# US 2025/0378301 A1 · Agent-based modeler using dynamic model-parameter and constraint generation

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

- Number: US 2025/0378301 A1
- Document ID: US20250378301A1
- Status: Published application
- Kind: A1
- Application number: 19/234,101
- Filed: 2025-06-10
- Published: 2025-12-11
- Earliest claimed: 2024-06-10
- Assignees: Aaru, Inc.
- Applicants: Aaru, Inc.
- Inventors: Edward Koh; Cameron Fink; John Kessler
- Web page: https://patentgrep.com/patent/US20250378301A1

## CPC

- G06N 3/006: based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
- G06F 9/451: Execution arrangements for user interfaces
- G06F 30/27: using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
- G06N 3/126: Evolutionary algorithms, e.g. genetic algorithms or genetic programming
- G06N 20/00: Machine learning
- B60W 60/001: {Planning or execution of driving tasks}
- B60W 60/0011: {involving control alternatives for a single driving scenario, e.g. planning several paths to avoid obstacles}
- G06N 20/20: Ensemble learning
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- B60W: CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT

## Abstract

Techniques for simulating multi-agent interactions within a virtual world in response to generated hypotheses using dynamically retrieved external data are disclosed herein. A computing system can receive instructions including a user input with a seed phrase and provide the user input to a configuration generation model to generate agent traits associated with agents in a virtual world and/or data-source identifiers associated with external data sources. The system can retrieve external data from the external data sources and provide the agent traits and external data to a hypothesis generation model to generate a hypothesis that includes a constraint associated with the external data. The system can instantiate agents that have the agent traits and execute a simulation session consistent with the constraint (e.g., by causing the agents to generate an output set using the hypothesis and question). The system can generate a representation of the output set for analysis.

## Claims

### Claim 1

A computer-implemented method for a multi-agent simulator, the method comprising:
receiving, via a graphical user interface (GUI) of a computing device, instructions comprising a user input including a seed phrase;
inputting the user input into a configuration generation model to generate (1) a set of agent traits of a set of agents in a virtual world associated with a physical configuration and (2) a set of data-source identifiers of a set of external data sources;
instantiating the virtual world comprising the set of agents;
in response to generating the set of agent traits of the set of agents in the virtual world, instantiating the set of agents including a first agent associated with the set of agent traits,
wherein each agent of the set of agents is associated with one or more machine-learning models, and
wherein each agent of the set of agents is associated with memory, one or more processors, and one or more agent interfaces configured to execute at least one action;
in response to generating the set of data-source identifiers of the set of external data sources, retrieving external data from one or more external data sources of the set of external data sources;
inputting the set of agent traits and the external data into a hypothesis generation model to generate a first hypothesis comprising a first question based on the external data,
wherein the first hypothesis includes a first constraint associated with the external data, and
wherein the first question is associated with the set of agents associated with the set of agent traits;
executing a simulation session in the virtual world, consistent with the first constraint, by causing the set of agents to generate, using the first hypothesis and the first question, a first output set; and
generating, for display on the GUI, a representation of the first output set.

### Claim 2 (depends on 1)

The method of claim 1, wherein instantiating the set of agents using the set of agent traits comprises:
generating, using the configuration generation model, the set of agent traits including at least two of: a processor identifier, an operating system identifier, a hardware component identifier, or a network configuration identifier;
generating, for a first agent of the set of agents, the set of agent traits;
modifying a first agent trait of the set of agent traits to generate a second set of agent traits associated with a second agent of the set of agents; and
instantiating the first agent associated with the set of agent traits and the second agent associated with the second set of agent traits.

### Claim 3 (depends on 1)

The method of claim 1, wherein retrieving the external data from the one or more external data sources comprises:
determining, based on the set of data-source identifiers, a first uniform resource locator associated with a first external data source of the one or more external data sources,
wherein the first external data source is associated with cybersecurity vulnerability data, economic data, news data, or social media feed data; and
retrieving, using the first uniform resource locator and an application programming interface (API), a first dataset associated with the first external data source.

### Claim 4 (depends on 3)

The method of claim 3, wherein generating the first hypothesis including the first constraint comprises:
providing the first dataset associated with the first external data source to a constraint generation model to generate the first constraint,
wherein the first constraint includes at least one of: an indication of a cybersecurity threat, an economic indicator, an environmental indicator, a regulatory requirement, or an agent-interaction constraint; and
generating a first data schema, associated with the first hypothesis,
wherein the first data schema includes an indication of the first constraint.

