Trusted Data, Shared Meaning: The Power of Data Products with Semantic Models

You've already invested in your data, analytics, and business logic. How can you get more out of that investment as you explore agentic AI?

As organizations explore agentic AI, one thing is becoming clear: AI needs more than access to accurate data. It also needs the business context to understand what that data means.

Throughout this blog series, we've looked at how your Qlik investment can help you put agentic AI to work. You can ask questions about your data, bring Qlik intelligence into the AI assistants you already use, and have agents watch for meaningful changes and help look ahead.

But behind these experiences is an important foundation: trusted data that can be reused, along with the business meaning that makes it useful. That's where Qlik Data Products and semantic models come together.

Turn the data you already trust into a data product

Qlik Data Products organize data into governed, reusable assets that support different analytics and AI use cases.

They make data easier to discover and reuse, with information that helps you understand its origin, ownership, and governance. And you don't have to start over to create one. You can build on the trusted data you already work with in Qlik, as well as data in platforms such as Snowflake and Databricks.

Instead of preparing the same data for every new use case, you can make trusted data easier to find, access, and reuse.

Add shared meaning and context with semantic models

Trusted data tells you what happened. Business context helps you understand what it means.

Consider a simple question:

Two applications might use the same accurate data but return different answers because they calculate margin differently. One might include discounts or shipping costs, while another doesn't.

Now imagine AI agents using that data across different business processes. Without consistent definitions, those differences can lead to conflicting answers and decisions.

That's why we've introduced Semantic Models in Qlik Data Products.

Semantic models let you add business meaning directly to your Qlik Data Product. You can define measures, dimensions and relationships, and connect business terms such as “sales,” “margin,” and “customer” directly to your organization's business glossary. This means your data products can provide more than trusted data. They can also carry the context needed to understand and interpret that data consistently.

For your teams, that can mean less time reconciling different definitions, more confidence in the metrics they use, and a clearer starting point for building new analytics and AI experiences.

And you don't have to recreate all your existing business logic from scratch. With the new public preview capability, you can extract a data product directly from an existing application. The resulting data product includes the application's data and a semantic model of its relationships, dimensions, and measures. Read the public preview post to learn more on how to enable the capability and get started.

VIDEO: https://videos.qlik.com/watch/FZD6LqhLT77sgWeLbKzh7P  

Build on this foundation for agentic AI

As AI becomes more embedded in everyday business processes, trusted data and consistent business meaning become even more important.

Each serves a different use case, but all depend on trusted data. Data Products are designed to work across these experiences, so you prepare the data once and build from there.

Qlik Data Products and semantic models extend into Qlik Answers and AI agents connecting through Qlik MCP Server, so the shared business definitions and context you build in a data product reach your AI workflows to interpret your data more consistently. You'll also be able to use AI to help build and maintain your data products, semantic models, and business glossary as your data evolves.

From Reactive to Proactive: Let Agents Watch and Forecast For You

Catch changes sooner, then figure out what could happen next, using the same governed data foundation as the rest of your Qlik environment.

In our blog last week, Ask Your Data Anything, From Anywhere, we looked at what happens when you ask your data a question, whether in Qlik or through the AI assistant you already use through Qlik MCP Server.

But what happens when an insight doesn't start with a question?

Sometimes, you don't know what to ask because you don't know that something has changed. A metric starts to slip, a trend shifts, or something unexpected shows up in your data. You don't know when to look for it until it shows up in the report, and it's already happened.

Now let's look at how Qlik can help you catch those changes sooner and figure out what could happen next.

Together, they help close that gap using the same governed data foundation as the rest of your Qlik environment.

Find out sooner with Discovery Agent

Instead of relying on someone to spot a change in a dashboard, Discovery Agent lets you set insight triggers around the measures and dimensions that matter most in your Qlik applications.

As your data updates, Discovery Agent evaluates those triggers and surfaces meaningful changes in a prioritized Feed and daily digest, helping bring important changes to your attention.

Open the insight to see what happened. When you want to understand why, jump into Qlik Answers and ask a follow-up question.

Ask what's coming next with Predict Agent

Knowing something changed is only half the answer. The next question is usually: “So what happens next?”

That's where Predict Agent comes in. Inside Qlik Answers, Predict Agent brings Qlik Predict's predictive capabilities into a conversation. You can ask questions about forecasts, trends, and outcomes in plain language, without starting with a modeling tool.

Start with the question:

Predict Agent helps turn those questions into predictive analysis and explains the results in language that's easier to understand and act on.

Notice sooner. Know what's next.

Discovery Agent and Predict Agent have different roles. Discovery Agent helps you notice. Predict Agent helps you look ahead. Together, they connect the steps between discover, investigate, forecast, and act.

