Granted Patent B1

US 12,282,719 · App. 18/775,938 · Granted Apr 22, 2025

Inventors: Roman Fedoruk (Cumming, GA); John Manton (Alpharetta, GA); Spencer Reagan (Marietta, GA); Gregory Roberts (Alpharetta, GA); Erich Stuntebeck (Johns Creek, GA)

Assignee: Airia LLC

G06F30/27

Abstract

Claims (61)

1. A method of creating an artificial intelligence (AI) pipeline, comprising:

receiving a selection of pipeline objects on a user interface (UI), the selected pipeline objects including a dataset object and an AI model, wherein the dataset object includes at least one vector database;

receiving inputs to the UI to arrange the selected pipeline objects in an AI pipeline, including causing an execution linking between the selected dataset object and the AI model to be established, wherein the UI visually represents the established execution linking between the pipeline objects, wherein the pipeline objects are moveable within the UI by a user, and wherein the AI pipeline is displayed as an execution flow within the UI,

wherein establishing the execution linking between the selected dataset object and AI model comprises validating that outputs from the selected dataset object can be used as inputs to the AI model;

causing a management policy to be applied to the dataset object, compliance with the management policy being a prerequisite for accessing the dataset object through an execution of the AI pipeline, wherein the UI visually represents the application of the management policy to the dataset object,

wherein enforcing the management policy includes verifying at least one of device location and a device security state based on a profile, and wherein the AI pipeline includes a conditional execution alternative based on noncompliance with the management policy;

causing a simulated execution of the AI pipeline to be displayed within the UI, including:

receiving a query in the UI;

receiving compliance information;

causing the selected pipeline objects to be executed in an order that follows the execution linking displayed within the UI and applies the management policy based on the compliance information; and

causing an output of the simulated execution to be displayed in the UI; and

causing the AI pipeline to be deployed, wherein the deployed AI pipeline is accessible by at least one AI application through a generated endpoint.

2. The method of claim 1 , wherein causing the execution linking between the selected dataset object and AI model to be established comprises causing a pipeline manifest file to be generated which stores the arrangement of pipeline objects in the AI pipeline.

3. The method of claim 2 , further comprising:

validating the pipeline manifest against a dependency rule, wherein the dependency rule dictates an event that must occur before at least one of the selected pipeline objects can execute; and

displaying, within the UI, a validation of the pipeline manifest.

4. The method of claim 1 , wherein the method further comprises, prior to displaying the selected pipeline objects, determining that a computing device through which the UI is displayed is compliant with at least one pipeline administrative user policy, wherein compliance with the pipeline administrative user policy is determined at least in part on information received from a management agent that executes on the computing device.

5. The method of claim 1 , wherein prior to displaying the selected pipeline objects in the UI, the method further comprises authenticating the user based on a user identifier and a tenant identifier.

6. The method of claim 1 , wherein the selected pipeline objects are selected from a displayed menu that includes at least one of AI model prompts, datasets, AI models, and executable code objects, and wherein one of the executable code objects is a conditional object that allows user selection of an if-then statement for at least two branches within the AI pipeline.

7. The method of claim 1 , wherein the management policy requires that an end user attempting to access the AI application is authorized to access the dataset object based at least in part on the end user being associated with an identifier of an authorized group and a client device of the end user being compliant with at least one pipeline end user policy, the client device of the end user being a computing device through which the attempt to access the AI application is attempted.

8. A non-transitory, computer-readable medium including instructions are executed by a processor and cause the processor to perform stages for creating an artificial intelligence (AI) pipeline, the stages comprising:

receiving a selection of pipeline objects on a user interface (UI), the selected pipeline objects including a dataset object and an AI model, wherein the dataset object includes at least one vector database;

receiving inputs to the UI to arrange the selected pipeline objects in an AI pipeline, including causing an execution linking between the selected dataset object and AI model to be established, wherein the UI visually represents the established execution linking between the pipeline objects, wherein the pipeline objects are moveable by a user within the UI, and wherein the AI pipeline is displayed as an execution flow within the UI,

wherein establishing the execution linking between the selected dataset object and AI model comprises validating that outputs from the selected dataset object can be used as inputs to the AI model;

causing a management policy to be applied to the dataset object, compliance with the management policy being a prerequisite for accessing the dataset object through an execution of the AI pipeline, wherein the UI visually represents the application of the management policy to the dataset object,

wherein enforcing the management policy includes verifying at least one of device location and a device security state based on a profile, and wherein the AI pipeline includes a conditional execution alternative based on noncompliance with the management policy;

causing a simulated execution of the AI pipeline to be displayed within the UI, including:

receiving a query in the UI;

receiving compliance information;

causing the selected pipeline objects to be executed in an order that follows the execution linking displayed within the UI and applies the management policy based on the compliance information; and

causing an output of the simulated execution to be displayed in the UI; and

causing the AI pipeline to be deployed, wherein the deployed AI pipeline is accessible by at least one AI application through a generated endpoint.

