Method and system for providing an interactive session for a user with a virtual agent

Abstract

The present disclosure discloses a method and a system for providing an interactive session for a user with a virtual agent. The method comprises receiving at least one user input comprising at least one image related to a design of a product. The method further comprises providing the interactive session with the user by: a) generating at least one query corresponding to the at least one user input by analysing the design of the product, b) receiving at least one user response to the at least one query via the user interface, c) generating one or more subsequent queries corresponding to the at least one user response. The one or more subsequent queries are generated based on past behaviour of the user and one or more properties related to the design of the product. The one or more properties are extracted from the at least one user response and subsequent.

Description

METHOD AND SYSTEM FOR PROVIDING AN INTERACTIVE SESSION FOR A USER WITH A VIRTUAL AGENTTECHNICAL FIELD

[0001] The present invention generally relates to the field of virtual agents, and more particularly relates to a method and a system for providing an interactive session for a user with a virtual agent.BACKGROUND

[0002] The following description includes information that may be useful in understanding the present invention. It is not an admission that any of the information provided herein is prior art or relevant to the presently claimed invention, or that any publication specifically or implicitly referenced is prior art.

[0003] A virtual agent refers to an interactive system that simulates human-like conversations or interactions using artificial intelligence (Al) techniques. The virtual agent is designed to communicate with a user in a natural language format, typically through text-based chat interfaces or voice-based interactions. The virtual agent may be capable of understanding user queries or requests, interpreting the context, and providing appropriate responses or actions.

[0004] The virtual agent's functionality is based on advanced algorithms and Al models, that enables processing and analyzing user input, extracting relevant information, and generating meaningful and contextually appropriate responses. The virtual agent may utilize various techniques such as natural language processing (NLP), machine learning, pattern recognition, and knowledge representation to achieve accurate understanding and effective communication with the user.

[0005] Conventionally, the virtual agents are known in this evolving landscape for providing automatic answers for the user queries. However, traditional methodologies of the virtual agents often involve fragmented communication and static data processing. Thus, the virtual agents may show inefficiencies and inaccuracies in providing response related to the user queries. For example, considering the virtual agent to a chatbot that provides answers to the user queries related to manufacturing and product development. A traditional chatbot may only answer to the queries raised by the user related to the manufacturing and product development. However, sometimes the queries raised by user and the corresponding answers may be insufficient for manufacturing a product, and the like.

[0006] Therefore, there exists a need for a real-time automated and dynamic virtual agent with improved efficiency and accuracy.SUMMARY

[0007] The present disclosure overcomes one or more shortcomings of the prior art and provides additional advantages. Embodiments and aspects of the disclosure described in detail herein are considered a part of the claimed disclosure.

[0008] In one non-limiting embodiment of the present disclosure, a method for providing an interactive session for a user with a virtual agent, is disclosed. The method comprises receiving at least one user input comprising at least one image related to a design of a product. The method further comprises providing the interactive session with the user. Herein, the method comprises generating at least one query corresponding to the at least one user input. The at least one query is generated by analysing the design of the product. The method further comprising receiving at least one user response to the at least one query via the user interface. Lastly, the method further comprises generating one or more subsequent queries corresponding to the at least one user response. The one or more subsequent queries are to be responded by the user. The one or more subsequent queries are generated based on past behaviour of the user and one or more properties related to the design of the product. The past behaviour and the one or more properties are extracted from the at least one user response and subsequent.

[0009] In one non-limiting embodiment of the present disclosure, a system for providing an interactive session for a user with a virtual agent, is disclosed. The system comprises a memory, processor coupled to the memory and a quantum processor coupled to the memory. The processor receives at least one user input comprising at least one image related to a design of a product. The processor further provides the interactive session with the user. The processor herein generates at least one query corresponding to the at least one user input. The at least one query is generated by analysing the design of the product. Moving ahead, the processor receives at least one user response to the at least one query via the user interface. Lastly, the processor generates one or more subsequent queries corresponding to the at least one user response. The one or more subsequent queries are to be responded by the user. The one or more subsequent queries are generated based on past behaviour of the user and one or more properties related to the design of the product. The past behaviour and the one or more properties are extracted from the at least one user response and subsequent.

[0010] The foregoing summary is illustrative only and is not intended to be in any way limiting. In addition to the illustrative aspects, embodiments, and features described above, further aspects, embodiments, and features will become apparent by reference to the drawings and the following detailed description.BRIEF DESCRIPTION OF DRAWINGS

[0011] The features, nature, and advantages of the present disclosure will become more apparent from the detailed description set forth below when taken in conjunction with the drawings in which like reference characters identify correspondingly throughout. Some embodiments of system and / or methods in accordance with embodiments of the present subject matter are now described, by way of example only, and with reference to the accompanying Figs., in which:

[0012] FIG. 1 depicts an exemplary environment for providing an interactive session for a user with a virtual agent, in accordance with embodiments of the present disclosure.

[0013] FIG. 2 depicts an exemplary block diagram illustrating a system for providing an interactive session for a user with a virtual agent, in accordance with embodiments of the present disclosure.

[0014] FIG. 3 illustrates a detailed block diagram for providing an interactive session for a user with a virtual agent, in accordance with embodiments of the present disclosure.

[0015] Fig. 4 represents a flowchart of an exemplary method for providing an interactive session for a user with a virtual agent, in accordance with embodiments of the present disclosure.