Put it to work in your role

The use case changes. The workflow stays the same:

Ask Your Data Anything, From Anywhere

Let people ask. Whether it's inside Qlik or from the AI assistant they already use, the foundation is the same: the governed data and business context you've already built.

In the first post in this series, From BI to Agentic AI: Why the Dashboards You Built Are the Foundation for Agents, we explored how the BI investment you've already built can become the foundation for what comes next. Now, let's look at what that can mean in practice.

One of the simplest places to start is changing how people access the insights already available in Qlik: let them ask.

A sales leader might want to know, “Why did margin fall in the Northeast last month?” Instead of finding the right dashboard, navigating filters, and interpreting several visualizations, they can simply ask the question.

And they can ask it in two ways: directly inside Qlik with Qlik Answers, or from an AI assistant they already use, such as ChatGPT, Claude, Gemini, Copilot, or Amazon Q, through Qlik MCP Server.

Ask more from your data inside Qlik

Answering all your questions, no matter how you ask them

You already have data and analytics inside your Qlik applications. Qlik Answers gives more people an accessible way to explore them using natural language, without requiring them to know how a dashboard was built or where to look for a particular insight.

Instead of searching through sheets, filters, and visualizations, you can ask a question such as “Why did margin fall in the Northeast last month?” and let Qlik determine the data and analysis needed to answer it.

And you don't have to stop with the first answer. You can ask follow-up questions to explore a finding further, request recommendations, and uncover insights in your data that could impact your business.

Behind the scenes, Qlik's analytics engine enables agents to perform advanced analysis and reasoning to determine what data and analysis are needed to respond. Agents can draw from Qlik data, internal unstructured content, and general knowledge. They can also break down compound questions and execute multiple queries to develop a more complete response.

Answers can include visualizations and supporting sources, and users can inspect the reasoning trace to better understand how a response was produced.

That means more people can move beyond simply seeing what happened and start exploring why it happened and what deserves their attention next.

And because Qlik Answers builds on your existing Qlik environment, you don't have to rebuild your analytics foundation to get started. You're extending the value of what you already have.

Video: 

Bring Qlik intelligence to the AI assistants you already use

But your users don't always start their questions inside an analytics application.

Increasingly, people are working from AI assistants such as ChatGPT, Claude, Gemini, and Copilot. Qlik MCP Server lets you bring Qlik's data and analytical capabilities into those experiences, so people can access trusted intelligence without leaving the tools where they're already working.

Take that same question: “Why did margin fall in the Northeast last month?”

Asked through a connected AI assistant, Qlik MCP Server can provide access to relevant governed data and Qlik's analytical capabilities to help answer it.

And this is about more than giving an AI assistant access to raw data. Qlik can also provide the business context captured in your analytics environment, helping the assistant work with the measures, dimensions, calculations, and terminology your organization already relies on.

That context matters. Rather than asking an AI model to interpret a table of data on its own, you can connect it to an analytics foundation designed to understand relationships in your data and perform the calculations needed to uncover meaningful insights.

No separate analytics experience to learn. No need to move between tools just to ask the next question.

Two doors. One trusted foundation.

Where should you start? It depends on your company's AI strategy and how your teams prefer to access insights from their data.

Your users can access trusted insights without needing to know how to build a dashboard or Qlik app. Or use both.

The goal isn't to force teams into a new way of working. It's to give them the flexibility to work in the tools that make sense for them while maintaining trust, governance, and business context.

You've already built the data and analytics foundation. Now you can extend that trusted intelligence across your organization, making it available wherever questions are asked and decisions are made.

And asking questions is only the beginning. In the next post, we'll look at what happens when agents don't wait for you to ask — when they continuously watch your data for meaningful changes and help you anticipate what's coming next.

I've been exploring Qlik's Agentic AI capabilities and recently tested the Discovery Agent using a retail scenario focused on a seasonal RainWear product category.

The use case was inspired by a real client engagement, with business labels anonymized to protect confidentiality.

Rather than waiting to search dashboards, the end users wanted to see whether AI could proactively monitor performance and surface business events that require attention.

Attached screenshots showcase how the Discovery Agent automatically identified:

• Record high sales during the monsoon season
• Record low sales during the winter period
• Higher-than-expected sales performance
• Regional sales downtrends requiring investigation

What stood out wasn't just anomaly detection. The Discovery Agent:

✅ Observed business activity continuously

✅ Detected meaningful changes automatically

✅ Explained insights in natural language

✅ Guided users toward further investigation

This felt less like traditional BI and more like having a digital business analyst continuously monitoring business performance and proactively surfacing insights that matter.

While the screenshots have been anonymized (RainWear, Store Cluster A, North Zone), the insight patterns are based on an actual client-inspired scenario and demonstrate how Agentic AI can proactively surface business-critical events.

Looking forward to hearing how others in the community are using Discovery Agent or other Qlik Agentic AI capabilities.