Create an AI Analyst

An AI Analyst in Athenic is your workspace to chat with your data and get answers in natural language.

You can create an AI Analyst in two ways:

• From the Home page: click "New AI Analyst"

• From the Data Sources page: click "Create AI Analyst" (if you have already connected your Data Source)

Create project in Home page

Create project in Data source page

Set Up your AI Analyst in 3-6 Steps

1. Connect a Data Source

Start by selecting an existing data source or uploading a new one. Athenic supports multiple formats, including CSV uploads and database connections.

For more information, see: Connect a Data Source

Select your data source

2. Create Dataset(s)

After connecting your data source, you’ll create a Dataset—a selection of the relevant tables and fields you want your AI Analyst to understand.

There are two ways to create datasets:

• Basic Datasets (no SQL required): Select tables and columns directly

• Advanced Datasets (SQL required): Write custom SQL for complex logic

After choosing or writing your datasets, you can:

• Adjust field names or data types for clarity

• Remove irrelevant or sensitive fields

• Select key tables and define relationships

Once you've created your datasets, use the Knowledge Graph’s drag-and-drop editor to:

• Define how datasets join together

• Embed shared business logic

• Link data to business-friendly concepts

For detailed instructions, see Creating Datasets and Data Preparation

Select Dataset and Columns

3. Add a Dataset to the Knowledge Graph

After saving a dataset:

1. Click the dataset to add it to the Knowledge Graph

2. Add context—describe what the dataset contains and how it should be interpreted

   Examples: definitions of key fields, business rules, KPI explanations

3. Define joins—connect this dataset to others to show how they relate

4. Link shared context—map reusable concepts like “region,” “channel,” or “active user” across datasets

Learn more in Understanding the Knowledge Graph

Add context to Dataset

Joins annotate how datasets are combined

Optional: Add Suggested Questions

To help users get started quickly, you can add example questions that will appear in the chat window before they type.

These are especially helpful for:

• Guiding non-technical users

• Demonstrating use cases

• Onboarding new team members

Suggested questions will continue to update based on conversation history and usage.

See Suggested Questions for more.

Add Suggested Questions

Optional: Define Key Terms

Key terms help guide users as they type questions—especially those less familiar with the data. Admins can define specific business terms (like “churned customers” or “Q1 revenue”) and provide helpful descriptions or context.

These definitions appear in the question builder and serve two purposes:

• Improve question clarity by anchoring ambiguous language to known business concepts

• Assist users by surfacing common terms they can ask about, along with explanations

This makes it easier for teams to ask precise questions and ensures consistent interpretation across your organization.

Add Key Terms

Optional: Add Default Question Prefixes

You can optionally set Default Question Prefixes to automatically add filters or context to every question. This is useful when there’s a default business assumption that should apply unless stated otherwise.

Add Default Question Prefixes

Optional: Add Natural Chat Prompt

Guide how the AI interacts with users by setting a custom prompt. Use it to define tone, behavior, or requirements—e.g., “Ask for a date range if not provided,” or “Warn if X data is joined with Y.”

Add Natural Chat Prompt

Finalize and Save

Once you’ve completed all steps:

• Give your AI Analyst a name

• Click Save and return to the Home page, where your new project will appear

You’re now ready to start asking questions, exploring data, and sharing insights.

Finish setting up AI Analyst

You will find the Suggested Questions, Key Terms, Default Filters in the Project as followed:

AI Analyst view