Try out a Chat Package

Click the button "Copy Chat Package to Clipboard".
Then paste the Chat Package into ChatGPT.
And start chatting with the Chat Package.

AiWerkz uses this copy/paste method so that your chats stay within your own ChatGPT account (or other AI system).

One minute to set up a document for chat!

We are able to make this process so simple and easy because we take your document and package it in a way where the user can simply paste the information into ChatGPT (or other AI systems) or if the AI System (like ChatGPT 4o) can accept a URL then the user only needs to paste the link and then start chatting on the content.

And if you have multiple related documents, we can tie them all together.

The trade off for all this simplicity is that the user will have to paste information in to the AI system that they are using. But there are some great benefits to this:

• Get your users chatting with your content very quickly.
• Your users chats stay private between them and the AI system that they want to use.
• Super simple and cost effective. No servers to set up. No complicated configuration. Very inexpensive.

To create Chat Packages we use a prompting style that we call Interactive Chat Enabled Packages (we call them ICE Packages). Chat Packages that can be shared with other people and are used by pasting them in to ChatGPT and other AI systems. This enables people to chat with your document.

ICEppl (ICE People) is the markup language that we use to describe ICE Packages. No need to learn ICEppl as ICE Packages are automatically created.

Learn more about ICE Packages and ICEppl at About Ice Packages

Rethinking RAG with Chat Packages

If you have documents that you want your users to be able to chat with you have likely come across RAG. We are an alternative.

Retrieval-Augmented Generation (RAG) integrates with Large Language Models (LLMs) to enable AI chat systems to utilize external documents that are not part of the model's training data. Typically, RAG involves the following steps:

1. Document Repository: A collection of documents is stored in a repository.
2. Vectorization: These documents are converted into vectors using a process called vectorization, often utilizing embeddings generated by a machine learning model.
3. Query Handling: When a user submits a query through a customer-facing web app, the query is also vectorized.
4. Search Function: The vectorized query is then used to search the document repository to find the closest matches based on vector similarity.
5. Retrieval and Generation: The relevant documents are retrieved and provided as context to the LLM, which then generates a response using both the retrieved documents and its own trained knowledge.

Along with cost and complexity, another downside to RAG systems is that the operator of the RAG system potentially has access to all information shared in the user chat session, which raises privacy and security concerns.

User Privacy AiWerkz, using ICE Packages and ICEppl, removes the operator of the RAG system from having access to the chat session. We do this by packaging content in a special Chat Package that the user can paste directly into their AI system of choice (like ChatGPT). So, one may consider an ICE Package as a sort of improvised RAG.

RAG done Easier, Cheaper and Faster AiWerkz Chat Packages are the fastest, easiest and least expensive way to share "chatable" documents with your customers.