# US 2026/0064103 A1 · Systems and methods for device advisors

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

- Number: US 2026/0064103 A1
- Document ID: US20260064103A1
- Status: Published application
- Kind: A1
- Application number: 19/267,184
- Filed: 2025-07-11
- Published: 2026-03-05
- Earliest claimed: 2024-09-04
- Assignees: Sensia Netherlands B.V.
- Applicants: Sensia Netherlands B.V.
- Inventors: Jeffery P. Anderson; Thony Brito Cardier; Srikanth G. Mashetty; Norman Andrew Weatherhead; Harshkumar Patel; Aquib Mustafa; Ryan McDonald; Jonathan Wun Shiung Chong
- Web page: https://patentgrep.com/patent/US20260064103A1

## CPC

- G05B 19/41835: {characterised by program execution}
- G06F 16/903: Querying (for retrieval from the web G06F16/953)
- G05B 13/0265: {the criterion being a learning criterion}
- G06F 16/3329: Natural language query formulation
- G06N 3/044: Recurrent networks, e.g. Hopfield networks
- G06N 3/045: Combinations of networks
- G06N 3/08: Learning methods
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06N 20/00: Machine learning
- G06N 20/20: Ensemble learning
- G05B: CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS (systems for controlling or regulating non-electric variables G05D; systems for regulating electric or magnetic variables G05F; control devices or systems insofar as characterised by mechanical features only G05G)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A system includes an edge device and a computing platform. The computing platform includes one or more processors and one or more non-transitory computer-readable media storing program instructions that cause the one or more processors to perform operations including fine-tuning, a plurality of advisor models to perform different types of advisor operations for an oil-and-gas facility using data relating to oil-and-gas equipment, sorting, a user query to a first advisor model by selecting, by at least one machine learning model, the first advisor model from the plurality of advisor models based on a content of the user query, generating, by the first advisor model, an actionable response to the user query, and causing the oil-and-gas facility to operate in accordance with the actionable response.

## Claims

### Claim 1

A system, comprising:
an edge device; and
a computing platform configured to assist operation of the edge device comprising a user interface, the computing platform comprising one or more processors and one or more non-transitory computer-readable media storing program instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
fine-tuning a plurality of advisor models to perform different types of advisor operations for an oil-and-gas facility using data relating to oil-and-gas equipment;
in response to receiving a user query via the user interface, sorting the user query to a first advisor model by selecting, by at least one machine learning model, the first advisor model from the plurality of advisor models based on a content of the user query;
generating, by the first advisor model, an actionable response to the user query; and
causing the oil-and-gas facility to operate in accordance with the actionable response.

### Claim 2 (depends on 1)

The system of claim 1, wherein the plurality of advisor models comprise a system advisor model, the system advisor model comprising a machine learning model configured to perform advisor operations comprising configuring the edge device and assisting in developing applications for the edge device.

### Claim 3 (depends on 1)

The system of claim 1, wherein the plurality of advisor models comprise an operational advisor model, the operational advisor model comprising a machine learning model configured to perform advisor operations comprising proposing improvements to and monitoring the edge device.

### Claim 4 (depends on 1)

The system of claim 1, wherein the plurality of advisor models comprise a domain advisor model, the domain advisor model comprising a machine learning model configured to perform advisor operations comprising providing training for the edge device to users, the machine learning model being continuously updated based on feedback from the training.

### Claim 5 (depends on 1)

The system of claim 1, wherein the at least one machine learning model is a context aware multimodal large language model (LLM) and the user query comprises at least one of a free-text user query, a visual user query, or an audio user query.

### Claim 6 (depends on 5)

The system of claim 5, wherein the user interface comprises a text interface, the at least one machine learning model to provide prompts to a user and receive the free-text user query via the text interface.

### Claim 7 (depends on 5)

The system of claim 5, wherein the user interface comprises a visual interface, the at least one machine learning model to provide prompts to a user and receive the visual user query via the visual interface.

### Claim 8 (depends on 5)

The system of claim 5, wherein the user interface comprises an audio interface, the at least one machine learning model to provide prompts to a user and receive the audio user query via the audio interface.

### Claim 9 (depends on 1)

The system of claim 1, wherein causing the oil-and-gas facility to operate in accordance with the actionable response comprises changing, based on the actionable response, a setting used by the edge device and controlling, by the edge device, equipment of the oil-and-gas facility using the setting.

### Claim 10 (depends on 1)

The system of claim 1, wherein the actionable response comprises parameter modifications for the edge device, the parameter modifications based on hyperparameters of the first advisor model.

### Claim 11

A method, comprising:
fine-tuning a plurality of advisor models to perform different types of advisor operations for an oil-and-gas facility using data relating to oil-and-gas equipment;
receiving a user query via a user interface;
sorting the user query to a first advisor model by selecting, by at least one machine learning model, the first advisor model from the plurality of advisor models based on content of the user query;
generating, by the first advisor model, an actionable response to the user query; and
operating, an edge device of the oil-and-gas facility, in accordance with the actionable response.

### Claim 12 (depends on 11)

The method of claim 11, wherein the plurality of advisor models comprise a system advisor model, the system advisor model comprising a machine learning model configured to perform advisor operations comprising configuring the edge device and assisting in developing applications for the edge device.

### Claim 13 (depends on 11)

The method of claim 11, wherein the plurality of advisor models comprise an operational advisor model, the operational advisor model comprising a machine learning model configured to perform advisor operations comprising proposing improvements to and monitoring the edge device.

### Claim 14 (depends on 11)

The method of claim 11, wherein the plurality of advisor models comprise a domain advisor model, the domain advisor model comprising a machine learning model configured to perform advisor operations comprising providing training for the edge device to users, the machine learning model being continuously updated based on feedback from the training.

### Claim 15 (depends on 11)

The method of claim 11, wherein the at least one machine learning model is a context aware multimodal large language model (LLM) and the user query comprises at least one of a free-text user query, a visual user query, or an audio user query.

### Claim 16 (depends on 15)

The method of claim 15, wherein the user interface comprises at least one of:
a text interface, the at least one machine learning model to provide prompts to a user and receive the free-text user query via the text interface,
a visual interface, the at least one machine learning model to provide the prompts to the user and receive the visual user query via the visual interface, or
an audio interface, the at least one machine learning model to provide the prompts to the user and receive the audio user query via the audio interface.

### Claim 17 (depends on 11)

The method of claim 11, wherein to cause the oil-and-gas facility to operate in accordance with the actionable response, the method further comprises:
changing, based on the actionable response, a setting used by the edge device; and
controlling, by the edge device, equipment of the oil-and-gas facility using the setting.

### Claim 18 (depends on 11)

The method of claim 11, wherein the actionable response comprises parameter modifications for the edge device, the parameter modifications based on hyperparameters of the first advisor model.

### Claim 19

A system, comprising:
an edge device;
a user interface; and
a computing platform configured to assist operation of the edge device, the computing platform comprising one or more processors and one or more non-transitory computer-readable media storing program instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
in response to receiving a user query from the user interface, providing, using at least one machine learning model, the user query to an advisor model of a plurality of advisor models based at least partially on a content of the user query, the plurality of advisor models to perform different types of advisor operations for an oil-and-gas facility using data relating to oil-and-gas equipment;
generating, by the advisor model, an actionable response to the user query; and
causing the user interface to display the actionable response.

### Claim 20 (depends on 19)

The system of claim 19, wherein the actionable response includes at least directions to a user.
