Preparing Your Data for AI in Power BI

Artificial Intelligence is no longer a supplementary feature in Power BI. It is rapidly becoming the primary way users discover, analyze, and interact with data. With Copilot in Power BI, business users can ask questions in natural language, uncover patterns automatically, build visuals instantly, and even generate complete reports with minimal manual effort.

Organizations need to strengthen their Power BI basics. They must focus on preparing and transforming data, building models, and optimizing performance. This will help them make the most of Power BI’s AI features. As a result, there is a growing demand for organized learning paths. An example of this is the Microsoft Power BI Data Analyst role.

What Is Prep Data for AI (Pre‑AI) in Power BI?

Prep data for AI is a set of capabilities in Power BI that allows semantic model authors to intentionally guide Copilot’s reasoning over data. Traditional data modeling focuses on relationships, measures, and performance optimization. Pre‑AI takes this one step further by optimizing semantic models specifically for AI‑driven experiences.

At its core, Pre‑AI answers one key question:

What data should AI focus on, and how should it interpret that data?

To achieve this, Microsoft introduced three tightly integrated features under the “Prepare your data for AI” section in Power BI.

AI Data Schema

Large enterprise Power BI models often contain hundreds of tables and columns – many of which are technical or rarely used. The AI data schema allows authors to define a focused subset of fields that Copilot should prioritize when answering questions.

By simplifying what AI sees, authors can significantly improve Copilot’s accuracy and reduce ambiguity. Microsoft provides detailed guidance on setting this up in AI data schemas for Power BI.

AI instructions allow model authors to embed business logic, terminology, and analytical preferences directly into the semantic model using natural language.

With AI instructions, you can:

• Define how revenue, churn, or utilization should be calculated
• Explain business‑specific terminology
• Guide how Copilot should frame insights and trends

This capability helps Copilot in Power BI think like your business, not just read your data. Learn more in Prepare your data for AI: AI instructions.

Verified Answers

Verified answers enable authors to confirm trusted explanations for key visuals. When users interact with Copilot, these validated insights are prioritized to ensure consistency and reliability across reports.

Why Preparing Data for AI Matters

Many organizations assume that enabling Copilot alone will unlock AI value. In practice, unprepared models often lead to vague, inconsistent, or misleading insights.

Improves Accuracy and Trust

Without guidance, Copilot in Power BI must interpret the entire semantic model, which can lead to incorrect field selection, especially when columns like Revenue, Sales, or Amount appear multiple times. A curated AI data schema removes this guesswork and increases confidence in AI‑generated outputs.

Aligns AI with Business Language

Every organization defines KPIs differently. Terms such as customer, attrition, or profitability are rarely universal. AI instructions ensure Copilot understands these definitions exactly as your business intends, improving relevance and interpretation accuracy.

Pre‑AI delivers smoother, more intuitive interactions, which directly improve Copilot adoption.

How to Use Pre‑AI in Power BI

Configuring Prep data for AI is a structured, iterative process that can now be done in both Power BI Desktop and the Power BI Service.

Step 1: Access Prep Data for AI

Open your semantic model and select Prep data for AI from the ribbon. As Microsoft recently announced, this capability is no longer limited to Desktop, making AI preparation easier and more scalable.

Step 2: Define the AI Data Schema

In the Simplify data schema tab:

• Select business‑relevant measures and columns
• Exclude technical keys and staging fields
• Prefer clean, descriptive naming

A simple rule: If a field confuses users, it will confuse Copilot even more.

Step 3: Add AI Instructions

Use the AI instructions tab to explain:

• Calculation logic for KPIs
• Preferred time intelligence behavior
• Which tables should Copilot prioritize

Well-written instructions dramatically improve Copilot’s ability to answer business questions correctly.

Step 4: Test with Copilot

Open the Copilot pane and ask realistic, business‑oriented questions. Validate that Copilot follows both your AI data schema and your instructions and refine as necessary.

Step 5: Publish and Evolve

Once published, Pre‑AI settings apply across all interactions with the model. As business needs evolve, revisit your AI schema and instructions to keep insights relevant and accurate.

Building AI-Ready Data

Well-prepared, high-quality data enables faster adoption and allows users to engage with AI more confidently. In an AI‑first world, semantic models are no longer designed solely to support visuals; they are built to support conversation. Preparing data for AI lays the foundation for meaningful, reliable, and truly business-ready conversations.