Kelvin Aims to Contribute to Environmental Conservation Through the Use of AI in Residential Energy Assessments

Exploring startup ideas with the potential to mitigate climate change might lead one to become proficient in conducting home energy evaluations. This was the path taken by the creators of Kelvin, a France-based enterprise employing computer vision alongside machine learning for simplifying energy audits in residential spaces.

The journey for Clémentine Lalande, Pierre Joly, and Guillaume Sempé into the realm of home energy efficiency began with the realization that renovation plays a pivotal role in cutting down energy use and CO₂ emissions. However, they noticed a lack of technological integration within most entities in this field, much like the larger trend in the construction sector.

“Europe alone has a target to refurbish 300 million homes in the coming three decades,” Kelvin’s Chief Executive, Lalande, shared with TechCrunch. “Yet, the construction sector ranks just ahead of agriculture in terms of digitization.”

In an aggressive push for sustainability, France’s National Housing Agency (ANAH) aims for 200,000 home renovations by 2024. However, the current pace of craftspeople cannot meet this demand, thereby exacerbating environmental strains. On a broader scale, European regulations seem to be leaning in favor of such innovative ventures.

Kelvin, which commenced operations in October 2023, operates purely on software innovation, steering clear of building a service provider marketplace or adopting a customer-facing stance as seen with Enter, a German counterpart in the home energy evaluation sector, also spotlighted by TechCrunch.

The startup has assembled a compact engineering team charged with the development of its proprietary AI model, focused exclusively on home energy evaluations via machine learning. They harness both open-source data such as satellite imagery and their extensive database comprising millions of photos and energy assessments.

“Our approach integrates over 12 sources of proprietary, semi-public, or open data, offering insights into the construction and thermal characteristics of buildings. By employing standard segmentation methods and analyzing satellite imagery through machine learning algorithms, we can identify features including solar panels, adjoining structures, and collective ventilation systems,” explained Lalande.

Additionally, Kelvin is pioneering a novel remote inspection mechanism, which employs a bot to guide users on capturing essential photos and videos. “Our technology can identify radiators from video content, discern doors, measure ceiling heights, and ascertain the types of boilers or ventilation systems present,” Lalande added.

Opting against the use of 3D technologies such as lidar, Kelvin aims to make its tool scalable and accessible, allowing for the use of regular photos and videos without necessitating advanced smartphones equipped with lidar sensors.

Potential beneficiaries of Kelvin’s services span construction entities, real estate agencies, and financial institutions keen on funding home renovation endeavors. These stakeholders are particularly interested in precise evaluations to inform their decisions.

Initial trials by the company have revealed that its assessments align closely, within a 5% margin, with traditional methods. Achieving widespread adoption as the preferred audit tool could greatly streamline comparisons of homes and renovations.

Kelvin has successfully secured €4.7 million ($5.1 million based on current exchange rates) in funding, led by Racine² with additional non-dilutive backing from Bpifrance. Seedcamp, Raise Capital, Kima Ventures, Motier Ventures, and a host of angel investors also contributed to the funding round.