US 2026/0278505 A1Published application

Generative artificial intelligence based workflow activites

OwnerRishabh PURWAR, Khyathi VAGOLU +3inventors

InventorsRishabh PURWAR, Khyathi VAGOLU, Himanshu GUPTA, Gaurav BHARGAVA, Ayush GUPTA

Life of this patent

Estimating the term…

Loading patent…

Abstract

In various examples, a workflow assistant tool utilizes a machine learning model to automate the creation and modification of workflow activities for an application based in a natural language input. For example, the workflow assistant tool leverages natural language processing (NLP) and machine learning models to interpret user queries, generate discrete segments and conditions, identify relevant data schemas, and produce executable data objects in a domain-specific language. In another example, the workflow assistant tool includes a multistep Retrieval-Augmented Generation-based prompt engineering pipeline in addition to schema filtering, segment and condition generation, and validation and correction mechanisms to generate workflow activities based on natural language inputs from a user.

Claims20

Dependent claims

An artificial intelligence (AI)-based system, comprising:

a workflow assistant tool having:
a machine learning model;
and
an interface element that obtains a user input through a user interface of an application that provides access to workflow activities, wherein a workflow activity is generated by the machine learning model based on the user input, the user input including at least a natural language description of the workflow activity to be generated by the workflow assistant tool, the workflow activity generated by the machine learning model by at least:
causing the machine learning model, based on the user input, to generate a set of segments, where a first segment of the set of segments represents a task indicated in the user input and includes a set of conditions, where a first condition of the set of conditions represents a subtask indicated in the user input;
determining a set of relevant schemas based on a comparison of table semantic vector embeddings and user input semantic vector embeddings, the user input semantic vector embeddings generated by the machine learning model based on the user input;
causing the machine learning model to generate a representation of the set of segments;
and
generating a structured data object executable by the application in a domain-specific language associated with the application based on the representation of the set of segments;
and
a relevant schema data store that stores the table semantic vector embeddings generated based on annotated tables stored in a database, where the database includes data processed by the application during execution of the workflow activity.

8 more elements

Dependencies are inferred from the claim text.