Hal9 is built on open source data technologies that allow you to do any predictive analytics and integrate to any system.

We are rolling out features to let you chat with any structured data source, no matter where it lives.


With Hal9, you get generative AI for your data, meaning that, you can connect Hal9 to your data source and answer ChatGPT-like questions, it is that simple!

In this example we are connected to a database to answer the following questions:

  • What are the sales by sales person by month?
  • What are the sales for a sales person by day?
  • How do sales people compare?


Users can use Hal9 online at hal9.com with the Basic and Pro plans. The Basic plan allows the user upload CSV files, ask data questions and get sharable data solutions in seconds. The Pro plan has other additional benefits, such as database support and it allows to embed solutions in any website. For enterprise customers we also have an Enterprise plan. You can check all our plans here


Hal9’s Enterprise plan enable your business users to answer data questions and find insights in seconds inside their own infrastructure with Hal9’s conversational AI.

This plan is recommended if:

  • you work at a company with 200+ people
  • you want to run everything on your own infrastructure, private cloud, AWS, Azure or Google account
  • you can not give access to third-party providers like Hal9

With this plan, your technical team can use Docker or Kubernetes to install Hal9 within your data center or private virtual cloud. Hal9 will then transform and visualize all data entirely in your infrastructure and only process metadata in Hal9’s LLM API.

Hal9’s LLM API lives in hal9.com and it is the only component that will be outside the enterprise. As we mentioned before, we never take a look at your data and we only use the metadata (table names, table description and the input prompt). This is usually more than enough for many of the organizations but if you happen to be in a case where the application you use cannot even access to the internet, talk to us, we have a workaround to also run the model that we are using within your organization but it is quite complex and in most cases it is not necessary at all.