All the text documents your organization has collected have untapped potential. Using natural language processing, machine learning and large language models (LLM), I can help unlock it. With tools such as GPT4 or Mistral, you can see results in less than a month. Get in touch to learn how I can help.
Below are two use cases that I’ve been focusing on:
Chatbots
I’ve been doing chatbots since 2018, when you still had to hand design the answer tree and chatbots would often misunderstand questions.
With the advent of LLMs, chatbots are much more powerful, understand language much better and can provide much more value. I’ve done chatbots that help employees navigate the internal policies of the company. This helped them save time that they previously used to ask their colleagues about how to proceed in various cases. I’ve also done customer support chatbots that reduce the load on customer support. Starting from an internal knowledge base, these chatbots can answer even novel questions a customer might have about the products or services a company has.
Data extraction
Language is messy and this makes understanding text data hard. I’ve done projects where I would take raw text data and then using LLMs extract structured information from them. The raw text can come from various sources, either from web scraping or from PDFs and Office files.
For example, a company was looking to process tens of thousands of real estate documents, looking for certain fields. These fields were legally required to be there, but there was no standard for how the documents should look like or even how the fields should be named. I wrote a parsing tool that used GPT4 to understand those documents and extracted just the fields that were of interest, in a well structured format.
If you would like a chatbot for your company or if you have a large amount of documents that you want to extract data from, reach out to me for a quote.