Back to results
US 2026/0277899 A1Published application
Systems and methods for federated task performance controlled by agents
AssigneeInstabase, Inc.
InventorsShaunak Godbole, Shih Ping Chang, Clemens Mewald, Anant Bhardwaj
Life of this patent
Estimating the term…
Abstract
Systems and methods for providing federated task performance controlled by agents are disclosed. Exemplary implementations may: receive user input representing a user prompt that entails a set of operations; generate a plan for executing the set of operations, including a first and second operation; provide a first prompt to a first client agent to execute the first operation; provide a second prompt to a second client agent to execute the second operation; generate and present a reply to the user prompt based on replies from the first and second client agents, and/or perform other steps.
Claims20
A system configured to provide federated task performance controlled by agents, the system comprising:
- one or more hardware processors configured by machine readable instructions to:
- receive, from a user, user input representing a user-generated prompt, wherein the user-generated prompt entails a set of operations, wherein the user is identified by identity information;
- generate a system-generated prompt for one or more machine learning models to create a plan for executing the set of operations;
- provide the system-generated prompt to the one or more machine learning models;
- obtain a particular reply from the one or more machine learning models, wherein the particular reply represents the plan for executing the set of operations, wherein the plan includes a first operation to be executed by a first client agent and a second operation to be executed by a second client agent, wherein execution of the first operation requires access to a first set of electronic files;
- generate a first plan-specific prompt for the first client agent to execute the first operation;
- provide the first plan-specific prompt to the first client agent to execute the first operation, wherein the first plan-specific prompt to the first client agent is accompanied by the identity information, and wherein the first client agent uses role-based access control (RBAC) to verify the access to the first set of electronic files based on the identity information;
- obtain a first plan-specific reply from the first client agent in response to the first plan-specific prompt;
- generate a second plan-specific prompt for the second client agent to execute the second operation;
- provide the second plan-specific prompt to the second client agent to execute the second operation;
- obtain a second plan-specific reply from the second client agent in response to the second plan-specific prompt;
- generate a reply to the user-generated prompt, wherein generation of the reply is based on the first plan-specific reply and the second plan-specific reply, and wherein the reply is based on execution of the plan;
- and
- effectuate a presentation of the reply to the user, through a user interface.
11 more elements
A method of providing federated task performance controlled by agents, the method being implemented in a computer system that includes one or more hardware processors, the method comprising:
- receiving, from a user, user input representing a user-generated prompt, wherein the user-generated prompt entails a set of operations, wherein the user is identified by identity information;
- generating a system-generated prompt for one or more machine learning models to create a plan for executing the set of operations;
- providing the system-generated prompt to the one or more machine learning models;
- obtaining a particular reply from the one or more machine learning models, wherein the particular reply represents the plan for executing the set of operations, wherein the plan includes a first operation to be executed by a first client agent and a second operation to be executed by a second client agent, wherein execution of the first operation requires access to a first set of electronic files;
- generating a first plan-specific prompt for the first client agent to execute the first operation;
- providing the first plan-specific prompt to the first client agent to execute the first operation, wherein the first plan-specific prompt to the first client agent is accompanied by the identity information, and wherein the first client agent uses role-based access control (RBAC) to verify the access to the first set of electronic files based on the identity information;
- obtaining a first plan-specific reply from the first client agent in response to the first plan-specific prompt;
- generating a second plan-specific prompt for the second client agent to execute the second operation;
- providing the second plan-specific prompt to the second client agent to execute the second operation;
- obtaining a second plan-specific reply from the second client agent in response to the second plan-specific prompt;
- generating a reply to the user-generated prompt, wherein generation of the reply is based on the first plan-specific reply and the second plan-specific reply, and wherein the reply is based on execution of the plan;
- and
- effectuating a presentation of the reply to the user, through a user interface.
10 more elements
Dependencies are inferred from the claim text.