Startup story #15 - Graphsight

Every day, inaccurate renewable energy forecasts lead to substantial financial losses for key stakeholders in the energy market. The complexity of this process is amplified by the large number of variables involved, often resulting in inefficiencies, imbalances and regulatory penalties. Graphsight tackles this issue by using an advanced graph-based machine learning approach, capable of analysing and predicting time-dependent data to optimise trading strategies and enhance grid stability. In this interview, Slobodan Lukovic, co-founder with Prof. Cesare Alippi and CEO, shares insights into their technology and vision.

How did the story of GraphSight begin?
I have been working in the field of AI applications for energy efficiency for over a decade. One day, during a conference in Brussels, I learnt about the spot energy market for renewable trading and found a strong fit with the graph-based AI models we were developing. Once back in Lugano, I started looking for potential partners and, in a short time, found a promising one in DXT Commodities. Our collaboration led to the creation of an Innosuisse project, as we realised that our solution could provide significant value to the market.

How does it work? 
Our technology applies graph-based deep learning models to time series data, a particularly valuable system in complex and interconnected markets, to analyse dynamic interaction across multiple solar portfolios. By capturing spatial and temporal dependencies, such as weather conditions or solar farm maintenance, our models adapt energy predictions to provide precise day-ahead and intraday energy forecasts for key players operating in the renewable energy sector. Currently, we are focusing on improving our system to account for extraordinary events like sudden weather fluctuations or unexpected grid failures.

Who would benefit the most from your solution?
Our main customers are Energy Traders (e.g., SDXs, DIEs), Distribution System Operators (DSOs), Grid Operators, Renewable Energy Developers, as well as other entities involved in managing or trading renewable energy. 

What have you achieved so far, and what are the next milestones?
We have successfully onboarded our first client, integrating our algorithms into their current system. The next step is to establish an API connection, enabling third parties to integrate with our solution and supporting our expansion into international markets in the next few months.

3 quick questions to wrap it up:

• How do you recharge your batteries? - Going to the gym, snowboarding, and spending time with my nephews.
• Which skill any founder should master? - There is no silver bullet, but curiosity and persistence always help.
• If you had one wish to be realised tomorrow, which would it be? - To live in a cleaner, safer, and more balanced world.