ChatGPT-based AI Expert Systems launched

Today we launched our AI Expert System. An AI Expert System is basically ChatGPT - The difference being that it knows everything about your problem domain. To illustrate the value go to ChatGPT and ask it to write a 5 paragraph blog article about your company's products. Unless you're IBM, Microsoft, or a huge company, chances are it will flat out hallucinate and generate random stuff.

This is because ChatGPT is a "general middle of the road" AI system. Our AI Expert Systems turns ChatGPT from a generalist into a specialist in 5 minutes.

What is an AI Expert System?

A ChatGPT-based AI Expert System is kind of like our chatbot product, except it can be secured, only allowing your employees access. This allows you to upload private data to it that you don't want to expose to people outside your organization. Do you have a contract that needs to be translated into 15 different languages? No problem, our AI Expert System's got your back! Need to run through 150 CVs to find the best candidate for a vacant position? This is a 5 second job for our AI Expert System. Use cases might be.

• Recruitment
• Legal aid
• Translation services
• Content production for digital marketing
• And many more

Train on your data

With our AI Expert Systems you can scrape websites, upload PDF files, XML files, CSV files, JSON files or YAML files - And you can "spice" your model with individual web pages from any URL. In addition you can manually add and edit your training snippets using our Magic dashboard. You can also choose which base model from OpenAI you want to use as the foundation of your AI Expert System. Our AI Expert System supports text-davinci-003, gpt-3.5-turbo, gtp4, and any other models that OpenAI have given you access to. If you have access to OpenAI's gpt4-32k model, you can even write books with 50+ pages using our AI Expert Systems. Below is an example of scraping a website for data.

Configuring your AI Expert System

The configuration options you get with our AI Expert Systems is quite frankly staggering. First of all you can create multiple "models", allowing you to have one AI Expert model for your marketing department, another for your legal department, and a third model for your HR department. Each model and employee can be given different access, based upon an RBAC (Role Based Access Control) system, giving you complete control over who have access to what. In addition you can configure individual models to use different based models from OpenAI, and each model in your AI Expert System being based upon different data.

For the extreme use cases you can even import training data and "fine-tune" your models, resulting in a custom Machine Learning model based upon one of OpenAI's base models. This is not for the faint at heart I must confess, but it is possible, and for some extreme use cases also necessary.