[0016] It should be appreciated by those skilled in the art that any block diagrams herein represent conceptual views of illustrative systems embodying the principles of the present subject matter. Similarly, it will be appreciated that any flow charts, flow diagrams, state transition diagrams, pseudo code, and the like represent various processes which may be substantially represented in a computer readable medium and executed by a computer or processor, whether or not such computer or processor is explicitly shown.DETAILED DESCRIPTION

[0017] The foregoing has broadly outlined the features and technical advantages of the present disclosure in order that the detailed description of the disclosure that follows may be better understood. It should be appreciated by those skilled in the art that the conception and specific embodiment disclosed may be readily utilized as a basis for modifying or designing other structures for carrying out the same purposes of the present disclosure.

[0018] The novel features which are believed to be characteristic of the disclosure, both as to its organization and method of operation, together with further objects and advantages will be better understood from the following description when considered in connection with the accompanying figures. It is to be expressly understood, however, that each of the figures is provided for the purpose of illustration and description only and is not intended as a definition of the limits of the present disclosure.

[0019] As described earlier, the traditional virtual agent may only answer the queries raised by the user related to the manufacturing and product development. Further, the answers provided by the virtual agent in response to the queries may be static. However, sometimes the queries raised by user and the corresponding answers may be insufficient for manufacturing a product, and the like. Thus, there is a requirement for a real-time automated and dynamic virtual agent.

[0020] The present disclosure provides a method and a system for providing an interactive session for a user with a virtual agent. In particular, the present disclosure proposes a method for interacting by the virtual agent with the user. In the present disclosure, one or more queries may be generated based on the product design received as input. The generated one or more queries may be provided to the user via the virtual agent. Then, the user may respond to the queries via the virtual agent. Relevant answers may be extracted from the response of the user to collect one or more properties related to the product design. Further, one or more subsequent queries may be generated iteratively, until all properties related to the product design are extracted from the response of the user from each of the respective queries. Thus, the present disclosure provides real-time dynamic queries for extracting all properties related to the product design. Further, in the present disclosure, the one or more queries may be generated based on past behavior of the user and confidence level of each of the properties. Thus, the present disclosure accurately extracts the properties related to the product design.

[0021] FIG. 1 exemplary environment 100 for providing an interactive session for a user with a virtual agent, in accordance with embodiments of the present disclosure. The exemplary environment 100 particularly depicts a user equipment 102 that may incorporate a system 103. The system 103 may provide an interactive session for a user 101 with a virtual agent 104 present in the user equipment 102. The environment 100 is exemplified for a scenario when the system 103 may have received an image of a product design, for which one or more property related to the product design, may be required for manufacturing and development of the product. In an exemplary embodiment, the product may be a hardware product that may be manufactured or developed in an industry. In an example without limitation, the product may refer to but not limited to turbine blade of a jet engine, stator of a motor, bearing of a motor. In an example without limitation, the virtual agent 104 may include but not limited to a chatbot, a voice assistant, a service assistant. In one exemplary embodiment, the system 103 may be incorporated with the user equipment 102. In another exemplary embodiment, the system 103 may reside outside the user equipment 102 as a server. Further, the system 103 may be coupled to one or more database 109.

[0022] In an exemplary embodiment, the virtual agent 104 may be an Artificial Intelligence (Al) powered application, often known as a chatbot. In a non-limiting embodiment, the virtual agent 104 may be software application. In a non-limiting embodiment, the virtual agent 104 may interact with the user 101 via interfaces such as a user interface, a web interface and the like. The virtual agent 104 may use natural language processing to understand and respond to users' questions and requests, automating tasks and providing customer service or support.

[0023] In some implementations, the system 103 may comprise a processor 105, a memory 106, a first Large Language Model (LLM) 107 and a second LLM 108. In some implementations, the system 103 may include other components (not shown in this fig.) to implement desired functions of the system 103. In an exemplary embodiment, the first LLM 107 and the second LLM 108 may be pretrained. In an exemplary embodiment, the first LLM 107 may be trained for interpreting at least one user response and the second LLM 108 may be trained for generating at least one query and one or more subsequent queries. In an exemplary embodiment, the processor 105 in the system 103 may use the first LLM 107 and the second LLM 108 for providing the interactive session to the user 101 with the virtual agent 104, for extracting one or more properties for manufacturing and developing the product.

[0024] In an exemplary embodiment, the system 103 may receiving at least one user input comprising at least one image related to a design of a product. In an exemplary embodiment, the image may be feed to the system 103 in a format of design file such as a Computer-Aided Design (CAD) file, a DWG (Drawing) file, a DXF (Drawing Exchange Format) file, a IGES (Initial Graphics Exchange Specification) file and the like. Upon receiving the design of the product, an interactive session may be generated with the user 101 by the system 103 via the virtual agent 104. In an embodiment, the virtual agent may be a LLM based virtual agent. The interactive session may include generating at least one query corresponding to the at least one user input on a user interface of the virtual agent 104, by the system 103. The at least one query may be generated by analysing the design of the product. Then, at least one user response to the at least one query may be received via the user interface. Upon receiving the at least one user response, the system 103 may generate one or more subsequent queries corresponding to the at least one user response on the user interface. Each of the one or more subsequent queries may be responded by the user 101 on the user interface. The one or more subsequent queries are generated based on past behaviour of the user 101 and one or more properties related to the design of the product. The past behaviour of the user 101 and one or more properties may be extracted from the at least one user response and subsequent user response. The one or more properties may include but not limited to a product function, a dimension of the product, material requirement of the product, weight of the one or more property, importance of the one or more property. Then, the system 103 may extract the one or more properties related to the design of the product from the user response. Further, the system 103 may generate a report comprising summary of the one or more properties related to the design of the product. The report may be used for manufacturing and development of the product. A detailed explanation of the system 103 is provided in the forthcoming paragraphs in conjunction with FIG.s 2, 3-4.

