchestrates the AOP engine and the RPA robot or AI agent.

2. The computer-implemented method of claim 1 , further comprising:

responsive to the execution of the automation being successful, continuing to facilitate execution of the AOP, by the conductor application, until the AOP is completed.

3. The computer-implemented method of claim 1 , wherein

the AI agent determines that another AI agent the RPA robot, or another RPA robot should be used for the automation, or the RPA robot determines that the AI agent or another AI agent should be used for the automation,

the other AI agent, the RPA robot, or the other RPA robot executes a portion of the automation of the AI agent, or the AI agent or the other AI agent executes a portion of the automation of the RPA robot,

the AI agent receives execution results from the other AI agent, the RPA robot, or the other RPA robot, or the RPA robot receive execution results from the AI agent or the other AI agent, and

by the first AI agent or the first RPA robot continues execution of the automation using the execution results.

4. The computer-implemented method of claim 3 , wherein the determining step is performed by the AI agent and the determining step comprises searching a database, by the agent, for AI agents and/or RPA robots that can assist with the automation, the database comprising natural language description of tasks that the AI agents perform.

5. The computer-implemented method of claim 3 , wherein the determining step is performed by the RPA robot and the determining step comprises searching a database, by the RPA robot, for AI agents that can assist with the automation.

6. The computer-implemented method of claim 3 , wherein the determining step is performed by the RPA robot and the determining step comprises determining to use the AI agent or the other AI agent based on one or more activities in an RPA workflow of the RPA robot.

7. The computer-implemented method of claim 3 , wherein responsive to the AI agent experiencing an error or failing to find a result, the one or more computer programs are further configured to cause the at least one processor to:

searching an agentic memory, searching an execution context, searching a context associated with a different industry or organization, or any combination thereof, by the AI agent; and

responsive to the search, outputting a result, by the AI agent.

8. The computer-implemented method of claim 3 , wherein initially, the method comprises:

requesting assistance from a human-in-the-loop in resolving a natural language request associated with the automation via a human-in-the-loop call, by the AI agent;

receiving a response from the human-in-the-loop, by the AI agent;

providing the response from the human-in-the-loop as output, by the AI agent; and

storing the response from the human-in-the-loop in agentic memory of the AI agent.

9. The computer-implemented method of claim 8 , further comprising:

fine-tuning one or more generative AI models using the response from the human-in-the-loop and responses from other humans in the loop and/or log traces, computing system state information, responses from third party employees, or a combination thereof;

determining whether the AI agent has achieved sufficient autonomy as defined by one or more metrics to operate autonomously; and

using the AI agent instead of the human-in-the-loop responsive to meeting or exceeding the one or more metrics.

10. One or more computing systems, comprising:

memory storing computer program instructions for a conductor application; and

at least one processor configured to execute the stored computer program instructions, wherein the computer program instructions are configured to cause the at least one processor to:

start a job for an agentic orchestration process (AOP) engine to execute an AOP, by the conductor application,

receive a request from the AOP engine to execute an automation for a step of the AOP via an artificial intelligence (AI) agent or a robotic process automation (RPA) robot, by the conductor application,

start a job for the requested AI agent or RPA robot, by the conductor application,

receive an indication from the AI agent or RPA robot that the execution of the automation was successful or failed, by the conductor application, and

inform the AOP engine that the execution of the automation was successful or failed, by the conductor application, wherein

the conductor application orchestrates the AOP engine and the RPA robot or AI agent.

11. The one or more computing systems of claim 10 , wherein the computer program instructions are further configured to cause the at least one processor to:

responsive to the execution of the automation being successful, continue to facilitate execution of the AOP until the AOP is completed.

12. The one or more computing systems of claim 10 , wherein

the AI agent determines that another AI agent the RPA robot, or another RPA robot should be used for the automation, or the RPA robot determines that the AI agent or another AI agent should be used for the automation,

the other AI agent, the RPA robot, or the other RPA robot executes a portion of the automation of the AI agent, or the AI agent or the other AI agent executes a portion of the automation of the RPA robot,

the AI agent receives execution results from the other AI agent, the RPA robot, or the other RPA robot, or the RPA robot receive execution results from the AI agent or the other AI agent, and

by the first AI agent or the first RPA robot continues execution of the automation using the execution results.

13. The one or more computing systems of claim 12 , wherein

the determining step is performed by the AI agent and the determining step comprises searching a database, by the agent, for AI agents and/or RPA robots that can assist with the automation, the database comprising natural language description of tasks that the AI agents perform, or

the determining step is performed by the RPA robot and the determining step comprises searching a database, by the RPA robot, for AI agents that can assist with the automation.

14. The one or more computing systems of claim 10 , wherein responsive to the AI agent experiencing an error or failing to find a result, the computer program instructions are further configured to cause the at least one processor to:

search an agentic memory, search an execution context, search a context associated with a different industry or organization, or any combination thereof, by the AI agent; and

responsive to the search, output a result, by the AI agent.

