.jpg)
In the current landscape of digital transformation, few cases illustrate the convergence of artificial intelligence, efficient data management, and disruptive business models as well as Float, a Danish startup that has managed to disaggregate the electricity consumption of hundreds of households down to the appliance level in real time. With a team of only three people, they have achieved what seemed reserved for large corporations: operating as a licensed energy retailer, processing 15 measurements per second per household, and offering billing based on market prices without additional margins. Behind this feat lies a fundamental technical decision: data compression in their time-series database, reaching 99.3%, making massive information storage economically viable. This milestone not only demonstrates the power of modern architectures but also opens the door for companies of any size, with the right support from technology partners like Q2BSTUDIO, to implement similar solutions through custom applications that integrate advanced analytics and cloud scalability.
Float was born from an obvious reality: most consumers do not know how much each appliance in their home consumes. European smart meters have a standardized interface that provides the total load every second, but that raw data is not useful without breaking it down. The company has developed its own hardware module that connects to that interface, a signal processing pipeline and neural networks to classify consumption, and an application with a proactive artificial intelligence agent. The storage challenge is immense: for a thousand households, 15,000 data points are generated per second. Without compression, infrastructure costs would make the flat subscription that constitutes their revenue model unfeasible. After evaluating multiple time-series databases, they found in Tiger Data (the managed platform of TimescaleDB) the necessary compression and continuous aggregations that eliminate batch processes. This choice allows them to retain the entire time series to train their machine learning models, calculate live bills, and detect anomalies like an oven left on for hours, all without overloading a minimal team. For any company looking to optimize its data architecture, the lesson is clear: a well-implemented compression strategy, combined with AI for businesses and cloud services like those offered by Q2BSTUDIO on AWS and Azure, can turn a technical challenge into a competitive advantage.
Float's platform is built on a stack that combines Azure IoT Hub, streaming on Google Cloud, and Tiger Data as the single time-series store. There, raw data, aggregations for billing, model training, and security alerts coexist. Integration with AI agents allows automating onboarding, customer service, and notifications, while detection of voltage spikes and frequency changes is done in real time. This approach, which integrates cybersecurity and business intelligence services, is replicable in sectors such as industry, logistics, or utilities. At Q2BSTUDIO, we develop custom software that allows companies to adopt similar architectures, including Power BI dashboards that visualize the behavior of IoT device fleets, or process automation systems that reduce operational load. Float's story demonstrates that, with the right tools, a small team can compete in regulated and complex markets, as long as the technological foundation is well designed. Artificial intelligence and intelligent agents are not the future: they are the present that is already transforming energy and business management.
