Your PDFs are Suddenly Magical Through AI

PDF Business Documents and GPT provide a powerful solution to corporate amnesia and all without code.

PDFs are great; they are like compiled and unchangeable pools of knowledge. But they can be difficult to keep track of despite the important information they contain and your company needs to perform its best. Workers shouldn’t have to keep track of the myriad of papers and documents the organization may have to get work done. Workers need answers, and they spend more than three hours a day foraging for the content that helps them accomplish stuff.

Fortunately, PDF business documents can be combined with generative AI resulting in a powerful solution to help avoid corporate amnesia. It’s simple - start with a PDF of reasonable size, upload it into CustomGPT, and begin asking questions.

For small or large businesses, this is the #no-code equivalent of AI.

Here’s an example I did with CyberLandr’s FAQ.

This is a new and emerging segment of the AI movement to leverage LLMs in ways that help businesses. There are many ways to transform PDF documents into something that is accessible through natural language queries.

One of the most common ways to transform PDF documents into something that is accessible through natural language queries is to use Optical Character Recognition (OCR) technology. OCR technology is a type of software that can recognize text from scanned documents, images, and PDFs. It can then convert the text into a format that can be read by a computer, such as HTML or XML. This allows the text to be indexed and searched using natural language queries.

Another way to transform PDF documents into something accessible through natural language queries is to use natural language processing (NLP) technology. NLP technology is a type of artificial intelligence that can understand and interpret natural language. It can be used to extract information from PDF documents and convert it into a format that can be read by a computer. This allows the text to be indexed and searched using natural language queries.

Finally, PDF documents can also be transformed into something that is accessible through natural language queries by using machine learning algorithms. Machine learning algorithms are a type of artificial intelligence that can learn from data and make predictions. They can be used to extract information from PDF documents and convert it into a format that a computer can read. This allows the text to be indexed and searched using natural language. This is where CustomGPT and other platforms like it, manage to create a near-instant solution.

CustomGPT can also add many PDFs to one project, and projects can contain documents from 90+ formats. It is designed to embrace a business with potentially many areas of expertise and knowledge through a single chat UI.

The Tech Behind This Approach

It’s all about embeddings, an integrated API from OpenAI. The PDF is dissected and each sentence is submitted to the LLM (large language model) to get a vector, like a pointer in the language model’s content.

As soon as your workers ask a question about the PDF document, it is also vectorised and compared with all the vectors from the PDF. The top hits are used to formulate the output.

But there is one additional step that occurs silently to the workers; the output is then shaped — also using the LLM — to provide a conversational style. This makes the answers far easier to read and use for work tasks.

CustomGPT API

I love products that have modern APIs; it allows me to automate many of the processes. CustomGPT’s API makes it possible to build AI DevOps so that you can train and deploy new projects.

The ability to test queries is one critical facet of AI DevOps. For that, I use the API to push new tests to the project without testing them manually in the chat UI.

This approach is made even more productive by using Coda with a Pack that makes the API calls when I click the Test button. Coda also has new AI features ready to launch next week. These are unrelated to the way I work with business documents and CustomGPT, but they are additionally powerful if Coda happens to be your chosen business document platform.

Here’s an example of using NodeJS to call the API directly. I’ve been using Coda Packs since late 2021 with OpenAI LLMs.