15. The one or more computing systems of claim 10 , wherein initially, the computer program instructions are configured to cause the at least one processor to:

request assistance from a human-in-the-loop in resolving a natural language request associated with the automation via a human-in-the-loop call, by the AI agent;

receive a response from the human-in-the-loop, by the AI agent;

provide the response from the human-in-the-loop as output, by the AI agent; and

store the response from the human-in-the-loop in agentic memory of the AI agent.

16. The one or more computing systems of claim 15 , wherein the computer program instructions are configured to cause the at least one processor to:

fine-tune one or more generative AI models using the response from the human-in-the-loop and responses from other humans in the loop and/or log traces, computing system state information, responses from third party employees, or a combination thereof;

determine whether the AI agent has achieved sufficient autonomy as defined by one or more metrics to operate autonomously; and

use the AI agent instead of the human-in-the-loop responsive to meeting or exceeding the one or more metrics.

17. One or more non-transitory computer-readable media storing one or more computer programs for a conductor application, the one or more computer programs configured to cause at least one processor to:

start a job for an agentic orchestration process (AOP) engine to execute an AOP, by the conductor application,

receive a request from the AOP engine to execute an automation for a step of the AOP via an artificial intelligence (AI) agent or a robotic process automation (RPA) robot, by the conductor application,

start a job for the requested AI agent or RPA robot, by the conductor application,

receive an indication from the AI agent or RPA robot that the execution of the automation was successful or failed, by the conductor application, and

inform the AOP engine that the execution of the automation was successful or failed, by the conductor application, wherein

the conductor application orchestrates the AOP engine and the RPA robot or AI agent.

18. The one or more non-transitory computer-readable media of claim 17 , wherein the one or more computer programs are further configured to cause the at least one processor to:

responsive to the execution of the automation or the other automation being successful, continue to facilitate execution of the AOP, by the conductor application, until the AOP is completed.

19. The one or more non-transitory computer-readable media of claim 17 , wherein

the AI agent determines that another AI agent, the RPA robot, or another RPA robot should be used for the automation or the other automation, or the RPA robot determines that the AI agent or another AI agent should be used for the automation or the other automation,

the other AI agent, the RPA robot, or the other RPA robot executes a portion of the automation or the other automation of the AI agent, or the AI agent or the other AI agent executes a portion of the automation or the other automation of the RPA robot,

the AI agent receives execution results from the other AI agent, the RPA robot, or the other RPA robot, or the RPA robot receive execution results from the AI agent or the other AI agent, and

by the first AI agent or the first RPA robot continues execution of the automation or the other automation using the execution results.

20. The one or more non-transitory computer-readable media of claim 19 , wherein the determining step is performed by the AI agent and the determining step comprises searching a database, by the AI agent, for AI agents and/or RPA robots that can assist with the automation, the database comprising natural language description of tasks that the AI agents perform.

21. The one or more non-transitory computer-readable media of claim 19 , wherein the determining step is performed by the RPA robot and the determining step comprises searching a database, by the RPA robot, for AI agents that can assist with the automation.

22. The one or more non-transitory computer-readable media of claim 19 , wherein the determining step is performed by the RPA robot and the determining step comprises determining to use the AI agent or the other AI agent based on one or more activities in an RPA workflow of the RPA robot.

23. The one or more non-transitory computer-readable media of claim 19 , wherein responsive to the AI agent experiencing an error or failing to find a result, the one or more computer programs are further configured to cause the at least one processor to:

search an agentic memory, search an execution context, search a context associated with a different industry or organization, or any combination thereof, by the AI agent; and

responsive to the search, output a result, by the AI agent.

24. The one or more non-transitory computer-readable media of claim 19 , wherein initially, the one or more computer programs are configured to cause the at least one processor to:

request assistance from a human-in-the-loop in resolving a natural language request associated with the automation via a human-in-the-loop call, by the AI agent;

receive a response from the human-in-the-loop, by the AI agent;

provide the response from the human-in-the-loop as output, by the AI agent; and

store the response from the human-in-the-loop in agentic memory of the AI agent.

25. The one or more non-transitory computer-readable media of claim 24 , wherein the one or more computer programs are further configured to cause the at least one processor to:

fine-tune one or more generative AI models using the response from the human-in-the-loop and responses from other humans in the loop and/or log traces, computing system state information, responses from third party employees, or a combination thereof;

determine whether the AI agent has achieved sufficient autonomy as defined by one or more metrics to operate autonomously; and

use the AI agent instead of the human-in-the-loop responsive to meeting or exceeding the one or more metrics.

Assignments (1)

Continuity (1)

References Cited (88)

Cited By (2)

US 12,694,227 US 12,711,308