Building Your First AI Model with AI Hub: A Step-by-Step Guide

We often hear about how AI can unlock huge value for businesses, but for many people, the hardest part is simply knowing where to start.

The good news? If you’re working with Microsoft Power Platform, you don’t need to be a data scientist or machine learning engineer to build something impactful.

With AI Hub, Microsoft has made AI accessible to low-code developers, business users, and citizen technologists. In this post, I’ll walk through how to build your first AI model using a no-code approach.

Why Start with AI Hub?

We all have a growing list of tasks and manual processes we’re trying to streamline. Automating even a few of these can free up valuable time and improve decision-making. That’s where AI Hub fits in. It allows you to embed intelligence directly into your apps and workflows without needing a background in data science.

• No code required
• Integrates easily with Power Apps and Power Automate
• Built on Azure AI
• Comes with practical templates: prediction, classification, object detection, and more

AI Hub Prerequisites and Licensing

Before you start, check that your environment meets the following requirements:

• Dataverse must be provisioned in the environment
• AI Builder capacity must be available, or you can start a free trial

Licensing options include:

• Power Apps or Power Automate per-user plans with AI Builder credits
• Standalone AI Builder add-on
• Free trial in a trial environment
• Microsoft 365 Developer tenant with trial capacity enabled

If you don’t see AI Hub or AI Models in your environment, check that you’re in the correct environment (top right of make.powerapps.com), that Dataverse is enabled, and that AI Builder credits or a trial is active.

Important Note Before You Begin: AI Hub requires at least 50 rows of data to train a prediction model. If you have fewer than 50, training will not begin. In most cases, training on 50–100 rows completes in under two minutes.

Step-by-Step: Build a Prediction Model

Step 1: Create the Model

• Go to https://make.powerapps.com
• Switch to an environment with Dataverse (make sure you have data that you want to use)
• In the left-hand menu, go to AI Models or to AI Hub and then AI Models
• Click + New AI model
• Under “Most popular”, select Prediction

Step 2: Define Your Business Outcome

Give your model a name like, and choose the binary outcome you want to predict (e.g., Yes/No).

Step 3: Choose Your Data

• Select a Dataverse table

The dataset must include:

• Input columns (like Age, Region, etc)
• An outcome column with two distinct values
• Have more than 50 lines

Step 4: Map the Outcome Column

Select the column that contains the value you want the model to predict (e.g., RenewalStatus). This column must contain binary values (e.g., Yes/No).

Step 5: Select Input Fields

Choose the fields that influence the outcome

Step 6: Train the Model

Click Train. It usually completes in under two minutes for smaller datasets.

Step 7: Publish the Model

Click Publish to make the model available for use in Power Apps and Power Automate or PowerBI.

After Training: Using Your Model

In Power Automate

• Use the “Predict” action to call the model in a flow
• Trigger actions based on the predicted outcome
• Log results to Excel or Dataverse

In Power Apps

• Add the AI model as a connector
• Use the Predict() function to return predictions

In Power BI

• Use Power Automate to write prediction results into a table or Excel
• Pull that dataset into Power BI for dashboards and insights

Final Thoughts

Getting started with AI inside the Microsoft ecosystem no longer requires a data science background or complex tooling. With AI Hub, you’re working within the same low-code interface that powers the rest of the Power Platform — and that means you can go from idea to implementation incredibly fast.

The key is to start with clarity: What decision are you trying to support? What behaviour are you trying to predict or influence? Once that’s clear, everything else becomes a structured process — identifying relevant historical data, selecting the right fields, and training the model using a tool that guides you each step of the way.

But — and this is important — AI is only as good as the foundations it’s built on. Clean, consistent, well-structured data makes all the difference. AI Hub lowers the barrier to entry, but it still rewards thoughtful preparation. The models you build will only be as good as the data they learn from.

As someone focused on Power Platform and AI, I see AI Hub as a major bridge between technical capability and business insight. It allows makers and analysts — not just developers — to embed intelligent decision-making directly into the apps and flows they create every day.

And the value doesn’t stop at predictions. Once you have a working model, the possibilities open up: automated escalation flows, dynamic user guidance, contextual Copilot experiences, and even natural language-driven business automation.

So if you’ve been waiting for the right moment to explore AI — this is it. Use what you already know, connect it to the data you already have, and let AI Hub accelerate your ideas into action.