# US 2026/0064370 A1 · Automated efficiency determination of components prior to deployment to artificial intelligence production pipeline

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0064370 A1
- Document ID: US20260064370A1
- Status: Published application
- Kind: A1
- Application number: 18/817,290
- Filed: 2024-08-28
- Published: 2026-03-05
- Earliest claimed: 2024-08-28
- Assignees: The Toronto-Dominion Bank
- Applicants: The Toronto-Dominion Bank
- Inventors: Peter Starszyk; Devinder Kumar; Maksims Volkovs
- Web page: https://patentgrep.com/patent/US20260064370A1

## CPC

- G06F 8/10: Requirements analysis; Specification techniques
- G06F 8/20: Software design
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

An example operation may include one or more of storing utility components for a production environment within a storage, receiving, via a software application, an AI model from a development environment, receiving a configuration file defining a configuration of a pipeline in the production environment which includes the AI model, generating, via the software application, a production pipeline of the AI model which includes a sequence of components including the AI model and at least one utility component based on the configuration file, determining whether the production pipeline satisfies at least one regulatory requirement based on the sequence of components, in response to the production pipeline satisfying the at least one regulatory requirement, executing the production pipeline on input data via the software application in the production environment and may further include an AI agent updating the configuration file of the production pipeline satisfying the at least one regulatory requirement.

## Claims

### Claim 1

An apparatus, comprising:
a memory; and
a processor, wherein the processor and the memory are communicatively coupled, the processor configured to:
store utility components for a production environment within the memory,
receive, via a software application, an artificial intelligence (AI) model from a development environment,
receive a configuration file defining a configuration of a pipeline in the production environment which includes the AI model,
generate, via the software application, a production pipeline of the AI model which includes a sequence of components including the AI model and at least one utility component from the utility components based on the configuration file,
determine whether the production pipeline satisfies at least one regulatory requirement based on the sequence of components, and
in response to the production pipeline satisfying the at least one regulatory requirement, execute the production pipeline of the AI model on input data via the software application in the production environment.

### Claim 2 (depends on 1)

The apparatus of claim 1, wherein the processor is further configured to train a second AI model using a neural network capability based on at least one of components of previous pipelines, source code of the previous pipelines, and model feedback data, and execute the second AI model on source code of the AI model to determine the at least one utility component.

### Claim 3 (depends on 2)

The apparatus of claim 2, wherein the processor is further configured to receive feedback about the at least one utility component in the production pipeline of the AI model via a graphical user interface of the software application, generate a model feedback record based on the feedback and the at least one utility component, add the model feedback record to the model feedback data, and retrain the second AI model based on the model feedback data with the model feedback record added thereto.

### Claim 4 (depends on 1)

The apparatus of claim 1, wherein the processor is configured to determine the at least one utility component from among the utility components to include with the AI model based on a model type of the AI model and model parameters of the AI model.

### Claim 5 (depends on 1)

The apparatus of claim 1, wherein the processor is further configured to wrap the AI model with an envelope to generate a wrapped AI model, and generate the production pipeline with the wrapped AI model interspersed among the sequence of components.

### Claim 6 (depends on 1)

The apparatus of claim 1, wherein the processor is configured to determine whether the production pipeline satisfies the at least one regulatory requirement based on at least one of bias checking, data scrubbing, and regression testing being included within the sequence of components of the production pipeline.

### Claim 7 (depends on 1)

The apparatus of claim 1, wherein, in response to the production pipeline not satisfying the at least one regulatory requirement, the processor is further configured to pause the production pipeline and display a warning via a graphical user interface of the software application, wherein an AI agent updates the configuration file of the production pipeline satisfying the at least one regulatory requirement.

