# US 2026/0099493 A1 · Database query method in centralized infrastructure

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

- Number: US 2026/0099493 A1
- Document ID: US20260099493A1
- Status: Published application
- Kind: A1
- Application number: 19/348,142
- Filed: 2025-10-02
- Published: 2026-04-09
- Earliest claimed: 2024-10-04
- Assignees: DoorDash, Inc.
- Applicants: DoorDash, Inc.
- Inventors: Matt Graveen; Brad Busch; Gordon Lee; Joanna Smulska; Zulfikar Imani
- Web page: https://patentgrep.com/patent/US20260099493A1

## CPC

- G06F 16/24542: {Plan optimisation}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A method is disclosed. The method includes receiving, by a server computer from a client computer, a query relating to stored records stored in a database, the query being from a user. The method also includes determining, by the server computer and using a large language model, a query structure from the query, and determining, by the server computer, features useful to answering the query. The method also includes generating, by the server computer, a database query using at least the query structure and the features, retrieving, by the server computer, data from the database using the database query, and outputting, the server computer, an answer to the query to the client computer.

## Claims

### Claim 1

A method comprising:
receiving, by a server computer from a client computer, a query relating to stored records stored in a database, the query being from a user;
determining, by the server computer and using a large language model, a query structure from the query;
determining, by the server computer, features useful to answering the query;
generating, by the server computer, a database query using at least the query structure and the features;
retrieving, by the server computer, data from the database using the database query; and
outputting, the server computer to the client computer, an answer to the query.

### Claim 2 (depends on 1)

The method of claim 1, wherein the server computer comprises query reformulator, and the query structure is a reformulated query produced using the query reformulator.

### Claim 3 (depends on 1)

The method of claim 1, wherein the stored records are accounting records.

### Claim 4 (depends on 1)

The method of claim 1, wherein the server computer comprises a query generation layer, and the method further comprises:
applying guardrails to query before determining the query structure, the query structure being a reformulated query produced using the query generation layer.

### Claim 5 (depends on 4)

The method of claim 4, wherein the query generation layer comprises a selector that calls a first LLM, a generator that calls a second LLM, and a validator.

### Claim 6 (depends on 5)

The method of claim 5, wherein the query generation layer is an SQL query generation layer, and the query is an SQL query.

### Claim 7 (depends on 1)

The method of claim 1, wherein the server computer comprises a feature extractor, and the feature extractor determines the features from the query.

### Claim 8 (depends on 1)

The method of claim 1, wherein the query structure is a prompt template.

### Claim 9 (depends on 1)

The method of claim 1, wherein the query structure is determined using one or more AI agents.

### Claim 10 (depends on 1)

The method of claim 1, wherein the server computer comprises a tool selector, and wherein the method further comprises:
selecting, by the tool selector in the server computer, a selected tool, and wherein the database query is further generated using the selected tool.

### Claim 11

A server computer comprising:
a processor; and
a non-transitory computer readable medium, the non-transitory computer readable medium comprising code executable by the processor, for performing a method comprising:
receiving, from a client computer, a query relating to stored records stored in a database, the query being from a user;
determining, using a large language model, a query structure from the query;
determining features useful to answering the query;
generating a database query using at least the query structure and the features;
retrieving data from the database using the database query; and
outputting an answer to the query to the client computer.

### Claim 12 (depends on 11)

The server computer of claim 11, wherein the server computer comprises query reformulator, and the query structure is a reformulated query produced using the query reformulator.

### Claim 13 (depends on 11)

The server computer of claim 11, wherein the stored records are accounting records.

### Claim 14 (depends on 11)

The server computer of claim 11, wherein the server computer comprises a query generation layer, and the method further comprises:
applying guardrails to query before determining the query structure, the query structure being a reformulated query produced using the query generation layer.

### Claim 15 (depends on 14)

The server computer of claim 14, wherein the query generation layer comprises a selector that calls a first LLM, a generator that calls a second LLM, and a validator.

### Claim 16 (depends on 15)

The server computer of claim 15, wherein the query generation layer is an SQL query generation layer, and the query is an SQL query.

### Claim 17 (depends on 11)

The server computer of claim 11, wherein the server computer comprises a feature extractor, and the feature extractor determines the features from the query.

### Claim 18 (depends on 11)

The server computer of claim 11, wherein the query structure is a prompt template.

### Claim 19 (depends on 11)

The server computer of claim 11, wherein the query structure is determined using one or more AI agents.

### Claim 20 (depends on 1)

The server computer of claim 1, wherein the server computer comprises a tool selector, and wherein the method further comprises:
selecting, by the tool selector in the server computer, a selected tool, and wherein the database query is further generated using the selected tool
