Skip to main content

      The first wave of mandatory ESG reporting is complete. What have we learned? Regulatory sustainability reporting is difficult! As preparers enter year 2 of reporting, they’re now asking: how can we get value from our data? And where can AI drive operational efficiency, cost savings, and innovation—while also managing risks?

      The answer starts with data quality. AI is only as strong as the foundation it stands on. Fragmented systems and inconsistent data create execution risk. That’s why Transformation is a team sport—and a marathon: success requires collaboration across finance, operations, and technology, and a commitment to long-term change.

      But the narrative doesn’t stop there. To unlock real value, companies must:

      • Balance ambition with pragmatism: AI-driven sustainability solutions promise efficiency and innovation, but they also bring challenges—energy consumption, integration complexity, and leveraging past investments.
      • Adopt Agentic AI with control: Plugging a chatbot into an ESG report won’t cut it. Controlled, purposeful adoption of AI agents embedded into workflows is key to minimizing risk and aligning with business objectives.
      • Build on what exists: Past investments in data platforms and governance are not sunk costs—they’re accelerators for AI-driven transformation.
      Michael Wagemans
      Michael Wagemans

      Partner, Head of Sustainability | Advisory

      KPMG in Belgium

      AI and Data Quality: Garbage In, Garbage Out

      Great AI outputs demand great inputs. Sustainability data is notoriously fragmented—spreadsheets from different plants, utility bills in PDFs, supplier info in portals. If this messy, inconsistent data feeds AI, you’ll get confidently wrong answers. The biggest barrier to using AI in sustainability isn’t fancy algorithms – it’s the data and process readiness. Before creating AI agents, companies must first clean house:

      • Break down data silos and enrich quality: Establish an ESG data hub that consolidates inputs from finance, operations, and supply chain, with versioning and quality checks. Automate data lineage and validation so every emission factor or KPI is traceable to its source.
      • Curate for completeness: Scope 3 data gaps are common. Use AI accelerators to classify supplier spend into emissions categories, parse invoices and logs with OCR/NLP, and normalize multi‑format spreadsheets—turning ESG data from messy to meaningful.
      • Anomaly detection with human‑in‑the‑loop: Deploy rule‑based and machine learning monitors to flag outliers (e.g., sudden energy spikes) and route them to analysts. AI handles volume; people handle exceptions and approvals.

      Transformation Is a Team Sport—and a Marathon

      Achieving mature, AI‑enabled sustainability workflows is multi‑year and cross‑functional. Success patterns include:

      • Strong governance: Establish a design authority spanning Sustainability, Data/AI, Finance/Audit, and Operations to set standards and prioritize use cases.
      • Assurance by design: Log inputs, rules, and outputs to produce audit‑ready evidence trails and reduce year‑end scramble.
      • Blended talent: Pair domain experts with data engineers and product/prompt designers; give them shared Objectives and Key Results tied to operational KPIs like listed below:
        • Data Readiness: Achieve 95% ESG data quality and governance compliance.
        • AI Integration: Deploy at least 3 Agentic AI workflows in core sustainability processes within 6 months.
        • Risk Mitigation: Reduce execution risk by 30% through controlled rollout and readiness assessments.
        • Efficiency Gains: Deliver measurable cost or energy savings of 10–15% in sustainability operations.

      High Value use cases: AI Agents That Scale Expertise

      Internal AI agents blend human judgment with automation, operating under clear guardrails to augment teams:

      • Data Agent: Classifies supplier documents and spend into Scope 3 categories, flags sustainability performance risks, and fills data gaps for reviewer approval.
      • Facilities & Operations Agent: Monitors IoT and utility data in real time; recommends setpoint or scheduling changes with quantified cost and CO₂ savings.
      • Reporting & Evidence Agent: Automates period‑close workpapers and assembles audit‑ready evidence linking sources, calculations, and assumptions.

      For example, a procurement team recently used an AI Vendor Spend Classification model to automate what used to be a tedious manual process. The Strategic Sourcing & Procurement (SSP) team provided a taxonomy of 92 spend categories, which the AI model learned to apply across thousands of vendor transactions. With each purchase automatically categorized to the correct Level-3 spend category, the Corporate Sustainability group could immediately map each category to the appropriate US EPA EEIO (Environmentally Extended Input-Output) emissions factor. The results were then fed into the company’s Carbon Accounting platform to calculate more complete and accurate Scope 3 Category 1 and 2 emissions. This automation dramatically reduced the number of “unclassified” vendor transactions and nearly eliminated manual sorting effort. In turn, data completeness improved, reporting risk fell (fewer gaps and errors), and the team gained confidence that their Scope 3 emissions numbers are auditable and defensible.

      What Good Looks Like in 12 Months

      CapabilityTarget State at 12 Months
      Trusted data foundationSingle source of truth for key ESG metrics with 90%+ automated ingestion and validation. Lineage and data quality scores visible for all reported figures.
      Embedded AI agents3 agents live in operations (e.g., procurement, reporting), each tied to a KPI (e.g., 50% manual effort reduction, 10% utility cost savings).
      Assurance-by-design

      Evidence trails generated automatically; reduced audit findings and no last-minute fire drills.

      Start Now: 6 Actions for Leaders in the Next Quarter

      • Appoint an executive sponsor and a cross‑functional task force (Sustainability, Data/AI, IT, Finance/Risk), engaging with your data office if one has been stood up.
      • Run a rapid data foundation sprint on one critical dataset (e.g., energy or procurement spend) to centralize, clean, and apply lineage/quality rules.
      • Prioritize 3 AI use cases and deliver one pilot with clear success criteria (time saved, accuracy improved, cost reduced).
      • Adopt Green AI guidelines: right size models, estimate compute/CO₂, schedule heavy jobs during greener grid windows.
      • Stand up a curated Sustainability knowledge base and connect it to an internal AI Q&A for consistent answers with citations.
      • Measure and communicate early wins; iterate quickly on what works and retire what doesn’t.

      Belgian perspective

      KPMG in Belgium’s 2026 Chief Sustainability Officer (CSO) Survey finds that a significant majority of sustainability leaders at domestic companies view Artificial Intelligence (AI) as part of the solution, not a threat. Indeed, there is widespread conviction that technological innovations, such as AI, will play an important role in realizing sustainability and business goals. It is not surprising then, that many CSOs and their teams in Belgium are planning and piloting AI, particularly in the context of compliance checks and tasks, as well as operational improvements (e.g. energy efficiency monitoring).

      At the same time, Belgian sustainability leaders share a strong conviction that the use of AI comes with great responsibility. That is, the adoption and deployment of AI means being accountable for its impact. This impact relates to environmental aspects like increased energy and water use, but also touches on social impact, trust, and the responsible use of AI.

      It bears repeating that the real potential and power of AI reside in people using it. Unlocking this value necessitates collaboration between functions (e.g. Sustainability, IT, HR, Legal) within organizations.

      Conclusion

      Companies that treat sustainability as a core data-and-technology opportunity will be better positioned for long-term success. With trustworthy data, efficient models, and workflow-native AI agents in place, AI becomes a practical engine for decarbonization and operational performance—shifting sustainability efforts from a compliance obligation to a source of measurable efficiency and risk reduction.

      This article is based on the original publication by KPMG US and has been localized by KPMG Belgium with additional Belgian perspectives.

      Sustainability

      Building a more sustainable future by helping companies find and fulfil a new more purpose-driven way of doing business.
      Read more
      fields

      Stay informed

      Be the first to know about top business trends that can drive success for your company.

      Subscribe now
      stay informed