<![CDATA[Background: In response to the increasing urgency of the global climate crisis, Indonesian regulations, as outlined in POJK No. 51 of 2017, mandate issuers to enhance transparency through the issuance of sustainability reports. However, these reports are primarily presented in static, non-standardized PDF documents, creating significant barriers for stakeholders seeking comparable industry data. This study develops Green Balance, an artificial intelligence-based platform designed to transform unstructured sustainability data into structured, measurable, and inter-company comparable information. Methods: The study employs the Waterfall Sysytem Development Life Cycle (SDLC) framework, integrating Natural Language Processing (NLP) and Machine Learning technologies, including Extreme Gradient Boosting and Random Forest. Macro-environmental feasibility is assessed using the PESTEL framework, while the Penta Helix model guides the collaborative development strategy. The research is grounded in Stakeholder Theory, emphasizing transparency as a fundamental right of information. Findings: The system successfully generates Green Scope, Green Trend, and Green Index features as objective parameters for comparing Environmental, Social, and Governance performance. In preliminary conceptual validation, the NLP-based extraction pipeline demonstrated a precision rate of approximately 87.3% in identifying ESG-relevant clauses from PDF-based sustainability reports, with an F1-Score of 0.84, benchmarked against manual expert annotation. Data processing time was reduced by an estimated 76% compared to conventional manual extraction methods. These results suggest that digitizing sustainability reports effectively mitigates greenwashing risks and enhances corporate accountability by providing accessible data for ethical investment decision-making. Conclusion: The application of artificial intelligence in sustainable accounting significantly improves information quality and transparency within the Indonesian capital market. Novelty/Originality of this article: This study contributes an original technical model integrating multi-dimensional analysis (PESTEL and Penta Helix) specifically tailored for the Indonesian sustainability reporting ecosystem, a context previously limited in academic research.]]>