[0025] Fig. 2 depicts an exemplary block diagram illustrating a system 200 (which is system 103 of Fig. 1) for providing the interactive session for the user 101 with the virtual agent 104 (of Fig. 1), in accordance with embodiments of the present disclosure. In an exemplary embodiment, the system 200 may be a computer system. In some implementations, the system 200 may further comprise a processor 201, a memory 202, a Question Answer (QA) option framer 203, an interpreter LLM module 204, a Question Answer (QA) LLL module 205, a properties module 206 and a user behaviour module 207.

[0026] In one implementation, the processor 201 may be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, logic circuitries, and / or any devices that manipulate signals based on operational instructions. Among other capabilities, the processor 201 may be configured to fetch and execute computer-readable instructions and other information stored in the memory 202. In an exemplary embodiment, the processor 201 may be used for execution instruction of a classical computer. In another exemplary embodiment, the processor 201 may be a general processor used for executing the instructions.

[0027] In an exemplary embodiment, the memory 202 may include one or more non-transitory computer-readable storage media, such as RAM, ROM, EEPROM, EPROM, one or more memory devices, flash memory devices, etc., and combinations thereof. In an exemplary embodiment, the input data 202a may be stored within the memory 202 in the form of various data structures. In a non-limiting example, input data 202a refers as the at least one user response, the user past behaviour, the one or more properties and the like. The memory 202 may also store other data 202b such as temporary data and temporary files, generated by the processor 201 or other any other parts of the system 200 including the virtual agent, the interpreter LLM module 204, the QA LLM module 205, the properties module 206 and the user behaviour module 207 for performing the various functions of the present invention.

[0028] In an exemplary embodiment, the QA option framer 203, the interpreter LLM module204, the QA LLM module 205, the properties module 206 and the user behaviour module 207 may be a pretrained for performing particular task use a set of technologies. A person of ordinary skill will appreciate that the QA option framer 203, the interpreter LLM module 204, the QA LLM module 205, the properties module 206 and the user behaviour module 207 may use the input data to mimic human cognitive functions like learning, problem-solving, and reasoning. The QA option framer 203, the interpreter LLM module 204, the QA LLM module205, the properties module 206 and the user behaviour module 207 may be pre-trained to perform particular task. In an exemplary embodiment, the QA option framer 203, the interpreter LLM module 204, the QA LLM module 205, the properties module 206 and the user behaviour module 207 may reside inside the processor 201. In another exemplary embodiment, the virtual agent, the interpreter LLM module 204, the QA LLM module 205, the properties module 206 and the user behaviour module 207 may reside outside the processor 201.

[0029] In an exemplary embodiment, the QA option framer 203 may frame options for the one or more queries. The options may be displayed along with the one or more queries on the user interface of the virtual agent. In an exemplary embodiment, the interpreter LLM module 204 may modify text received from the QA LLM module 205. In an exemplary embodiment, the QA LLM module 205 may generate at least one query and the one or more subsequent queries. In an exemplary embodiment, the properties module 206 may evaluate a confidence level of the one or more properties. In an exemplary embodiment, the user behaviour module 207 may analyse behaviour of the user.

[0030] In an exemplary embodiment, the one or more properties may include, but not limited to of the product function, the dimension of the product, the material requirement of the product and the weight of the one or more property. In a non-limiting embodiment, the one or more properties may refer to a characteristics or qualities that describe and identify a substance or a component. The properties of the product may not be limited to, physical structure of the product, chemical traits, behaviour of the product under various conditions. In a non-limiting embodiment, the product function may refer to task or action performed by the product. In a non-limiting embodiment, the dimension of the product may refer to measure of the product in terms of length, breath, width, height and the like. In a non-limiting embodiment, the material requirement of the product may refer to type of material required for manufacturing the product. In a non-limiting embodiment, the weight of the one or more property may refer to importance of the property with respect to the design of the product.

[0031] In one implementation, a properties storage 208a, a project storage 208b and a user behaviour storage 208c may refer to a database. In an exemplary embodiment, the properties storage 208a, the project storage 208b and the user behaviour storage 208c may store respective data received from the system 200. In an exemplary embodiment, the properties storage 208a may store the one or more properties of the design of the product. In an exemplary embodiment, the properties storage 208a, may provide the one or more properties for further evaluation. The project storage 208b may store the project details. The user behaviour storage 208c may store past behaviour of the user. In an exemplary embodiment, the user behaviour storage 208c may provide data related to past behaviour of the user to the interpreter LLM module 204. The nonlimiting examples of the properties storage 208a, the project storage 208b and the user behaviour storage 208c may be a server, data storage and the like. In an exemplary embodiment, the properties storage 208a, the project storage 208b and the user behaviourstorage 208c may may reside outside the system 200. In another exemplary embodiment, the properties storage 208a, the project storage 208b and the user behaviour storage 208c may reside inside the system 200.

[0032] In an exemplary embodiment, the processor 201 may receive the at least one user input comprising the at least one image related to the design of the product. The at least one image may be related to design of the product. In an exemplary embodiment, the image may be related to either a two-dimensional image or a three-dimensional image. The image may specify the design of the product. In a non-limiting example, the design of the product may refer to structure of a hardware product. In a non-limiting example, the image may be received in a CAD fde format.

[0033] The processor 201 may provide the interactive session for the user with the virtual agent. While providing the interactive session to the user, the processor 201 may generate the at least one query corresponding to the at least one user input on the user interface of the virtual agent. The at least one query may be generated by analysing the design of the product. The at least one query may be provided to the user on the user interface of the virtual agent. In one exemplary embodiment, the at least one query generated initially, when the image is received, may be based on the design of the product and in further process the at least one query may be generated based on the at least one user response provided to the previous at least one query. In an exemplary embodiment, the QA LLM module 205 associated with the processor 201 may generate the at least query and the interpreter LLM module 204 may format the at least one query in proceeded format and provide to the formatted query to the user. In a non-limiting embodiment, the processed format may refer to standard natural language that is readable by the user. In an exemplary embodiment, the QA option framer 203 may formulate at least one option with respect to the at least one query to be sent to the user on the user interface. The Upon providing the at least query to the user, the user may respond to the query and the processor 201 may receive the at least one user response to the at least one query via the user interface.

