Customized AI Chatbots Using Structured Data Mediation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AI/ML models, such as large language models, are limited by their need for extensive training data and computing resources, and their responses are often generalized, approximate, and lack precision, leading to inaccuracies and inconsistencies.
Innovation Solution
A computer system enables administrators to customize chatbots using AI/ML models without re-training, by integrating non-AI/ML data processing systems to generate reliable and accurate responses, leveraging database management systems to provide precise data processing, and combining these systems to enhance privacy and reduce data exposure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If AI/ML models are used to generate chatbot responses, then the chatbot can provide natural language responses, but the responses are generalized and approximate, leading to inaccuracies and inconsistencies
Solution Approach 1:
The patent introduces an intermediary data processing layer between the user prompt and the AI/ML model. This layer retrieves precise data from structured data sources (databases, data warehouses) and prepares it for the AI/ML model, which then generates the final response. The intermediary ensures that the AI/ML model works with accurate, relevant data rather than generating entirely approximate responses from scratch.
Solution Approach 2:
The system combines multiple components into a composite chatbot architecture: structured data sources (for precision), AI/ML models (for natural language generation), and data processing systems (for integration). This composite approach leverages the strengths of each component while mitigating their individual weaknesses, producing responses that are both accurate and naturally expressed.
2Reliability
If extensive training data and computing power are used to train capable AI/ML models, then the models become highly trained and capable, but the training requires very large amounts of computing resources and time
Solution Approach 1:
The patent performs data preparation and retrieval operations in advance, before the AI/ML model needs to generate responses. Structured data is pre-processed, validated, and organized in accessible formats, reducing the computational burden during actual inference. This preliminary action ensures that when the model operates, it receives ready-to-use data rather than needing to process raw information in real-time.
Solution Approach 2:
The system extracts only the necessary and relevant data from structured sources that is needed for specific chatbot responses, rather than using the AI/ML model to process all available training data during operation. This extraction approach reduces computational requirements by focusing the model's processing power on generating responses from pre-selected, relevant information.
3Ease of operation
If AI/ML models are used to process data, then natural language responses can be generated, but data exposure and privacy risks increase
Solution Approach 1:
The patent introduces an intermediary data processing layer that acts as a gatekeeper between sensitive structured data and the AI/ML model. This intermediary retrieves, validates, and prepares data before presenting it to the model, ensuring that only necessary and authorized data is exposed. The intermediary can also mask or anonymize sensitive information while preserving the utility needed for response generation.
4Adaptability or versatility
If customized chatbots are created for specific uses or contexts, then the chatbots can operate as intended, but re-training AI/ML models for customization requires significant resources
Solution Approach 1:
The patent enables customization through preliminary configuration of data sources, retrieval rules, and processing parameters rather than requiring re-training of the AI/ML model itself. Administrators can pre-configure which structured data sources the chatbot should access, how data should be retrieved and processed, and what business rules should be applied. This preliminary setup allows rapid customization without the computational cost of model re-training.
Solution Approach 2:
The system allows different chatbots to have customized local configurations (specific data sources, retrieval parameters, processing rules) while sharing the same underlying AI/ML model infrastructure. Each chatbot is customized at the configuration level rather than requiring a completely different trained model, enabling efficient resource utilization while maintaining adaptability to specific use cases.
Data Source
AI summary
Systems, methods, and apparatus, including computer-readable media, for managing and deploying customized artificial intelligence chatbots. In some implementations, a system receives data indicating user input instructing a chatbot to be created and an indication of a data set for the chatbot to use. The system creates the chatbot based on data objects in the data set. The system provides a code segment or module configured to cause a chatbot interface for interacting with the chatbot to be embedded in a user interface. The system receives a user prompt provided for the chatbot through the user interface. The system provides a response to the user prompt from the chatbot, and the response from the chatbot includes text generated by one or more artificial intelligence and/or machine learning (AI/ML) models using values from the data set.