### Claim 5 (depends on 1)

The method of claim 1, wherein executing the simulation session, consistent with the first constraint, comprises:
generating a first data schema including a representation of the first constraint;
providing the first data schema and the first question to the first agent of the set of agents;
in response to providing the first data schema and the first question to the first agent, generating a first output for the first output set.

### Claim 6 (depends on 1)

The method of claim 1, wherein generating the first hypothesis comprises:
providing the set of agent traits and the external data to the hypothesis generation model to generate a set of questions based on the external data;
generating, for display on the GUI, representations of the set of questions;
in response to generating, for display on the GUI, the representations of the set of questions, receiving, via the GUI, a user selection of the first question of the set of questions; and
generating the first hypothesis comprising the first question associated with the user selection.

### Claim 7 (depends on 1)

The method of claim 1, wherein generating the first hypothesis comprises:
generating, for display on the GUI, a representation of a preliminary question;
in response to generating, for display on the GUI, the representation of the preliminary question, receiving, via the GUI, the first question,
wherein the first question is a modified version of the preliminary question; and
generating the first hypothesis including the first question.

### Claim 8 (depends on 1)

The method of claim 1, wherein generating the first hypothesis comprising the first question comprises:
determining, based on the external data, a first value associated with the first constraint;
generating the first question including an indication of the first value associated with the first constraint; and
generating the first hypothesis including the first question.

### Claim 9

One or more non-transitory, computer-readable media having computer-executable instructions thereon for generating hypotheses for multiagent simulations in response to dynamically-retrieved external data, the instructions, when executed by one or more processors, causing a computing system to perform operations comprising:
receiving instructions comprising a user input including a seed phrase;
inputting the user input into a configuration generation model to generate a set of agent traits of a set of agents in a virtual world;
instantiating the virtual world comprising the set of agents;
in response to generating the set of agent traits of the set of agents in the virtual world, instantiating the set of agents including a first agent associated with the set of agent traits;
retrieving external data from one or more external data sources,
inputting the set of agent traits and the external data into a hypothesis generation model to generate a first hypothesis comprising a first question based on the external data,
wherein the first hypothesis includes a first constraint associated with the external data, and
wherein the first question is associated with the set of agents associated with the set of agent traits;
executing a simulation session in the virtual world, consistent with the first constraint, by causing the set of agents to generate, using the first hypothesis and the first question, a first output set; and
generating a representation of the first output set.

### Claim 10 (depends on 9)

The one or more non-transitory, computer-readable media of claim 9, wherein the instructions for instantiating the set of agents using the set of agent traits cause the computing system to perform operations comprising:
generating, using the configuration generation model, the set of agent traits including at least two of: a role indicator, an institution identifier, an age value, a salary value, a residence address identifier, a political affiliation identifier, and an educational attainment identifier;
generating, for a first agent of the set of agents, the set of agent traits;
modifying a first agent trait of the set of agent traits to generate a second set of agent traits associated with a second agent of the set of agents; and
instantiating the first agent associated with the set of agent traits and the second agent associated with the second set of agent traits.

### Claim 11 (depends on 9)

The one or more non-transitory, computer-readable media of claim 9, wherein the instructions for retrieving the external data from the one or more external data sources cause the computing system to perform operations comprising:
determining a set of data-source identifiers associated with the one or more external data sources;
determining, based on the set of data-source identifiers, a first uniform resource locator associated with a first external data source of the one or more external data sources,
wherein the first external data source is associated with cybersecurity data, economic data, news data, or social media feed data; and
retrieving, using the first uniform resource locator and an application programming interface (API), a first dataset associated with the first external data source.

### Claim 12 (depends on 11)

The one or more non-transitory, computer-readable media of claim 11, wherein the instructions for generating the first hypothesis including the first constraint cause the computing system to perform operations comprising:
providing the first dataset associated with the first external data source to a constraint generation model to generate the first constraint,
wherein the first constraint includes at least one of: an indication of a cybersecurity threat, an economic indicator, an environmental indicator, a regulatory requirement, or an agent-interaction constraint; and
generating a first data schema, associated with the first hypothesis,
wherein the first data schema includes an indication of the first constraint.