[0034] Upon receiving the user response, the processor 201 may generate one or more subsequent queries corresponding to the at least one user response where each of the one or more subsequent queries may be responded by the user. In an exemplary embodiment, the one or more subsequent queries are generated based on past behaviour of the user and one or moreproperties related to the design of the product. The past behaviour of the user and the one or more properties may be extracted from the at least one user response and subsequent user response. Thus, the present disclosure iteratively generates multiple question for extracting all the properties required for manufacturing the product. The question generated are particularly related to the design of the product. Thereby the present disclosure provides real-time automated and dynamic virtual agent with improved efficiency and accuracy.

[0035] Referring to Fig. 3, initially, upon receiving the design of the product, a Question Answer (QA) LLM 304 may generate the at least one query based on the received design of the product. Then, an interpreter LLM 303 may modify the at least one query received from the QA LLM 304. In an exemplary embodiment, the interpreter LLM 303 may modify the at least one query into processed text format and provide the processed text format to a Question Answer (QA) option framer 302. The QA option framer 302 may receive the processed text format of the at least one query from the Interpreter LLM 303 and may decide whether any selectable options need to be presented along with the at least one quey based on the processed text format. The output of the QA OF 302 may be provided as the at least quey to a user 301. In a non -limiting example, the QA option framer 302 may not provide options along with the question. In an exemplary embodiment, providing the options may be based on the at least on query and the user behaviour with respect to the at least one quey and a confidence level of the at least one query.

[0036] Then, the user 301 may response to the at least one query by providing the user response via the user interface of the virtual agent. The user response may be forwarded to the QA LLM 304 for generating further one or more subsequent queries based on the user response. In an exemplary embodiment, upon receiving the user response, the QA LLM 304 may generate further at least one query based on the user response. Then, iteratively the QA LLM 304 may generate the one or more subsequent queries based on the received user response, until all the one or more properties are received with respect to the design of the product. In an exemplary embodiment, the one or more subsequent queries may also be generated based on the past behaviour of the user and the extracted one or more properties. Thus, the present disclosure extracts the one or more properties based on the at least one user response provided by the user initially and followed by the subsequent user responses. The extracted one or more properties may be stored in properties storage 305.

[0037] Referring back to Fig. 2, Upon generating the at least one query and the one or more subsequent queries corresponding to the at least one user response, the processor 201 may identify the one or more properties related to the design of the product from the at least one user response. In an exemplary embodiment, the one or more properties may include but are not limited to the product function, the dimension of the product, the material requirement of the product, the weight of the one or more property, the importance of the one or more property . Then, the processor 201 may determining a confidence level of the one or more properties based on a weight of each of the property among the one or more properties. The weight may be indicative of an importance of the property. Further, the confidence level may be compared with a predefined threshold value for determining whether the at least one property of the one or more properties is missing from an internal database. In case the confidence level is less than the predefined threshold value then the at least one property may be missing. In such case, the processor 201 may continue generating the one or more subsequent queries for the user till the confidence level of the missing at least one property crosses the predefined threshold value. In other case, if the processor 201 determines that the confidence level of the at least one property is more than the predefined threshold value, then the one or more properties provide required information related to the design of the product. Further, the one or more properties may be stored in the internal database. In an exemplary embodiment, the internal database may refer to the properties storage 208a. In other exemplary embodiment, the properties module 206 may identify the confidence level of the one or more properties and extract all the required information related to the design of the product by receiving all the required one or more properties from the user response. Thereby, the present disclosure extracts all the required information related to design of the product by generating one or more subsequent queries. Further, as the present disclosure generates the one or more subsequent queries based on the past behaviour of the user and the one or more properties, accurate answers may be extracted, and thereby identifying more accurate properties related to the design of the product. Hence, the efficiency of the system may be increased.

[0038] Referring back to Fig. 3, the one or more properties related to the design of the product may be identified from the at least one response. In an embodiment, the properties storage 305 may be a database that stores the one or more properties identified from the user responses. Further, the one or more properties may be evaluated for determining the confidence level of the one or more properties. The properties storage 305 may further provide the one or more properties to a properties adder function 314 for further evaluation. The properties storage 305may receive the input in terms of property value in text format from QA LLM 304, and output from the properties storage 305 may be in terms of property importance value in numerical format to properties adder function 314. In an embodiment, the properties adder function 314 may increase a confidence metric of the virtual agent by adding a numerical value to a variable. The value to be added may be decided based on a weight / importance of the one or more properties. In a non-limiting embodiment, the properties adder function 314 may receive the input in terms of property importance in numerical format from the properties storage 305 and provide an output in terms of adjusted confidence metric based on the property importance.

[0039] In an embodiment, the properties confidence 306 determine the confidence level of the one or more properties based on the weight of the property of the one or more properties. The weight may be indicative of the importance of each of the property from the one or more properties. Then, the confidence level may be compared with the predefined threshold value. In an exemplary embodiment, the properties confidence 306 may decide whether the confidence metric (confidence level) is high, based on a threshold for the varying confidence level determined. In a non-limiting embodiment, the input for the properties confidence 306 may be the current confidence level and industry threshold and the output may be the decision on whether the confidence is high or not.

