EcoAssist Shows Devs Greener Ways to Code

April 22, 2026

Whether we think about it or not, the environmental impact of the internet continues to climb. Every line of code written for a website leads to power consumption and energy demand. 

While current AI-assisted coding systems prioritize speed, computer science researchers with an interest in sustainability began to wonder how they could highlight more energy-efficient code without sacrificing the benefits of AI-supported coding.

The team created a new digital assistant, EcoAssist, that empowers front-end web developers to see – and reduce – the carbon footprint of their code as they write it. 

The paper, “EcoAssist: Embedding Sustainability into AI-Assisted Frontend Development,” received an Honorable Mention Award at the 2026 ACM Conference on Human Factors in Computing (CHI) this month, an honor reserved for the top 5% of accepted papers. 

EcoAssist lives directly within a web developer’s workspace. Unlike traditional AI coding assistants like GitHub Copilot or Claude Code, which focus primarily on functional correctness and developer speed, EcoAssist analyzes the energy “footprint” of the front-end code in real-time.

The project came about while HCII Assistant Professor Nikolas Martelaro was on a visiting research trip to Portugal last year.

“The EcoAssist project is exciting given my interest in sustainable human-computer interaction, an area I believe is important yet understudied,” said Martelaro. 

Martelaro was on a walk in a garden with ITI student researcher André Barrocas while wondering how AI coding agents might change web development practices in the future. 

“My extensive discussions with ITI Lisbon students brought new perspectives,” Martelaro said. “If we don’t write as much code in the future because AI-supported coding is doing that work, how can we help to make the software more energy-aware?” 

While web sustainability guidelines do exist, many web developers do not know about them.

For example, the World Wide Web Consortium (W3C) published a set of recommendations last year designed to reduce the environmental impact of digital products. AI coding assistants can take digital sustainability recommendations like those and apply them while the developer is working, highlighting more energy-aware code options.

This human-computer interaction (HCI) collaboration came together due to the strong line of environmental and ecologically-minded design work happening at ITI Lisbon.

“AI is consuming a lot of resources. It matters that we understand how to make AI coding a more sustainable practice – our joint research shows how it can happen inside the developer's everyday workflow,” said Valentina Nisi, coauthor and professor at Interactive Technologies Institute (ITI) and Instituto Superior Técnico (IST) - University of Lisbon.

The importance of this work lies in the sheer scale of the internet. With billions of page loads occurring globally every hour, even a small reduction in the percent of energy used to render a site can lead to a massive cumulative reduction in digital emissions.

“Sustainable computing is a research strand I've worked on for almost two decades under the CMU Portugal international partnership,” said Nuno Jardim Nunes, also a coauthor and professor at ITI and IST. “EcoAssist is a fresh collaboration with Martelaro and a forthcoming dual-degree PhD student, extending that line into AI-assisted development.”

The team’s research found that current AI assistants often suggest code that works but is not optimized for energy efficiency. By training EcoAssist to identify energy-intensive patterns and suggest "greener" alternatives, they were able to reduce the per-website energy consumption by an average of 13% to 16%.

EcoAssist did more than just fix code – it changed how developers thought about their work. The tool successfully increased energy awareness without hindering productivity, proving that sustainability doesn't have to come at the cost of performance.

“EcoAssist got developers thinking about the energy impacts that their work could have,” Martelaro said. “Small changes add up over millions of page views, so helping developers and their AI code assistants be more energy-aware will help reduce the impact that the software we make has on the planet.”