How can hospitality teams turn scattered sustainability data into a dashboard that helps people make decisions? That was the practical focus of my 90-minute webinar for Data Science master's students at Modul University Vienna, delivered as part of the "Practical AI & Data" series.
I am Kasia Sadowska, founder of FutureHabits.Tech. In this session, I showed how vibe coding tools such as Lovable and Claude Code can support building interactive applications, with a hands-on workflow centred on using Claude to prepare the brief and Lovable to build and refine the dashboard. The goal was not another static chart. It was a clearer way to explore sustainability information in the hospitality sector.
The challenge: sustainability data needs a usable interface
Data science and operational decision-making do not automatically meet in the same place. A hotel team may have information in several files or systems, but still struggle to answer a straightforward question: what changed, where, and what should we investigate next?
An interactive dashboard can become a central hub for that information. Examples worth exploring include energy use, water consumption and waste across properties or reporting periods. These are practical dashboard ideas, not claims that we connected live hotel systems or measured environmental improvements during the webinar.
The session brought the discussion back to the business question before choosing a tool. Students raised questions about how to choose the right model for a specific business goal, whether it always needs to be an LLM, and where to start when teams and data are fragmented.
What happened in the 90-minute webinar
I walked through the path from an initial dashboard idea to a published, interactive interface: structure the problem, prepare a prompt, build in Lovable, review the user interface and publish. This is the same principle I use in my founder workshops: prepare a clear brief before spending time on the build.
1. Start with Claude and a prompt template
My starting point was Claude, not an empty Lovable conversation. Using prompt templates helps turn a broad idea into a structured brief: who will use the dashboard, what decision it should support, what data is available and what the first version needs to do.
For a hospitality sustainability dashboard, a useful brief specifies the indicators, units, reporting period and comparisons. It also describes interactions such as filtering by property or date. Claude helps organise these requirements into a Lovable prompt; it does not verify the source data or decide whether a metric is scientifically valid.
Practical tip: ask Claude to identify missing information before writing the build prompt. If the brief says "show sustainability performance", clarify how performance will be defined. A well-structured prompt is more useful than a long list of features without a purpose.
2. Give Lovable a clear first-build prompt
The next step was prompting Lovable with that structured brief. Instead of asking for "a nice dashboard", describe the intended user, the main view, the required filters and how the information should be presented.
A first version should be narrow enough to review. A summary view, a small number of charts and one meaningful filter are a sensible starting point. If real data is not ready, use clearly labelled sample data rather than allowing illustrative figures to look like verified results.
Practical tip: specify what should happen when data is missing. An empty value must not silently become zero, and a sample dataset must not be presented as an operational report.
3. Edit the user interface around the decision
Generating the first interface is only the beginning. I showed how to refine it through follow-up prompts: improve the hierarchy, adjust the layout, make labels clearer and change the way a user navigates or filters the information.
In sustainability reporting, clarity matters more than visual decoration. A chart needs a readable title, a unit and an understandable comparison. A manager should be able to distinguish total consumption from consumption per guest night, rather than treating those measures as interchangeable.
Practical tip: change one thing at a time and check the result. Test whether filters update the relevant charts, whether labels remain legible and whether the layout works on a smaller screen. A polished interface is not proof that the calculations are correct.
4. Review and publish
The workflow also covered publishing: moving beyond a private build to a shareable application that others can open and review. A published prototype gives stakeholders something concrete to react to, instead of asking them to imagine a dashboard from a description.
Publishing is not the same as production readiness. Before sharing real hospitality data, review access, permissions, sensitive information and the accuracy of calculations. A public demonstration should use data you are allowed to share.
Practical tip: open the published link as a visitor and test the central task from start to finish. Check that the intended information is visible and that private data has not become publicly accessible.
Practical lessons for a hospitality sustainability dashboard
- Define the decision first. Start with one question, such as which property needs further investigation, rather than displaying every available metric.
- Make comparisons fair. Show reporting periods, units and relevant denominators. Properties of different sizes cannot be compared meaningfully through totals alone.
- Keep the data traceable. Make the source and last update understandable. Separate missing data, estimates and observed values.
- Check the maths outside the interface. Validate aggregations and conversions against trusted source data. Neither Claude nor a generated dashboard replaces that review.
- Build for the people using it. Ask an operational user whether the dashboard helps them answer a real question, not only whether it looks impressive.
- Add complexity only when needed. Start with a useful prototype before introducing additional integrations or AI features.
The takeaway: tailored dashboards are within reach
You no longer need to begin every dashboard project with a large custom-development effort or a stack of extra tools. Vibe coding can make it easier to build a tailored interface around the data and decisions that matter. That interface can serve as a central hub for sustainability information rather than another disconnected presentation.
That does not mean spreadsheets, data engineering or specialist review are obsolete. The dashboard still depends on reliable inputs, sensible calculations and clear ownership. AI accelerates the path to an interface; people remain responsible for interpreting and using it.
For Data Science students, the useful connection is between analytical work and an application someone can actually use. The lesson was not that every problem requires an LLM. It was that a clear brief, a focused build and careful review can bring data closer to a practical decision.
Thank you to Bahareh Heidarnezhad for the invitation. My core takeaway remains: don't start with tools, start with people.
Build your next dashboard with a clearer brief
Start with the PRD Builder to structure your application idea, explore guides and resources, or see the related INiTS hands-on AI workshop. For a tailored session on AI, data or vibe coding, get in touch.

