Granted Patent B1

US 12,423,069 · App. 19/227,007 · Granted Sep 23, 2025

Inventors: John Morello (Baton Rouge, LA); Ben Bernstein (New York, NY); Dima Stopel (Herzliya, IL)

Assignee: MINIMUS LTD

G06F8/35G06F8/10G06F8/20G06F8/75G06F8/77G06F11/328G06F11/3684G06F11/3688G06F11/3698G06F21/554G06F21/577G06N3/0475G06N3/08G06N3/0895G06N20/00

Abstract

Claims (43)

1. A method for building software images, comprising:

generating a prompt based on an input portion of application-identifying data;

prompting a generative artificial intelligence model using the prompt in order to output text indicating a plurality of software packages, wherein each software package is a set of code for performing a plurality of functions of an application;

generating a software image recipe based on the output text, wherein the software image recipe is a file including a set of instructions that cause a processing circuitry to build the software image by combining the plurality of software packages when executed; and

building the software image by executing the software image recipe via the processing circuitry, wherein the software image causes deployment of a software component when executed, wherein the software component is configured to perform the plurality of functions of the application.

2. The method of claim 1 , wherein the set of instructions further indicates a location of the set of code of each software package.

3. The method of claim 1 , further comprising:

providing the generative artificial intelligence model access to a tool having at least one function, wherein the generative artificial intelligence model calls the at least one function of the tool in order to obtain data of the application, wherein the output text is generated by the generative artificial intelligence model based on the data of the application obtained by calling the at least one function of the tool.

4. The method of claim 3 , wherein providing the generative artificial intelligence model access to the tool further comprises:

prompting the generative artificial intelligence model using text defining at least one function call for the tool, wherein the generative artificial intelligence model calls the at least one function of the tool using the at least one function call in order to obtain the data of the application.

5. The method of claim 1 , wherein each software package includes code of a subset of the plurality of functions of the application.

6. The method of claim 1 , wherein the generative artificial intelligence model is a language model.

7. The method of claim 1 , further comprising:

training the generative artificial intelligence model using a supervised machine learning process based on a labeled training data set including training inputs and corresponding training outputs, wherein the training inputs include a plurality of portions of training application-identifying data, wherein the training outputs include a plurality of corresponding labels representing respective software image recipe text outputs for the training inputs.

8. The method of claim 1 , further comprising:

deploying the software component by executing the software image.

9. The method of claim 8 , further comprising:

redeploying the software component when a cyber threat is detected with respect to the software component, wherein redeploying the software component further comprises rebuilding the software image using an updated version of at least one software package of the plurality of software packages.

10. A non-transitory computer readable medium having stored thereon instructions for causing a processing circuitry to execute a process, the process comprising:

generating a prompt based on an input portion of application-identifying data;

prompting a generative artificial intelligence model using the prompt in order to output text indicating a plurality of software packages, wherein each software package is a set of code for performing a plurality of functions of an application;

generating a software image recipe based on the output text, wherein the software image recipe is a file including a set of instructions that cause a processing circuitry to build the software image by combining the plurality of software packages when executed; and

building the software image by executing the software image recipe via the processing circuitry, wherein the software image causes deployment of a software component when executed, wherein the software component is configured to perform the plurality of functions of the application.

11. A system for building software images, comprising:

a processing circuitry; and

a memory, the memory containing instructions that, when executed by the processing circuitry, configure the system to:

generate a prompt based on an input portion of application-identifying data;

prompt a generative artificial intelligence model using the prompt in order to output text indicating a plurality of software packages, wherein each software package is a set of code for performing a plurality of functions of an application;

generate a software image recipe based on the output text, wherein the software image recipe is a file including a set of instructions that cause a processing circuitry to build the software image by combining the plurality of software packages when executed; and

build the software image by executing the software image recipe via the processing circuitry, wherein the software image causes deployment of a software component when executed, wherein the software component is configured to perform the plurality of functions of the application.

12. The system of claim 11 , wherein the set of instructions further indicates a location of the set of code of each software package.

13. The system of claim 11 , wherein the system is further configured to:

provide the generative artificial intelligence model access to a tool having at least one function, wherein the generative artificial intelligence model calls the at least one function of the tool in order to obtain data of the application, wherein the output text is generated by the generative artificial intelligence model based on the data of the application obtained by calling the at least one function of the tool.

14. The system of claim 13 , wherein the system is further configured to:

prompt the generative artificial intelligence model using text defining at least one function call for the tool, wherein the generative artificial intelligence model calls the at least one function of the tool using the at least one function call in order to obtain the data of the application.

15. The system of claim 11 , wherein each software package includes code of a subset of the plurality of functions of the application.

16. The system of claim 11 , wherein the generative artificial intelligence model is a language model.

17. The system of claim 11 , wherein the system is further configured to:

train the generative artificial intelligence model using a supervised machine learning process based on a labeled training data set including training inputs and corresponding training outputs, wherein the training inputs include a plurality of portions of training application-identifying data, wherein the training outputs include a plurality of corresponding labels representing respective software image recipe text outputs for the training inputs.

18. The system of claim 11 , wherein the system is further configured to:

deploy the software component by executing the software image.

19. The system of claim 18 , wherein the system is further configured to:

redeploy the software component when a cyber threat is detected with respect to the software component, wherein redeploying the software component further comprises rebuilding the software image using an updated version of at least one software package of the plurality of software packages.

Assignments (2)

References Cited (38)