Dedalo AI

Problem

The Internet is now responsible for 1.6 billion tons of CO2 emitted per year and this number will double in the next three years for two reasons: 1. The number of connected devices will jump from 13.8Bn to 30.9Bn in 2025 2. Access to the Internet represents 60% of the world's population and is growing in the most carbon intensive areas (India, China, Africa) with the aim of reaching 100% by 2030. Internet emissions are caused for 5% by data centers, 25% by network and 70% by users’ devices. 

Emissions of data centers are already measured and controlled by cloud providers, those of the network are estimated with mathematical formulas, but in any case, they cannot be controlled or reduced by companies, while those of users' devices can only be calculated with wattmeters. The problem with this hardware measurement method is that it is not scalable: to be accurate you need to buy all the existing devices, run the page a thousand times per device and manually measure the power consumption with a wattmeter. 

This process is too expensive (in the order of € millions), too time-consuming (one year with 3 full-time technicians) and too energy intensive (constant use of electricity). Needless to say, although companies are interested in quantifying this data, they do not have access to it because of the previously listed points. Moreover, without this quantification, companies have less ways of doing assessments to understand how to reduce their emissions according to changes applied to the code.

Solution

Dedalo AI developed a software to measure the CO2 emissions of websites on end-users' devices. By maintaining the same level of accuracy as hardware measurements, our software measurements allow us to be at least 40 times cheaper, 50 times faster and completely energy efficient. The first product is a tool that calculates the CO2 emissions of a web page across its usage by all end-users’ devices. Our proprietary AI algorithm quantifies how much energy a web page consumes on the CPU of any type of hardware architecture. 

To do this, the algorithm performs a static analysis of the code available on the front-end of the page, such as HTML, JS and CSS. Together with this analysis, the algorithm processes data that are imported from the website analytics through our one-click API. The processed data include: the type of hardware devices that visited the website, the browser used, the time spent per visit and their location. The device location is used to retrieve the carbon intensity of its specific area. In less than 7 days we provide the company with a CO2 emissions report that is accurate, holistic and ESG-compliant, which companies can use to make data-driven decisions and for their ESG strategies. The company can monitor their performance with a dedicated dashboard that performs the analysis on a weekly or monthly basis (depending on the chosen subscription); while at the end of the fiscal year, we provide the company with the annual report to include in their ESG disclosures. 

To compensate for these emissions, companies can buy verified carbon credits from us in order to offset their emissions and result being carbon neutral. With the second version of the product, our technology will be able to identify which pieces of code use unnecessary energy and provide companies with optimization suggestions to help them reduce their emissions and improve their ESG score.

Check out our demo: https://www.youtube.com/watch?v=uiAT6KyjH80

What we achieved during the hackathon

We launched the beta version of our measurement tool, then we studied the GSF SDK and integrated it with our proprietary AI algorithm which allows us to calculate the energy consumption caused by the rendering of the webpages. Thanks to the GSF SDK we managed to add to our service the carbon intensity of each region, and therefore we are able to measure the CO2 caused by each web session on each type of user's device. Moreover, our results are scientifically validated through our partnership with the university “Politecnico di Torino”. 

Our impact

Our solution has a big impact on the ESG data industry and most importantly a big impact for the planet. At the moment no one focus on users’ devices related emissions, everyone focuses the network, but the code’s energy waste related to the rendering of the webpage and that impacts the CPU of billion devices is something that can be reduced.

Next steps

To continue our project we need people, network and money. Our goal is to join the GSF and help them to develop the Green IT industry. We won’t stop with websites, we will also address mobile apps, videogames, blockchain etc. At the moment we are co-piloting with Reply & KPMG and we are in closed beta with 6 big companies. Moreover, we are accelerating our business and product with Berkeley Skydeck, the accelerator of the University of Berkeley, Cariplo Factory and Lendlease.

The value of our solution

Our primary target audience are large companies that do ESG reporting and whose websites are among the 110,000 most trafficked sites in the world (those with more than 8M monthly visits). According to SAP research, 79% of ESG reporting companies are dissatisfied with their environmental impact data because it is not accurate. However, ESG reporting companies need accurate data to be transparent 8 with their customers and investors as well as to be compliant with regulations. What we provide is an accurate measurement tool, 50 times faster and 40 times less expensive, that gives companies the emissions data from their software. It is important to mention that we also offer a free version of our tool, so if you want to test it or give us suggestions for improvements you can visit our website.

https://dedalo.ai