[0040] In case of the confidence level is higher than the predefined threshold value, the properties refiner 307 may modify the one or more properties stored in the properties storage 305 into a suitable form for providing as the output from the virtual agent and further provide the output to an output format checker 308. The output format checker 308 may verify whether the one or more properties are in a preferred format to provide as output from the virtual agent. Then, the formatted one or more properties may be sent to a project storage 315 for storing all the one or more properties related to the particular project. In an embodiment, the project storage 315 virtual agent may be a database used to store the one or more properties of all the parts that are included in the particular project.

[0041] In case the confidence level is lower than the predefined threshold value a lacking property checker 316 may be triggered. In case of absence of the at least on property related to the particular project may provide low confidence level. In such case, the lacking property checker 316 may check the properties storage 305 to find which important property may be missing and sends a notification to the QA LLM 304, indicating that a property among one ormore properties required for a particular project is missing. In a non-limiting embodiment, the lacking property checker 316 may receive the input from the properties storage 305 when properties confidence is low, and the output from the lacking property checker 316 the notification of missing property name with a highest priority to the QA LLM 304. Then, the QA LLM 304 may generate the one or more subsequent queries to receive user inputs from the user related to the missing property. The QA LLM 304 may generate the one or more subsequent queries until all the one or more properties required for a particular project are extracted from the user response.

[0042] In an embodiment, a property matcher 317 may compare the one or more properties related to the particular project, present in the project storage 315 with the identified at least one property to be fill in the missing properties. The property matcher 317 may compare to obtain changes confirmed by the user and the property matcher 317 may sends the changes for user confirmation to the interpreter LLM 303. Thus, the present disclosure accurately identifies each and every property required for manufacturing and developing the product proposed in the particular project.

[0043] Referring back to Fig. 2, the processor 201 may generate the one or more subsequent queries corresponding to the at least one user response based on past behaviour. The processor 201 may analyse the past behaviour of the user based on user behaviour data stored in a database. In an exemplary embodiment, the user behaviour module 207 may analyse the past behaviour of the user. In an exemplary embodiment, the past behaviour comprises at least one response of the user to at least one previous query. The previous query may be part of the one or more subsequent queries generated by the processor 201 with respect to same design of the product. The user behaviour may be stores in the user behaviour storage 208c. Thus, as the present disclosure considers user behaviour for generating the queries, the system provides a user-friendly virtual agent with improved efficiency.

[0044] Referring back to Fig. 3, the QA LLM 304 and Interpreter LLM 303 may generate the one or more subsequent queries by analysing the past behaviour of the user. In an embodiment, a user behaviour analysis 309 may take the responses of the user as input to analyse the behaviour and store it in a correct format via the user storage format checker 310. For example, input may be response of the user in the text format and the output may be user behavior data in the correct format to the user storage format checker 310.

[0045] In an exemplary embodiment, the user storage format checker 310 may check whether the output from the user behaviour analysis 309 matching the required format to store it in a user long term storage 311.

[0046] In an exemplary embodiment, the user behaviour formatter 312 may modify the user behaviour data from the user long term storage 311 to be used by the Interpreter LLM 303 after the format is checked by the user behaviour format checker 313. In an exemplary embodiment, the user behaviour format checker 313 may ensure that the input to the interpreter LLM 303 is in the correct standardized format. The correct format may be the standardized data format in text format for the interpreter LLM 303.

[0047] In an exemplary embodiment, the virtual agent may initiate the process when the user, for example such as a product engineer, interacts with the virtual agent to create a specific component / product. The interaction starts with the user responding to initial questions from the QA LLM 304, which is designed to gather information about the component the user intends to manufacture.

[0048] In an embodiment, the QA LLM 304 may create questions for the user based on the properties that are present in the properties storage 305 in the form of key-value pairs. The question to ask may be based on the importance value of the property which ranges from 1 to 5 and varies based on the industry identified by the initial questions. The importance value may be stored in the properties storage 305 from the beginning of the interaction. As the user provides details about the component, such as the products function, dimensions, material requirements, etc., the QA LLM 304 captures the information and stores the information in the properties storage 305.

[0049] The Interpreter LLM 303 may utilize insights from the user behaviour analysis 309 and refined by the user behaviour formatter 312, tailors the interaction to the communication preferences and style of user by making the process more efficient and user-friendly. This customization ensures that the questions and options presented are aligned with the user's past interactions and preferences.

[0050] In an exemplary embodiment, the adjustments / refinement may include but are not limited to:Technical Expertise Adjustment:For a highly technical user: The Interpreter LLM 303 might use more jargon and complex technical terms. For instance, in case the user has consistently provided detailed technical responses, a question might be phrased as, "What are the specific tensile strength requirements for the material in MPa?"For a less technical user: The virtual agent may simplify the language and provide more context or explanations, such as "Could you tell us how strong the material needs to be? Strength means how much the product may stretch or pull without breaking."Question Formatting:For the users who prefer structured options: The virtual agent may present questions with multiple-choice answers, such as "For the finish of the component, do you prefer: a) Matte, b) Glossy, c) Textured?"For the users who prefer open-ended discussions: The virtual agent may pose more exploratory questions, like "What type of finish are you envisioning for the component, and what considerations are influencing this choice?" Previous behaviour of the user:In case the user has shown a preference for certain materials or processes in the past, the Interpreter LLM 303 might refer these preferences in its questions, like "Based on your previous projects, you've preferred aluminium for its lightweight properties. Would aluminium be suitable for this component as well?"

