AI Lead Distribution2

Digital Innovation

Engineering Services

Technology Solutions

How Generative AI

Led to a 30% Increase IN Leads

For Mortgage Brokers

Introduction

2023 & AI

2023 was a remarkable year which made an undeniable awareness about AI in the general public.

ChatGPT

Large part of this credit goes to Generative AI with end-user oriented platforms like ChatGPT.

Large Language Models

Generative AI is built with a core component called Large Language Models (aka LLM). Power of LLMs can be used in a wide variety of non-trivial use cases.

Challenge

Analytics is a common tool for gaining insights out of large amount of “structured data”. Many organizations widely use Analytics for their day-to-day decision making which affects success of their business functions.

Businesses also deal with a verity of “unstructured data” such as emails, notes, support tickets etc. Those data sources contain insights of immense business value. Unfortunately, analytics don’t go very well with unstructured data. Usually unstructured data require “human eye-balling” or very expensive tools to extract data in a useful format.

Solution

Instead of traditional use of Natural Language Understanding (NLU) models, we employed the “reasoning” capability in LLMs.

With adequate pre-processing and creative prompt engineering, LLMs tend to perform a good job in converting unstructured data into a structured form that can be used in Business Intelligence tools, eventually present insights in a graphical form.

Result

As a result of above methodology, iTelaSoft was able to come up with an initial analysis of a large Survey feedback dataset within weeks.

Usual method of human review and reporting was estimated to take months, costing a significant amount.

Tech Behind It

There are certain LLMs that are fine tuned for reasoning, which is a good contender for these type of use cases. Anyway, data privacy is a major concern from many of our customers against using public services like OpenAI. Die to that iTelaSoft used it’s internally managed LLM infrastructure which is securely ring-fenced. Traditional method of achieving the same result by using NLU and Knowledge Graphs is much more faster and power efficient than LLM Reasoning method. But initial solution development time is significantly shorter with LLMs which make this method far beneficial for short to medium term use cases.

Want AI to improve
your business outcomes?

We’ve been designing and implementing cutting edge AI and machine learning solutions for over a decade.

If you want to know what’s possible for your business,

Please contact our team

Let us know how to help

We love to hear what challenges you go through in your business and discuss how we can assist you.