# US 2025/0094187 A1 · Automation discovery platform

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

- Number: US 2025/0094187 A1
- Document ID: US20250094187A1
- Status: Published application
- Kind: A1
- Application number: 18/368,407
- Filed: 2023-09-14
- Published: 2025-03-20
- Earliest claimed: 2023-09-14
- Assignees: ServiceNow, Inc.
- Applicants: ServiceNow, Inc.
- Inventors: Patrick Ryan; Aparna Mahalingam; Rammohan Narendula; Phani Nivarthi
- Web page: https://patentgrep.com/patent/US20250094187A1

## CPC

- G06F 9/453: {Help systems}
- G06F 40/30: Semantic analysis
- G06F 40/40: Processing or translation of natural language (natural language analysis G06F40/20; semantic analysis G06F40/30)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

In various embodiments, a process for providing an automation discovery platform includes obtaining training data indicating conversation utterances and labels that are associated with the conversation utterances, where each of the labels of at least a subset of the labels corresponds to a virtual agent automation topic. The process includes obtaining a language machine learning model, where the language machine learning model has been trained to a first trained state using unlabeled data. The process includes updating the language machine learning model from the first trained state to a second trained state by applying the training data to the language machine learning model, where updating the language machine learning model includes generating an automation discovery model configured to provide outputs corresponding to virtual agent automation opportunities.

## Claims

### Claim 1

A method, comprising:
obtaining training data indicating conversation utterances and labels that are associated with the conversation utterances, wherein each of the labels of at least a subset of the labels corresponds to a virtual agent automation topic;
obtaining a language machine learning model, wherein the language machine learning model has been trained to a first trained state using unlabeled data; and
updating the language machine learning model from the first trained state to a second trained state by applying the training data to the language machine learning model, wherein updating the language machine learning model includes generating an automation discovery model configured to provide outputs corresponding to virtual agent automation opportunities.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
providing an incident description of an incident record to the automation discovery model; and
suggesting an automation opportunity based on a result of the automation discovery model in response to the incident description.

### Claim 3 (depends on 1)

The method of claim 1, wherein the training data indicating conversation utterances and labels that are associated with the conversation utterances includes at least one of: an application log, a chat conversation, an incident table, or a human resources (HR) case tables.

### Claim 4 (depends on 1)

The method of claim 1, wherein the training data is user-specific and includes data different from the unlabeled data with which the language machine learning model is trained.

### Claim 5 (depends on 1)

The method of claim 1, wherein the automation opportunity includes utilization of a machine learning based tool.

### Claim 6 (depends on 5)

The method of claim 5, further comprising determining a quantitative assessment of using the machine learning based tool, wherein suggesting the automation opportunity includes providing the quantitative assessment.

### Claim 7 (depends on 6)

The method of claim 6, wherein:
the automation opportunity includes at least one use case specific to a user; and
a quantitative assessment of using the machine learning based tool includes potential time savings associated with the at least one use case.

### Claim 8 (depends on 1)

The method of claim 1, wherein each of the labels of at least a subset of the labels corresponds to a natural language intent.

### Claim 9 (depends on 1)

The method of claim 1, further comprising performing language detection on the training data.

### Claim 10 (depends on 1)

The method of claim 1, further comprising generating another automation discovery model, wherein the automation discovery model is associated with a first automation opportunity and the other automation discovery model is associated with a second automation opportunity.

### Claim 11 (depends on 1)

The method of claim 1, further comprising identifying at least one cluster of data from a set of unclassified data, wherein the unclassified data are unclassified with respect to the outputs corresponding to virtual agent automation opportunities.

### Claim 12 (depends on 11)

The method of claim 11, wherein the at least one cluster of data is determined using a clustering model.

### Claim 13 (depends on 11)

The method of claim 11, further comprising determining analytical information associated with the at least one cluster and providing the at least one cluster of data and the associated analytical information.

### Claim 14 (depends on 11)

The method of claim 11, further comprising adding another virtual agent automation topic based at least on the identified at least one cluster of data.

### Claim 15 (depends on 1)

The method of claim 1, further comprising pre-processing the training data including by at least one of: cleaning the training data, applying de-duplication to the training data, measuring at least one statistic with respect to the training data, or removing small punctuation in the training data.

### Claim 16 (depends on 1)

The method of claim 1, wherein the automation opportunity is included in a discovery report.

### Claim 17 (depends on 16)

The method of claim 16, wherein the discovery report includes a plurality of automation opportunities.

### Claim 18 (depends on 16)

The method of claim 16, wherein the discovery report is provided in a graphical user interface.

### Claim 19

A system, comprising:
a processor configured to:
obtain training data indicating conversation utterances and labels that are associated with the conversation utterances, wherein each of the labels of at least a subset of the labels corresponds to a virtual agent automation topic;
obtain a language machine learning model, wherein the language machine learning model has been trained to a first trained state using unlabeled data; and
update the language machine learning model from the first trained state to a second trained state by applying the training data to the language machine learning model, wherein updating the language machine learning model includes generating an automation discovery model configured to provide outputs corresponding to virtual agent automation opportunities; and
a memory coupled to the processor and configured to provide the processor with instructions.

### Claim 20

A computer program product embodied in a non-transitory computer readable medium and comprising computer instructions for:
obtaining training data indicating conversation utterances and labels that are associated with the conversation utterances, wherein each of the labels of at least a subset of the labels corresponds to a virtual agent automation topic;
obtaining a language machine learning model, wherein the language machine learning model has been trained to a first trained state using unlabeled data; and
updating the language machine learning model from the first trained state to a second trained state by applying the training data to the language machine learning model, wherein updating the language machine learning model includes generating an automation discovery model configured to provide outputs corresponding to virtual agent automation opportunities.