[0051] Throughout the process, the properties adder function 314 may increment the virtual agent confidence level based on the importance value of the extracted one or more properties. The Properties Confidence 306 may evaluate whether the collected information is sufficient or if additional details are needed. The Properties Confidence 306 may compare the total confidence level value with the predefined threshold value. In varying industries, the threshold value for the virtual agent’s confidence metric may set according to the specific requirements and risks associated with the components being manufactured. For example, aerospace components and medical devices demand highest thresholds due to their critical safety standards, precision needs, and potential risk to life, requiring exhaustive property validation. Automotive parts, while slightly less stringent, still maintain a high threshold to meet safety, durability, and compliance standards. While consumer electronics have a lower threshold, focusing more on functionality, aesthetics, and cost-effectiveness, though reliability and the user safety remain essential considerations. The adaptive threshold ensures the manufacturing process meets the unique demands of each industry while maintaining efficiency and quality.

[0052] In case critical information is missing i.e., the confidence level is low, the lacking property checker 316 may identify gaps between the properties related to the particular project, prompting the user for the specific missing details. Thus, ensuring that all necessary properties of the component are accurately captured.

[0053] The process repeats until the confidence level crosses the predefined threshold value. The properties refiner 307 may then process the collected data, ensuring it is in the correct format and refined for standardization and specific information. The refining process includes but isn’t limited to:Standardization of Units: For instance, if the collected data includes dimensions in a mix of imperial and metric units, the refiner will standardize these to a single unit system, such as converting all measurements to metric units for consistency.Geometric Tolerances: For a component where flatness is critical, the original data might just say "flat surface required". The refiner would specify this as "Flatness tolerance: 0.01 mm over 100 mm".Environmental and Operational Conditions: For a component that needs to operate in high temperatures with a generic note like "heat resistant", the refiner would detail this as "Operational temperature range: -20°C to 600°C".Electrical Conductivity Requirements: For a component described as "conducting", the refiner might specify "Electrical conductivity: Minimum 30 MS / m", providing clear criteria for material selection and testing.

[0054] The above-mentioned step may be crucial for ensuring the component may be manufactured to meet the required specifications.

[0055] In an exemplary embodiment, to fill the gaps in the properties, the property matcher 317 may identify similar previously manufactured parts / requested services and compares their properties with the current one or more properties. The property matcher 317 identifies may recommend any necessary adjustments to fill the missing properties of the current part with the previous part’s properties, with the suggestions communicated to the user for confirmation through the Interpreter LLM 303. For example, the Property Matcher 317 may identify a missing mechanical property, such as tensile strength, in the specifications of a new component. The property was specified in previous similar components within the same project but is absent in the current part's specifications. Utilizing the previous information, theInterpreter LLM 303 communicates with the user, asking, "Your previous components in this project specified a tensile strength of 500 MPa. The property is missing in the current component specifications. Would you like to include this tensile strength requirement for the new component?". This interaction ensures that critical mechanical properties are consistently applied across similar components in the project, maintaining quality and performance standards.

[0056] Upon all the necessary one or more properties are refined and confirmed, the output format checker 308 ensures that the data is in the correct format to export and save the one or more properties.

[0057] The Project Storage 315 may be updated with the one or more properties of the new component, enriching the virtual agent knowledge base for future components. The Project Storage 315 serves as a valuable resource for comparing and contrasting with future components that are required by the user.

[0058] Throughout the process, the system 300 may refines its understanding of the user's behaviour and preferences through continuous interaction, updating the user long term storage 311 with new insights. This learning process enhances future interactions, making the system 300 more responsive and tailored to the requirements of the user. For example, in case the user consistently focuses on the mechanical properties of materials, such as fatigue strength, yield strength, and Young's modulus, during the design of high-stress components, the system 300 takes a note of the same. The user behaviour Analysis 309 identifies the pattern, emphasizing the user's recurrent interest in these specific properties. This information is processed and encapsulated into the user preferences like "High Focus on Mechanical Strength Properties" and stored in the user long term Storage 311.

[0059] In a non-limiting example, the system 300 may initiates a conversation via a virtual agent with an aerospace engineer, Rajesh, who has uploaded a CAD file of a turbine blade for a jet engine but may have not provided any additional information. The chatbot’s objective is to extract all the necessary properties for this component through an interactive Q&A session.Virtual Agent: "Welcome, Rajesh. I see you've uploaded a CAD file for a component. Can you confirm if this part is intended for the aerospace industry?"Rajesh: "Yes, it's a turbine blade for a jet engine."Virtual Agent: "Great! Let's start detailing your turbine blade. Given its application, high- temperature resistance is crucial. Will this component operate in extreme temperature conditions?"Rajesh: "Absolutely, it needs to withstand temperatures from -50°C up to 1500°C."Virtual Agent: "Understood. Considering the operational stress, is the blade exposed to dynamic stress due to high-speed rotation?"Rajesh: "Yes, it will be subjected to significant dynamic stresses."Virtual Agent: "Noted. Aerospace components often face harsh environments. Will this blade be exposed to chemical substances, like fuel or combustion by-products?"Rajesh: "Indeed, it needs to be resistant to those kinds of chemical exposures."Virtual Agent: "Precision is key in aerospace. What level of dimensional accuracy are you aiming for?"Rajesh: "We need a very high precision, so a dimensional accuracy of ±0.01 mm is necessary." Virtual Agent: "Do you have a specific material in mind, or would you like recommendations based on the blade's conditions?"Rajesh: "I'm open to recommendations, but the material must meet our thermal and strength criteria."Virtual Agent: "Understood. In terms of production, are we looking at a low-volume batch for testing purposes, or is this for mass production?"Rajesh: "This is for a prototype batch, let's say around 5 units to start with."Virtual Agent: "Perfect. With prototypes, time is often of the essence. What's your expected lead time for this batch?"Rajesh: "We're aiming for a 3-week turnaround."Virtual Agent: "Alright, that gives us a good framework. Moving on to post-processing, are there any specific surface treatments or finishes required to enhance the blade's performance or longevity?"Rajesh: "Yes, a surface treatment for thermal resistance and a smooth finish are necessary for aerodynamics."Virtual Agent: "Lastly, compliance with aerospace regulations is critical. Will this component need to meet any specific regulatory standards?"Rajesh: "Absolutely, it must comply with all relevant aerospace standards and regulations." Virtual Agent: "Thank you, Rajesh. We've gathered all the essential details for your turbine blade. Is there anything else you'd like to add or any other requirements we should consider?" Rajesh: "That covers everything. I look forward to seeing the review."