### Claim 8

A method comprising:
storing utility components for a production environment within a storage;
receiving, via a software application, an artificial intelligence (AI) model from a development environment;
receiving a configuration file defining a configuration of a pipeline in the production environment which includes the AI model;
generating, via the software application, a production pipeline of the AI model which includes a sequence of components including the AI model and at least one utility component from the utility components based on the configuration file;
determining whether the production pipeline satisfies at least one regulatory requirement based on the sequence of components; and
in response to the production pipeline satisfying the at least one regulatory requirement, executing the production pipeline of the AI model on input data via the software application in the production environment.

### Claim 9 (depends on 8)

The method of claim 8, further comprising training a second AI model using a neural network capability based on at least one of components of previous pipelines, source code of the previous pipelines, and model feedback data, and executing the second AI model on source code of the AI model to determine the at least one utility component.

### Claim 10 (depends on 9)

The method of claim 9, further comprising receiving feedback about the at least one utility component in the production pipeline of the AI model via a graphical user interface of the software application, generating a model feedback record based on the feedback and the at least one utility component, adding the model feedback record to the model feedback data, and retraining the second AI model based on the model feedback data with the model feedback record added thereto.

### Claim 11 (depends on 8)

The method of claim 8, wherein the generating comprises determining the at least one utility component from among the utility components to include with the AI model based on a model type of the AI model and model parameters of the AI model.

### Claim 12 (depends on 8)

The method of claim 8, further comprising wrapping the AI model with an envelope to generate a wrapped AI model, wherein the generating comprises generating the production pipeline with the wrapped AI model interspersed among the sequence of components.

### Claim 13 (depends on 8)

The method of claim 8, wherein the determining comprises determining whether the production pipeline satisfies the at least one regulatory requirement based on at least one of bias checking, data scrubbing, and regression testing being included within the sequence of components of the production pipeline.

### Claim 14 (depends on 8)

The method of claim 8, wherein, in response to the production pipeline not satisfying the at least one regulatory requirement, the method further comprises pausing the production pipeline and displaying a warning via a graphical user interface of the software application, wherein an AI agent updates the configuration file of the production pipeline satisfying the at least one regulatory requirement.

### Claim 15

A computer-readable storage medium comprising instructions which when executed by a computer cause a processor to perform:
storing utility components for a production environment within a storage;
receiving, via a software application, an artificial intelligence (AI) model from a development environment;
receiving a configuration file defining a configuration of a pipeline in the production environment which includes the AI model;
generating, via the software application, a production pipeline of the AI model which includes a sequence of components including the AI model and at least one utility component from the utility components based on the configuration file;
determining whether the production pipeline satisfies at least one regulatory requirement based on the sequence of components; and
in response to the production pipeline satisfying the at least one regulatory requirement, executing the production pipeline of the AI model on input data via the software application in the production environment.

### Claim 16 (depends on 15)

The computer-readable storage medium of claim 15, wherein the processor is further configured to perform training a second AI model using a neural network capability based on at least one of components of previous pipelines, source code of the previous pipelines, and model feedback data, and executing the second AI model on source code of the AI model to determine the at least one utility component.

### Claim 17 (depends on 16)

The computer-readable storage medium of claim 16, wherein the processor is further configured to perform receiving feedback about the at least one utility component in the production pipeline of the AI model via a graphical user interface of the software application, generating a model feedback record based on the feedback and the at least one utility component, adding the model feedback record to the model feedback data, and retraining the second AI model based on the model feedback data with the model feedback record added thereto.

### Claim 18 (depends on 15)

The computer-readable storage medium of claim 15, wherein the generating comprises determining the at least one utility component from among the utility components to include with the AI model based on a model type of the AI model and model parameters of the AI model.

### Claim 19 (depends on 15)

The computer-readable storage medium of claim 15, wherein the processor is further configured to perform wrapping the AI model with an envelope to generate a wrapped AI model, wherein the generating comprises generating the production pipeline with the wrapped AI model interspersed among the sequence of components.

### Claim 20 (depends on 15)

The computer-readable storage medium of claim 15, wherein the determining comprises determining whether the production pipeline satisfies the at least one regulatory requirement based on at least one of bias checking, data scrubbing, and regression testing being included within the sequence of components of the production pipeline.
