Best Practices for Using AI Chat

Earlier this year, Domo launched AI Chat as part of its latest wave of features designed to amplify the impact of our platform for customers. And we’ve been thrilled to see so many Domo users embracing the power of this personal data assistant to quickly extract deeper insights that can generate real value for their companies. But not everybody is familiar with how AI Chat works, and even those who may need a refresher on how to use the tool.

In part 5 of our AI Insights Livestream series, we asked a group of Domo software engineers and architects a simple question: What do you need to know to make the most of AI Chat? If you’re curious about using AI to its full potential, read on to learn what our experts had to say. (Links to the recorded session are also available below.)

What is AI Chat? 

Before we get into best practices, let’s take a moment to review what AI Chat actually does.  

AI Chat is a software tool that allows people to ask questions in natural language about their data and quickly generate answers in both narrative and visual responses, for smarter, faster decision-making. The tool can speed up your data analysis by identifying the most relevant information based on the specific dashboard or app in use. Plus, it can suggest follow-up questions, encouraging a real conversation with the data.  

It also leans heavily on transparency. AI Chat provides a clear, step-by-step breakdown of how it answered the question, so you can easily verify each step in the process, eliminating “black-box AI.” Any charts or visualizations that are generated can be saved so you can continue to dig deeper into the data.  

For a deeper understanding of the mechanics behind the tool, Jeremy Rodgers, a senior software engineer at Domo, reviewed AI Chat’s architecture in our livestream:

How to get the most out of AI Chat  

Now that you understand why this tool is a game-changer, it’s important to understand how to use it effectively.  

1. Pick the right context 

Chat can determine which data to use to answer your questions accurately by relying on two types of context: inferred and supplied. 

The first type is inferred context, which includes everything visible on the page or referenced on the page when you enter your question into the chat. This can cover cards, dashboards, apps, and data sets. If you aren’t sure what is included, simply click the chat’s focus icon to see a list of everything being used. This can come in handy if the chat is generating an answer that appears to be based on the wrong data set—all you have to do is toggle certain elements off.  

2. Prepare your data to be AI-ready 

If you’ve been working with data long enough, you are probably familiar with the phrase “Garbage in, garbage out.” It may be a cliché, but your results really are only as good as the data you put in. If you ask the AI Chat to work with a data set that has poorly named columns or names that don’t accurately reflect how you otherwise refer to the information, the responses you receive are unlikely to meet your needs.  

3. Ask good questions 

To get the best answers out of your AI Chat agent, you need to be as explicit as possible when asking questions. When you leave out important details, the agent will try to fill in the blanks based on a probability distribution of possible answers. It can certainly get the answer right, but it leaves more room for an incorrect response.  

4. Review your answers 

With AI Chat, we are prioritizing transparency. At each step of the process, you have the ability to verify the pieces used to generate the answer. For every question, you can see the data set that was provided, the SQL query that was generated, and the resulting data that was returned. And we do this so that you can review important steps to check the logic applied.  

Put your AI personal agent to work   

You now have the knowledge to put it to good use, so accelerate your data analysis and workflow with AI Chat. Domo’s AI tools and services are giving users the ability to streamline their data exploration and understand their data in new ways that can transform how their organizations operate.