[0060] Using the information from above conversation the final summary report may be generated as output is represented in specific format as given below.” 'json{"materialType": "Metallic","productstage": "Functional Testing","industry": "Aerospace","thermalStrength": "Yes","temperatureRange": "-50°C to 1500°C","strengthMPa": 1000,"stressExposure": "Dynamic","chemicalExposure": "Yes","accuracyResolution": "±0.01 mm","partDimensions": {"Length": "200 mm","Width": "150 mm","Thickness": "30 mm"},"productionVolume": "Low (Prototype batch of 50 units)","leadTime": "3 months","budget": "$100,000 for prototype batch","waterProof : "No","flexibilityRequirement": "No","colorRequirement" : "Default","transparencyRequirement" : "No","temperatureRequirement": "-50°C to 1500°C","strengthRequirement": 1000,"partType": "Fitting","staticStress": "No","material": "High-Temperature Alloy","color": "Default","dimensionalAccuracy": "±0.01 mm","surfaceFinish": "High", "additionalProperties": { "mechanicalProperties": "High", "thermalProperties": "High", "chemicalResistance": "Yes", "electrical Conductivity": "No", "regulatoryCompliance": "Yes", "postProcessing": "Yes", "biocompatibility" : "No", "environmentallmpact": "Low", "lightweightstructure": "Yes" } }

[0061] In an embodiment, the present disclosure may find the uses in the domain of manufacturing. For example, the virtual agent may be used to extract specific details from the user’s natural responses. Due to the nature of the system, in one of the exemplary embodiments, the virtual agent may be integrated with any product lifecycle management (PLM) system as well as digital manufacturing solutions. Extracting key requirements from the users serves as the key to consulting in the manufacturing domain. Conventionally this has been done manually with human experts. The present disclosure seeks to automate this expensive and time-consuming process, thus serving as an automation solution.

[0062] The output report produced by the system serves as the base on which a manufacturing consultation and recommendation system may work. The structured data allows for easily comparing the user requirements and matching them with manufacturing methods. The integration of the current software system with similar software / system makes the whole process end-to-end automated.

[0063] FIG. 4 represents flowchart of an exemplary method system for providing an interactive session for a user with a virtual agent, in accordance with embodiments of the present disclosure. The order in which the method 400 is described is not intended to be construed as a limitation, and any number of the described method blocks may be combined in any order to implement the method. Additionally, individual blocks may be deleted from themethods without departing from the spirit and scope of the subject matter described. Furthermore, the method can be implemented in any suitable hardware, software, firmware, or combination thereof. However, for ease of explanation, in the embodiments described below, the method 400 may be implemented by the respective components and / or by the processor 201 using the virtual agent, the interpreter LLM module 204, the QA LLM module 205, the properties module 206 and the user behaviour module 207 of FIG. 2.

[0064] At step 401, the method may include receiving the at least one user input comprising the at least one image related to the design of the product. In one implementation, the processor 201 may receive the at least one image related to the design of the product.

[0065] At step 402, the method may include generating the at least one query corresponding to the at least one user input, on the user interface of the virtual agent. The at least one query may be generated by analysing the design of the product. In one implementation, the processor 201 may generate the at least one query. In another implementation, the combination of the QA option framer 203, the interpreter LLM module 204 and the QA LLM module 205 may generate the at least one query.

[0066] At step 403, the method may include receiving the at least one user response to the at least one query via the user interface. In one implementation, the processor 201 may receive the at least one user response.

[0067] At step 404, the method may include generating the one or more subsequent queries corresponding to the at least one user response on the user interface. Each of the one or more subsequent queries may be responded by the user and the one or more subsequent queries may be generated based on the past behaviour of the user and the one or more properties related to the design of the product. The one or more properties may be extracted from the at least one user response and subsequent. In one implementation, the processor 201 may generating the one or more subsequent queries. In other implementation, the combination of the QA option framer 203, the interpreter LLM module 204 and the QA LLM module 205 may generate the one or more subsequent queries. In one implementation, the past behaviour of the user may be analysed by the processor 201. In other implementation, the past behaviour of the user may be analysed by the user behaviour module 207. In one implementation, the one or more properties may be identified by the processor 201. In other implementation, the one or more properties may be identified by the properties module 206.

[0068] The order in which the method 400 is described is not intended to be construed as a limitation, and any number of the described method blocks may be combined in any order to implement the method. Additionally, individual blocks may be deleted from the methods without departing from the spirit and scope of the subject matter described.

[0069] The illustrated steps are set out to explain the exemplary embodiments shown, and it should be anticipated that ongoing technological development will change the manner in which particular functions are performed. These examples are presented herein for purposes of illustration, and not limitation. Further, the boundaries of the functional building blocks have been arbitrarily defined herein for the convenience of the description. Alternative boundaries can be defined so long as the specified functions and relationships thereof are appropriately performed.