9. The non-transitory, computer-readable medium of claim 8 , wherein causing the execution linking between the selected dataset object and AI model to be established comprises causing a pipeline manifest file to be generated which stores the arrangement of pipeline objects in the AI pipeline.

10. The non-transitory, computer-readable medium of claim 9 , the stages further comprising:

validating the pipeline manifest against a dependency rule, wherein the dependency rule dictates an event that must occur before at least one of the selected pipeline objects can execute; and

displaying, within the UI, a validation of the pipeline manifest.

11. The non-transitory, computer-readable medium of claim 8 , wherein the stages further comprise, prior to displaying the selected pipeline objects, determining that a computing device through which the UI is displayed is compliant with at least one pipeline administrative user policy, wherein compliance with the pipeline administrative user policy is determined at least in part on information received from a management agent that executes on the computing device.

12. The non-transitory, computer-readable medium of claim 8 , wherein prior to displaying the selected pipeline objects in the UI, the stages further comprise authenticating the user based on a user identifier and a tenant identifier.

13. The non-transitory, computer-readable medium of claim 8 , wherein the selected pipeline objects are selected from a displayed menu that includes at least one of AI model prompts, datasets, AI models, and executable code objects, and wherein one of the executable code objects is a conditional object that allows user selection of an if-then statement for at least two branches within the AI pipeline.

14. The non-transitory, computer-readable medium of claim 8 , wherein the management policy requires that an end user attempting to access the AI application is authorized to access the dataset object based at least in part on the end user being associated with an identifier of an authorized group and a client device of the end user being compliant with at least one pipeline end user policy, the client device of the end user being a computing device through which the attempt to access the AI application is attempted.

15. A system for creating an artificial intelligence (AI) pipeline, comprising:

a memory storage including a non-transitory, computer-readable medium comprising instructions; and

a hardware-based processor that executes the instructions to carry out stages comprising:

receiving a selection of pipeline objects on a user interface (UI), the selected pipeline objects including a dataset object and an AI model, wherein the dataset object includes at least one vector database;

receiving inputs to the UI to arrange the selected pipeline objects in an AI pipeline, including causing an execution linking between the selected dataset object and AI model to be established, wherein the UI visually represents the established execution linking between the pipeline objects, wherein the pipeline objects are moveable by a user within the UI, and wherein the AI pipeline is displayed as an execution flow within the UI,

wherein establishing the execution linking between the selected dataset object and AI model comprises validating that outputs from the selected dataset object can be used as inputs to the AI model;

causing a management policy to be applied to the dataset object, compliance with the management policy being a prerequisite for accessing the dataset object through an execution of the AI pipeline, wherein the UI visually represents the application of the management policy to the dataset object,

wherein enforcing the management policy includes verifying at least one of device location and a device security state based on a profile, and wherein the AI pipeline includes a conditional execution alternative based on noncompliance with the management policy;

causing a simulated execution of the AI pipeline to be displayed within the UI, including:

receiving a query in the UI;

receiving compliance information;

causing the selected pipeline objects to be executed in an order that follows the execution linking displayed within the UI and applies the management policy based on the compliance information; and

causing an output of the simulated execution to be displayed in the UI; and

causing the AI pipeline to be deployed, wherein the deployed AI pipeline is accessible by at least one AI application through a generated endpoint.

16. The system of claim 15 , wherein causing the execution linking between the selected dataset object and AI model to be established comprises causing a pipeline manifest file to be generated which stores the arrangement of pipeline objects in the AI pipeline.

17. The system of claim 16 , the stages further comprising:

validating the pipeline manifest against a dependency rule, wherein the dependency rule dictates an event that must occur before at least one of the selected pipeline objects can execute; and

displaying, within the UI, a validation of the pipeline manifest.

18. The system of claim 15 , wherein the stages further comprise, prior to displaying the selected pipeline objects, determining that a computing device through which the UI is displayed is compliant with at least one pipeline administrative user policy, wherein compliance with the pipeline administrative user policy is determined at least in part on information received from a management agent that executes on the computing device.

19. The system of claim 15 , wherein prior to displaying the selected pipeline objects in the UI, the stages further comprise authenticating the user based on a user identifier and a tenant identifier.

20. The system of claim 15 , wherein the selected pipeline objects are selected from a displayed menu that includes at least one of AI model prompts, datasets, AI models, and executable code objects, and wherein one of the executable code objects is a conditional object that allows user selection of an if-then statement for at least two branches within the AI pipeline.

Assignments (1)

Continuity (2)

References Cited (152)

Cited By (11)

US 12,432,240 US 12,493,473 US 12,493,540 US 12,493,829 US 12,547,677 US 12,683,941 US 12,711,263 US 12,737,492 US 12,744,783 US 12,750,208 US 12,750,368