### Claim 13 (depends on 9)

The one or more non-transitory, computer-readable media of claim 9, wherein the instructions for executing the simulation session, consistent with the first constraint, cause the computing system to perform operations comprising:
generating a first data schema including a representation of the first constraint;
providing the first data schema and the first question to the first agent of the set of agents;
in response to providing the first data schema and the first question to the first agent, generating a first output for the first output set.

### Claim 14 (depends on 9)

The one or more non-transitory, computer-readable media of claim 9, wherein the instructions for generating the first hypothesis cause the computing system to perform operations comprising:
providing the set of agent traits and the external data to the hypothesis generation model to generate a set of questions based on the external data;
generating, for display on a GUI associated with a user device, representations of the set of questions;
in response to generating, for display on the GUI, the representations of the set of questions, receiving, via the GUI, a user selection of the first question of the set of questions; and
generating the first hypothesis comprising the first question associated with the user selection.

### Claim 15 (depends on 9)

The one or more non-transitory, computer-readable media of claim 9, wherein the instructions for generating the first hypothesis cause the computing system to perform operations comprising:
generating, for display on a GUI associated with a user device, a representation of a preliminary question;
in response to generating, for display on the GUI, the representation of the preliminary question, receiving, via the GUI, the first question,
wherein the first question is a modified version of the preliminary question; and
generating the first hypothesis including the first question.

### Claim 16 (depends on 9)

The one or more non-transitory, computer-readable media of claim 9, wherein the instructions for generating the first hypothesis comprising the first question cause the computing system to perform operations comprising:
determining, based on the external data, a first value associated with the first constraint;
generating the first question including an indication of the first value associated with the first constraint; and
generating the first hypothesis including the first question.

### Claim 17

A computer system having one or more processors and one or more memory units having computer-executable instructions stored therein generating hypotheses for multiagent simulations in response to dynamically-retrieved external data, the instructions, when executed by the one or more processors, causing the computer system to:
receiving instructions comprising a user input including a seed phrase;
inputting the user input into a configuration generation model to generate a set of data-source identifiers of a set of external data sources;
in response to generating the set of data-source identifiers associated with the set of external data sources, retrieving external data from one or more external data sources of the set of external data sources,
instantiating a virtual world comprising a set of agents;
generating, using the external data, a set of agent traits associated with the virtual world;
instantiating the set of agents including a first agent associated with the set of agent traits,
wherein the set of agent traits is determined based on the external data;
inputting the set of agent traits and the external data into a hypothesis generation model to generate a first hypothesis comprising a first question based on the external data,
wherein the first hypothesis includes a first constraint associated with the external data, and
wherein the first question is associated with the set of agents associated with the set of agent traits;
executing a simulation session in the virtual world, consistent with the first constraint, by causing the set of agents to generate, using the first hypothesis and the first question, a first output set; and
generating a representation of the first output set.

### Claim 18 (depends on 17)

The computer system of claim 17, wherein the instructions for instantiating the set of agents using the set of agent traits cause the computer system to:
generate, using the configuration generation model, the set of agent traits including at least two of: a role indicator, an institution identifier, an age value, a salary value, a residence address identifier, a political affiliation identifier, and an educational attainment identifier;
generate, for a first agent of the set of agents, the set of agent traits;
modify a first agent trait of the set of agent traits to generate a second set of agent traits associated with a second agent of the set of agents; and
instantiate the first agent associated with the set of agent traits and the second agent associated with the second set of agent traits.

### Claim 19 (depends on 17)

The computer system of claim 17, wherein the instructions for retrieving the external data from the one or more external data sources cause the computer system to:
determine a set of data-source identifiers associated with the one or more external data sources;
determine, based on the set of data-source identifiers, a first uniform resource locator associated with a first external data source of the one or more external data sources,
wherein the first external data source is associated with economic data, news data, or social media feed data; and
retrieve, using the first uniform resource locator and an application programming interface (API), a first dataset associated with the first external data source.

### Claim 20 (depends on 19)

The computer system of claim 19, wherein the instructions for generating the first hypothesis including the first constraint cause the computer system to:
provide the first dataset associated with the first external data source to a constraint generation model to generate the first constraint,
wherein the first constraint includes at least one of: an indication of an economic indicator, an environmental indicator, a regulatory requirement, or an agent-interaction constraint; and
generate a first data schema, associated with the first hypothesis,
wherein the first data schema includes an indication of the first constraint.
