We are experimenting with Microsoft copilot to check the progress of ÄLLITÄ project based on published articles on Savonia web. A prompt containing links of all published articles and Gantt chart of the project was given as input. We were interested in checking the progress of WP2 and WP3 only and asked Microsoft Copilot to generate a report.

The response below was generated by Copilot.

The Story of ÄLLITÄ: Progress in WP2 & WP3

Once upon a time in Savonia, a team of innovators set out on a mission: to make heating and energy systems smarter and kinder to our planet. This mission became the ÄLLITÄ project. Its vision was simple yet bold—use Artificial Intelligence (AI) to predict, optimize, and automate energy use, while making everyday tasks easier.

Why does this matter?

Energy efficiency is not just about saving money—it’s about reducing emissions and creating sustainable campuses. The team knew that smarter systems could make a big difference for the future.

The Challenges

The journey was not easy. Data was scattered, solar panels produced energy at unpredictable times, and routine tasks consumed valuable hours. The team faced missing data, complex models, and the need for tools that could work in real life.

What the Team Did

• Collected electricity and weather data to understand patterns.

• Installed smart sensors (IoT) across campuses for real-time insights.

• Analyzed solar panel performance and discovered seasonal trends.

• Built AI tools to summarize articles and design products automatically.

• Created prediction models that reached up to 93% accuracy.

Achievements So Far

WP2 has strong models for predicting energy use and solar performance. WP3 has working AI tools that save time and spark creativity. The campuses are becoming smarter every day.

The Road Ahead

The story is far from over. Next, the team will forecast solar power more accurately, connect 5G and heating data, and make AI tools even smarter. The goal? A future where energy systems think for themselves and routine tasks take seconds, not hours.

From this line, we are writing what we learned from this experiment/pilot.

Copilot was able to understand the Gantt Chart and all the published papers. It was able to map planned tasks to the completed tasks. It gave PV forecasting task in recommended next steps because we didn’t publish article about it. It is clear from this experiment that Copilot is helpful monitoring the project performance and related tasks where context is clear and measurable whether from published articles as in our case or through some quantified results. It is useful where results are published regularly and readily available on web or results can be shared with Copilot with attached documents.

---

Authors:

Shahbaz Baig, RDI Specialist, DigiCenter, Savonia ammattikorkeakoulu, shahbaz.baig@savonia.fi

Aki Happonen, Digital Development Manager, DigiCenter, Savonia-ammattikorkeakoulu, aki.happonen@savonia.fi

Mika Leskinen, RDI Specialist, DigiCenter, Savonia-ammattikorkeakoulu, mika.leskinen@savonia.fi

Laura Leppänen, RDI Specialist, Savonia-ammattikorkeakoulu Oy, laura.leppanen@savonia.fi

---