[0070] Advantageous• The present disclosure provides the adaptive virtual agent interaction model which significantly enhances user engagement by tailoring questions and options to the user's preferences and past behaviour by making the process more intuitive and less timeconsuming. Thereby, increases accuracy and efficiency of the virtual agent.• The virtual agent’s dynamic confidence metric ensures that all necessary information is captured before progressing. Thereby reducing the risk of errors or omissions that could lead to costly manufacturing defects.• An embodiment of present disclosure provides an ability to learn from and leverage past project data for property matching which introduces a level of efficiency and accuracy in capturing component specifications that is unprecedented.• The continuous refinement of user behaviour insights makes the system increasingly efficient over time. This helps in providing a personalized and user-centric experience that is unmatched by traditional static virtual agent.

[0071] Alternatives (including equivalents, extensions, variations, deviations, etc., of those described herein) will be apparent to persons skilled in the relevant art(s) based on the teachings contained herein. Such alternatives fall within the scope and spirit of the disclosed embodiments. Furthermore, one or more computer-readable storage media may be utilized in implementing embodiments consistent with the present disclosure. A computer-readable storage medium refers to any type of physical memory on which information or data readableby a processor may be stored. Thus, a computer-readable storage medium may store instructions for execution by one or more processors, including instructions for causing the processor(s) to perform steps or stages consistent with the embodiments described herein. The term “computer- readable medium” should be understood to include tangible items and exclude carrier waves and transient signals, i.e., are non-transitory. Examples include random access memory (RAM), read-only memory (ROM), volatile memory, non-volatile memory, hard drives, CD ROMs, DVDs, flash drives, disks, and any other known physical storage media.

Claims

We claim:

1. A method for providing an interactive session for a user with a virtual agent, the method comprising: receiving at least one user input comprising at least one image related to a design of a product; and providing the interactive session with the user by: generating, on a user interface of the virtual agent, at least one query corresponding to the at least one user input, wherein the at least one query is generated by analysing the design of the product; receiving at least one user response to the at least one query via the user interface; and generating, on the user interface, one or more subsequent queries corresponding to the at least one user response, wherein each of the one or more subsequent queries are to be responded by the user and wherein the one or more subsequent queries are generated based on past behaviour of the user and one or more properties related to the design of the product, extracted from the at least one user response and subsequent;2. The method as claimed in claim 1, wherein the virtual agent is a Large Language Model (LLM) based virtual agent.

3. The method as claimed in claim 1, wherein generating the one or more subsequent queries corresponding to the at least one user response, comprising: identifying the one or more properties related to the design of the product from the at least one user response; determining a confidence level of the one or more properties based on a weight of a property of the one or more properties, wherein the weight is indicative of an importance of the property; comparing the confidence level with a predefined threshold value; determining that at least one property of the one or more properties is missing from an internal database if the confidence level is less than the predefined threshold value; and continue generating the one or more subsequent queries for the user till the confidence level of the missing at least one property crosses the predefined threshold value.

4. The method as claimed in claim 3, wherein comparing the confidence level with the predefined threshold value, further comprising: determining that the one or more properties present in the internal database, provides required information on the design of the product, if the confidence level is more than the predefined threshold value; and storing the one or more properties in the internal database.

5. The method as claimed in claim 1, wherein generating the one or more subsequent queries corresponding to the at least one user response, comprising: analysing past behaviour of the user based on user behaviour data stored in a user behaviour database, wherein the past behaviour comprises at least one response of the user to at least one previous query.

6. The method as claimed in claim 4, further comprising: generating a report based on the one or more properties that provides required information on the design of the product.

7. The method as claimed in claim 1, wherein the one or more properties comprises at least one of a product function, a dimension of the product, material requirement of the product and weight of the one or more property.

8. A system for providing an interactive session for a user with a virtual agent, the system comprises: a memory; at least one processor coupled with the memory configured to: receive at least one user input comprising at least one image related to a design of a product; and provide the interactive session with the user by: generating, on a user interface of the virtual agent, at least one query corresponding to the at least one user input, wherein the at least one query is generated by analysing the design of the product; receiving at least one user response to the at least one query via the user interface; and generating, on the user interface, one or more subsequent queries corresponding to the at least one user response, wherein each of the one or more subsequent queriesare to be responded by the user and wherein the one or more subsequent queries are generated based on past behaviour of the user and one or more properties related to the design of the product, extracted from the at least one user response and subsequent;9. The system as claimed in claim 8, wherein the virtual agent is a Large Language Model (LLM) based virtual agent.

10. The system as claimed in claim 8, wherein to generate the one or more subsequent queries corresponding to the at least one user response, the at least one processor is configured to: identify the one or more properties related to the design of the product from the at least one user response; determine a confidence level of the one or more properties based on a weight of a property of the one or more properties, wherein the weight is indicative of an importance of the property; compare the confidence level with a predefined threshold value; determine that at least one property of the one or more properties is missing from an internal database if the confidence level is less than the predefined threshold value; and continue to generate the one or more subsequent queries for the user till the confidence level of the missing at least one property crosses the predefined threshold value.

11. The system as claimed in claim 10, wherein to compare the confidence level with the predefined threshold value, the at least one processor is configured to: determine that the one or more properties present in the internal database, provides required information on the design of the product, if the confidence level is more than the predefined threshold value; and store the one or more properties in the internal database.

12. The system as claimed in claim 8, wherein to generate the one or more subsequent queries corresponding to the at least one user response, the at least one processor is configured to: analyse past behaviour of the user based on user behaviour data stored in a user behaviour database, wherein the past behaviour comprises at least one response of the user to at least one previous query.

13. The system as claimed in claim 11, further the at least one processor is configured to:generate a report based on the one or more properties that provides required information on the design of the product.

14. The system as claimed in claim 8, wherein the one or more properties comprises at least one of a product function, a dimension of the product, material requirement of the product and weight of the one or more property.

Images