Kaleidescope ML+AI Demo

In this video, we demonstrate an example of combining machine learning and language models into one real-world customer use case. Often these two types of models are confused as being the same thing.

The use case is a custom application for creating job estimates. A salesperson enters a natural language job description. A language model then converts the description into estimate line items using a set of rules and mappings.

A machine learning model has been trained on past job data to evaluate the initial line item estimates and improve their accuracy based on actual historical job requirements and outcomes.

The process shows how the language model handles the natural language processing and conversion using rules, while the machine learning model brings in predictive analytics to refine the outputs based on patterns in the data.

By combining both models, the application can take unstructured text descriptions, turn them into structured estimates, and then optimize those estimates to be more accurate based on historical data analysis. This improves the profitability of projects by inserting machine learning into the estimation process while still allowing a conversational user experience via the language model.

You can see here a clear delineation between the distinct roles that language models and machine learning models can play together in an integrated real-world application.