# US 2024/0202225 A1 · Enterprise generative artificial intelligence architecture

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

- Number: US 2024/0202225 A1
- Document ID: US20240202225A1
- Status: Published application
- Kind: A1
- Application number: 18/542,536
- Filed: 2023-12-15
- Published: 2024-06-20
- Earliest claimed: 2022-12-16
- Assignees: C3.ai, Inc.
- Applicants: C3.ai, Inc.
- Inventors: Thomas M. Siebel; Nikhil Krishnan; Louis Poirier; Romain Juban; Michael Haines; Yushi Homma; Riyad Muradov
- Web page: https://patentgrep.com/patent/US20240202225A1

## CPC

- G06F 16/345: {Summarisation for human users}
- G06F 16/3326: {using relevance feedback from the user, e.g. relevance feedback on documents, documents sets, document terms or passages}
- G06F 16/334: Query execution (filtering based on additional data G06F16/335)
- G06F 16/3347: {using vector based model}
- G06F 16/335: Filtering based on additional data, e.g. user or group profiles (filtering in web context G06F16/9535, G06F16/9536)
- G06F 16/338: Presentation of query results
- G06F 40/20: Natural language analysis (semantic analysis of natural language G06F40/30)
- G06F 40/40: Processing or translation of natural language (natural language analysis G06F40/20; semantic analysis G06F40/30)
- G06N 3/045: Combinations of networks
- G06N 3/0475: Generative networks
- G06N 3/092: Reinforcement learning
- G06N 5/04: Inference or reasoning models
- G06N 20/00: Machine learning
- G06N 3/063: using electronic means
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Systems and methods managing, by an orchestrator, a plurality of agents to generate a response to an input. The orchestrator employs one or more multimodal models such as a large language models to process or deconstruct the prompt into a series of instructions for different agents. Each agent employs one or more machine-learning models to process disparate inputs or different portions of an input associated with the prompt. The system generates, by the orchestrator, a natural language summary of the structured and unstructured data records. The system formulates output and transmits the natural language summary of the data records.

## Claims

### Claim 1

A method comprising:
managing, by an orchestrator, a plurality of agents to generate a response to an input, wherein the orchestrator employs one or more multimodal models to process or deconstruct a prompt into a series of instructions for different agents, wherein each agent employs one or more machine-learning models to process disparate inputs or different portions of an input associated with the prompt;
instructing, by the orchestrator, retrieval requests related to the input to the one or more agents of the plurality of agents;
receiving, from the one or more agents of the plurality of agents, data from multiple data domains based on instructions from the orchestrator;
analyzing, by the orchestrator, the received data to formulate one or more responses to the prompt, wherein the orchestrator provides additional retrieval requests to the one or more agents to retrieve additional data to satisfy a context validation criteria associated with the input; and
outputting, by the orchestrator, at least one of a validated response of the one or more responses to the input that satisfies context validation criteria and a portion of data retrieved by the one or more agents related to the input.

### Claim 2 (depends on 1)

The method of claim 1, wherein the orchestrator generates intermediate instructions to the plurality of agents, wherein intermediate instructions comprise at least one of portions of the input, questions about the input generated by the one or more multimodal models, and follow-up questions about answers generated by the one or more multimodal models.

### Claim 3 (depends on 1)

The method of claim 1, wherein outputting the portion of data retrieved by the one or more agents related to the input includes a source citation for the at least a portion of the validated response.

### Claim 4 (depends on 1)

The method of claim 1, wherein the context validation criteria includes a threshold for identifying source material from an enterprise data system that corroborate the response.

### Claim 5 (depends on 1)

The method of claim 1, wherein managing the plurality of agents comprises iterative processing or multiple instructions from the orchestrator.

### Claim 6 (depends on 1)

The method of claim 1, wherein the retrieving the data from multiple data domains includes time series data, structured data, and unstructured data.

### Claim 7 (depends on 2)

The method of claim 2, wherein at least one agent of the one or more agents instantiates a tool to perform an operation on the instruction, the retrieved data, or the intermediate instructions.

### Claim 8 (depends on 7)

The method of claim 7, wherein the operation includes at least one of calculation, translation, formatting, visualization.

### Claim 9 (depends on 1)

The method of claim 1, wherein the one or more agents are trained on different domain specific machine-learning models.

### Claim 10 (depends on 1)

The method of claim 1, wherein at least one agent employs a type system to unify incompatible data from disparate data sources.

### Claim 11

A method comprising:
selecting, by an orchestrator based on pre-processed input generated from an initial input by the orchestrator, one or more agent of a plurality of different agents;
retrieving, by the selected one or more agent, data records from an unstructured dataset and additional data records from a structured dataset;
generating, by the orchestrator, a natural language summary of the data records from the unstructured dataset and the additional data records from the structured dataset; and
transmitting the natural language summary of the one or more data records and the one or more additional data records as a response to the input.

### Claim 12 (depends on 1)

The method of claim 1, wherein the orchestrator comprises one or more multimodal models that includes at least a large language models.

### Claim 13 (depends on 11)

The method of claim 11, wherein the agent comprises one or more multimodal models, and an additional agent of the plurality of different agents comprises one or more additional multimodal models.

### Claim 14 (depends on 13)

The method of claim 13, wherein the agent implements an iterative generative artificial intelligence process using one or more unstructured data retrieval tools to retrieve the one or more data records from an unstructured dataset.

### Claim 15 (depends on 13)

The method of claim 13, wherein the additional agent implements a non-iterative generative artificial intelligence process using one or more structured data retrieval tools to retrieve the one or more additional data records from a structured dataset.

### Claim 16 (depends on 14)

The method of claim 14, wherein the agent instructs the one or more unstructured data retrieval tools based upon embeddings in a vector store.

### Claim 17 (depends on 15)

The method of claim 15, wherein the additional agent instructs the one or more structured data retrieval tools based upon a data model describing relationships of one or more types of the data model.

### Claim 18 (depends on 13)

The method of claim 13, wherein the selecting, by the orchestrator based on the processed input, the agent of the plurality of different agents, further comprises generating intermediate input based on a first portion of the input and routing the intermediate input to the agent.

### Claim 19 (depends on 18)

The method of claim 18, wherein the selecting, by the orchestrator based on the processed input, the additional agent of the plurality of different agents, further comprises generating a third input based on a second portion of the input and routing the third input to the additional agent.

### Claim 20 (depends on 13)

The method of claim 13, wherein the agent and the additional agent execute in parallel and perform their respective retrievals in parallel.
