# US 2026/0093740 A1 · Systems and methods for providing recommendations to cloud service requests

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

- Number: US 2026/0093740 A1
- Document ID: US20260093740A1
- Status: Published application
- Kind: A1
- Application number: 18/902,392
- Filed: 2024-09-30
- Published: 2026-04-02
- Earliest claimed: 2024-09-30
- Assignees: Sap SE
- Applicants: Sap SE
- Inventors: Robert LOWSKI; Florian FRICKE; Varsha SHETTY
- Web page: https://patentgrep.com/patent/US20260093740A1

## CPC

- G06F 16/3347: {using vector based model}
- G06F 16/383: using metadata automatically derived from the content
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Described herein are techniques for processing cloud service requests. Cloud service providers may allow users to submit service requests to perform functions on the cloud. Most service requests may be based on service request templates available in a service catalog. However, users may be unable to find what they would like to do in the service catalog and therefore may submit an assisted service request that includes text description of what the user is trying to do. These assisted service requests may be analyzed and recommended service request templates may be provided from the service catalog.

## Claims

### Claim 1

A method, comprising:
applying a plurality of service request templates of a service catalog to a large language embedding model to generate a vector database;
receiving an assisted service request containing text in natural language, wherein the text describes a customer service request associated with a cloud service;
applying the text contained in the assisted service request to the large language embedding model to generate a vector representation from the text;
performing a similarity search on the vector representation in the vector database, wherein the similarity search returns at least one vector;
retrieving, based on the at least one vector returned by the similarity search, at least one service request template from the plurality of service request templates of the service catalog; and
presenting the at least one service request template as a suggestion for the assisted service request.

### Claim 2 (depends on 1)

The method as in claim 1, wherein the vector database is based on a plurality of software request templates in the service catalog.

### Claim 3 (depends on 2)

The method as in claim 2, wherein each software request template in the plurality of software request templates includes a description in natural language text describing the corresponding software request template and the vector database is generated from applying a word embedding technique to the description of the plurality of software request templates.

### Claim 4 (depends on 1)

The method as in claim 1, wherein generating the vector representation from the text includes applying a word embedding technique to the text.

### Claim 5 (depends on 1)

The method as in claim 1, wherein the at least one service request template is presented as a suggestion in a pop up window.

### Claim 6 (depends on 1)

The method as in claim 1, wherein a first service request template is presented in a first window and a second service request template is presented in a second window.

### Claim 7 (depends on 6)

The method as in claim 6, wherein the first window includes an icon configured to indicate that the first service request template has a higher similarity score than the second service request template.

### Claim 8

A system comprising:
one or more processors;
a non-transitory computer-readable medium storing a program executable by the one or more processors, the program comprising sets of instructions for:
applying a plurality of service request templates of a service catalog to a large language embedding model to generate a vector database;
receiving an assisted service request containing text in natural language, wherein the text describes a customer service request associated with a cloud service;
applying the text contained in the assisted service request to the large language embedding model to generate a vector representation from the text;
performing a similarity search on the vector representation in the vector database, wherein the similarity search returns at least one vector;
retrieving, based on the at least one vector returned by the similarity search, at least one service request template from the plurality of service request templates of the service catalog; and
presenting the at least one service request template as a suggestion for the assisted service request.

### Claim 9 (depends on 8)

The system of claim 8, wherein the vector database is based on a plurality of software request templates in the service catalog.

### Claim 10 (depends on 9)

The system of claim 9, wherein each software request template in the plurality of software request templates includes a description in natural language text describing the corresponding software request template and the vector database is generated from applying a word embedding technique to the description of the plurality of software request templates.

### Claim 11 (depends on 8)

The system of claim 8, wherein generating the vector representation from the text includes applying a word embedding technique to the text.

### Claim 12 (depends on 8)

The system of claim 8, wherein the at least one service request template is presented as a suggestion in a pop up window.

### Claim 13 (depends on 8)

The system of claim 8, wherein a first service request template is presented in a first window and a second service request template is presented in a second window.

### Claim 14 (depends on 13)

The system of claim 13, wherein the first window includes an icon configured to indicate that the first service request template has a higher similarity score than the second service request template.

### Claim 15

A non-transitory computer-readable medium storing a program executable by one or more processors, the program comprising sets of instructions for:
applying a plurality of service request templates of a service catalog to a large language embedding model to generate a vector database;
receiving an assisted service request containing text in natural language, wherein the text describes a customer service request associated with a cloud service;
applying the text contained in the assisted service request to the large language embedding model to generate a vector representation from the text;
performing a similarity search on the vector representation in the vector database, wherein the similarity search returns at least one vector;
retrieving, based on the at least one vector returned by the similarity search, at least one service request template from the plurality of service request templates of the service catalog; and
presenting the at least one service request template as a suggestion for the assisted service request.

### Claim 16 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein the vector database is based on a plurality of software request templates in the service catalog.

### Claim 17 (depends on 16)

The non-transitory computer-readable medium of claim 16, wherein each software request template in the plurality of software request templates includes a description in natural language text describing the corresponding software request template and the vector database is generated from applying a word embedding technique to the description of the plurality of software request template.

### Claim 18 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein generating the vector representation from the text includes applying a word embedding technique to the text.

### Claim 19 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein a first service request template is presented in a first window and a second service request template is presented in a second window.

### Claim 20 (depends on 19)

The non-transitory computer-readable medium of claim 19, wherein the first window includes an icon configured to indicate that the first service request template has a higher similarity score than the second service request template.
