Before jumping into the hands-on work, it’s important to understand how all the pieces fit together. The diagram bellow shows the full picture, from your data sources to the final user-facing solution, and it should help you see exactly where you come in.

Let’s walk through it step-by-step:

Everything starts with the data. With Denodo Express, once you’ve connected your public datasets (via CSVs, APIs, JSON, etc.), the real work begins: creating a semantic layer that makes your data both usable and intelligent.

• You start by building base views, which directly reflect each connected dataset. Then, you create derived views to enrich and combine your data by applying joins, filters, unions, or calculated fields to craft meaningful relationships between sources.
• Each view should be well documented. You can (and should) add view descriptions and column-level metadata to clarify what each field represents. This step improves usability for others and enhances how well the large language model (LLM) can understand and interpret your data.

Once your data is ready, the Denodo AI SDK acts as the bridge between your organized data and the LLM. The SDK includes endpoints (like getMetadata, answerQuestions) that let users ask natural language questions and get smart, data-based answers.

This whole setup follows a method called RAG (retrieval-augmented generation). Here's how it works:

• The LLM doesn’t know your data by default.
• When someone asks a question, the AI SDK retrieves the relevant information from Denodo, based on the views you created.
• Then, that info is passed to the LLM, which generates a helpful response using the data.

This means your AI solution won’t rely on guesses. It answers using real, up-to-date data that you control.

To test this setup, you can start with the built-in Example Chatbot that comes with the SDK. Later, you can build your own interface (like a chatbot, web app, or dashboard), and even connect it to tools like Microsoft Teams, Power BI, or Copilot.

In this part, your goal is to prepare the data backend, the foundation for your intelligent solution.

Your Task:

• Manage the data:
• What’s the goal?
• Integrate at least two publicly available datasets using Denodo Express to create a coherent, optimized data model aligned with your chosen scenario.
• Focus:
• Connect your data
• You’ll start by importing datasets (CSV, JSON, APIs, etc.) into Denodo Express.
• Make sure the connection is correctly configured.
• Create base views for each dataset, with clear and relevant field selection.
• Important note: Do not manually modify the datasets. All data transformations or manipulations should be done using Denodo Express.
• Create useful views
• Once your data is in, build base views with the most important fields.
• Then, create derived views by using joins, unions, groupings, and filters, to combine your datasets in meaningful ways (these views are what the AI will use later to generate answers).
• Optimize your model
• Make sure your queries run efficiently.
• Minimize the number of calls made from your final solution to the Denodo Platform by performing combinations and operations inside the Denodo Platform.
• Important note: Clearly document your view and column metadata. Add meaningful descriptions and synchronize them with the Denodo Data Catalog using the “Sync with VDP” option. This step not only improves documentation and governance, but also plays a key role in how well the AI SDK understands your data.
• Spoiler alert! The better you describe your data, the better the LLM will understand it.

Not sure where to start in defining your project? Jump to our list of suggestions below: Open Data Sources & Sample Topics

Once the data foundation is ready, it's time to build an intelligent solution powered by the Denodo AI SDK.

Your Task:

• Develop an AI-powered Solution
• What’s the goal?
• Use the AI SDK to solve a real world problem. You can:
  • Develop a functional, AI-powered application (chatbot, web app, or custom interface).
  • Or use one of the existing integrations (AI SDK Chatbot or the integrations document in the KB articles).
• Focus:
• Test your data model
• Before building your solution, make sure your data views work well with natural language. You can test this using the built-in Example Chatbot included with the SDK.
• This helps ensure your views are correctly built and provide accurate answers.
• Build your solution
• Once everything works, plug your data into:
• Tools that are already compatible with Denodo AI SDK.
• Your own custom solution like a chatbot, web app, dashboard, or anything else you imagine.
• (Optional) Make it even better:
• Add user interface features, advanced filtering, dynamic visualizations, or custom interactions.
• Make your solution intuitive, engaging, and impactful.

Challenge Guiding Questions:

• What real-world problem are you trying to solve, and why is it important?
• Which datasets are most relevant to your scenario, and why did you choose them?
• What insights or patterns can you uncover by combining and analyzing your data?
• How does your AI solution make it easier to explore or interpret the data?
• What real-world impact could your project have if applied at scale?

You’re free to focus your project on any real-world challenge that can be explored through data and AI, but if you need inspiration, we provide a list of suggestions on open data sources and a set of optional thematic scenarios. These examples are a great starting point!

Where to Find Data?

• Our World in Data – Global statistics on health, environment, energy, and economy.
• World Bank Open Data – Economic and social indicators for countries worldwide.
• Global Health Observatory (WHO) – Worldwide health indicators.
• Humanitarian Data Exchange (HDX) – Open data related to humanitarian crises and development.

Looking for Inspiration? Explore These Thematic Scenarios:

1. What Makes a Country More Sustainable?

Goal: Identify key environmental, social, and economic factors driving long-term sustainability.

Possible factors to analyze:

• Renewable energy usage
• Waste management and recycling
• Access to clean water
• Economic growth

Some datasets:

• Renewable energy consumption (% of total final energy consumption), Source: World Bank
• Municipal waste recycled, Source: OurWorldInData
• CO2 emissions , Source: OurWorldInData

2. Where Are People Going, and Why?

Goal: Analyze global travel trends to understand tourist behavior and destination popularity.

Possible factors to analyze:

• Top destinations by region
• Seasonal tourism trends (summer vs. winter)
• Travel behavior by generation (Gen Z vs. Millennials vs. Boomers)
• High vs. low season dynamics

Some useful datasets:

• International tourism, number of arrivals: WEB | API, Source: World Bank
• International tourism, number of departures: WEB | API, Source: World Bank
• International trips for personal vs. business and professional reasons, Source: OurWorldInData
• Continent Map (suggested for mapping), Source: OurWorldInData

3. What Drives a Population’s Quality of Life?

Goal: Identify key factors influencing global health and longevity.

Possible factors to analyze:

• Health expenditure (% of GDP)
• Life expectancy at birth
• Access to healthcare services
• Nutrition, education, or income levels

Some useful datasets:

• Health Expenditure (% of GDP): WEB | API, Source: World Health Organization (WHO)
• Life Expectancy (years) , Source: OurWorldInData

Each team must submit the following materials as part of their final project. These will be used to evaluate both the technical quality and the clarity of communication:

• Denodo VQL file. Export your VQL from Denodo Express, including all base views, derived views, and final views used in your project.
• AI-Driven Solution. Submit your working AI-powered solution (chatbot, web app, dashboard, etc.) that uses the Denodo AI SDK. If applicable, include source code or scripts.
• Slide Deck (in English). Include a clear, well-structured slide presentation (PDF or PPTX) that explains your process. An adequate length is between 15 and 20 slides. The slides must include:
  • A problem definition and scenario (in your own words)
  • A description of selected datasets and why they were chosen
  • A solution architecture diagram
  • A documented data management process using the Denodo Platform
  • The outcome of your AI solution
  • Final conclusions and strategic recommendations
• Video explanation (5 minutes). Record a short video, in English, explaining your project and demonstrating your solution. This must include:
  • A high-level overview of the slides
  • A live demo of your application or solution
  • Use spoken language, in English (do not use a text-to-voice app)

You can find some tips here, for the project execution.