General details
EDIHs involved
Challenges
Marine ecosystems are under increasing threat from pollution, overexploitation, and climate change. Many companies aim to reduce their environmental impact, but struggle to quantify their efforts and align them with marine sustainability goals. Traditional sustainability assessments often overlook the marine dimension, lack comparability across firms, and provide limited guidance on which actions deliver the most tangible improvements.
Sea the Change sought to address this gap by developing a solution capable of translating companies’ sustainability reports into actionable insights, enabling companies to understand their marine impact, benchmark their performance, and identify priority areas for action.
Beyond this challenge, the project aimed to guide the SME in adopting artificial intelligence models as part of its digital transformation journey. The challenge was therefore to explore how AI could enhance marine sustainability assessments, and to design a step-by-step pathway for AI adoption in a microenterprise context.
Solutions
Sea the Change designed a machine learning based software to assess companies’ sustainability in a marine context.
The envisioned system integrates:
- Automated data extraction from standardized sustainability reports (PDFs), identifying quantitative indicators and company actions.
- A regression model trained to generate a sustainability score (0–1 scale) for each company, reflecting its environmental alignment.
An interpretability layer that ranks the costumer’s company actions by their relative contribution to the sustainability score, providing clear guidance on which initiatives yield the highest impact.
Resources invested include methodology development, Python software development, dataset construction, manual labelling of company reports, model training and validation. The project requires at least 2,000 labelled samples, 14 quantitative marine indicators and 21 action-related features to build a robust model.
Results and Benefits
The solution is an AI-based software that analyses corporate sustainability reports and produces a marine sustainability score from 0 to 1. Public/private investment is justified because the tool helps companies respond to ESG, CSRD and marine sustainability requirements with a scalable, transparent and data-driven approach.
Key benefits include:
- Digitalization of manual report review, reducing repetitive work and allowing staff to focus on strategic analysis.
- Support for informed decision-making, providing the SME with actionable insights to guide sustainability strategy.
- Empowerment of the human team, as the AI system automates routine tasks but does not replace human judgment.
The deployment is still in development, so the measurable results are currently defined as expected pilot outcomes rather than final achieved impacts. Success will be measured through the number of company reports processed, the accuracy of the model, the reduction in assessment time, and the improvement of company sustainability scores before and after recommended actions.
Expected benefits include faster analysis of sustainability reports, improved comparability between companies and clearer prioritisation of sustainability actions. The model is expected to reach a Mean Absolute Percentage Error below or equal to 20%, which is identified in the project requirements as an acceptable performance target.
The software can unlock opportunities in ESG assessment, CSRD support, blue economy due diligence, portfolio screening and marine sustainability advisory. It is especially relevant for companies in sectors such as shipping, ports, aquaculture, offshore energy, coastal tourism and marine infrastructure.
Perceived social/economic impact
The project strengthens the link between business decision-making and ocean conservation by embedding sustainability intelligence into the corporate ecosystem. The wider impact is the creation of a tool that helps companies take better sustainability decisions in relation to marine ecosystems. Measurable impact can include the number of companies assessed, number of action plans generated, time saved per assessment and improvement in company marine sustainability scores.
Socially, it highlights the interdisciplinary nature of digital sustainability, showing how advanced AI technologies can be applied in SMEs working on ethical and environmental challenges. It raises awareness about the marine dimension of sustainability and empowers organizations to take measurable, informed actions.
Economically, it promotes innovation-driven competitiveness and supports compliance with the Corporate Sustainability Reporting Directive (CSRD). By enabling data-driven marine sustainability assessments, Sea the Change contributes to the broader transition toward a blue economy, where digitalization, environmental protection, and economic growth reinforce one another.
A pilot company could be assessed before and after implementing recommended actions. For example, if its score increases from 0.45 to 0.60, this would represent a 33% improvement in its marine sustainability performance score.
DMA score and results - Stage 0
Sea the Change's score shows that the company has already achieved an average level of digital maturity, performing slightly above SMEs of the Environment sector across Italy and Europe. The company's current investments in digital technologies cover a range of core business operations, relying on a number of mainstream technologies. The adoption of more advanced and disruptive technologies (like AI) is facilitated by the EDIH service, which relied on the support of CINECA. Sea the Change personnel have an average level of digital skills and would need well planned and executed training of personnel. The company's business information is handled in digital form but could also benefit by a comprehensive data strategy, including data security.
Lessons learned
Do’s
- Maintain step-by-step interactions and validation meetings with the SME to correctly design the project and align expectations.
- Ensure continuous collaboration between AI developers and domain experts for a realistic, useful solution.
Don’ts
- Don’t start by developing multiple models in parallel; instead, clarify business needs early and focus on designing one well-defined model architecture.
- Avoid overcomplicating the workflow — prioritize interpretability and usability for non-technical users.
Need support?
Consult our catalogue to locate the Eupopean Digital Innovation Hub nearest to you and accelerate your company's digital transformation.