<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Neno Duplan on Medium]]></title>
        <description><![CDATA[Stories by Neno Duplan on Medium]]></description>
        <link>https://medium.com/@nenoduplan?source=rss-57827543a31a------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*W-xsDH_yauMyAWDHTkBtmw.jpeg</url>
            <title>Stories by Neno Duplan on Medium</title>
            <link>https://medium.com/@nenoduplan?source=rss-57827543a31a------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 02:04:45 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@nenoduplan/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[You Cannot Build an EHS AI Compliance Strategy on Data You Don’t Control]]></title>
            <link>https://nenoduplan.medium.com/you-cannot-build-an-ehs-ai-compliance-strategy-on-data-you-dont-control-8356466db818?source=rss-57827543a31a------2</link>
            <guid isPermaLink="false">https://medium.com/p/8356466db818</guid>
            <category><![CDATA[ai-ready-ehs-compliance]]></category>
            <dc:creator><![CDATA[Neno Duplan]]></dc:creator>
            <pubDate>Mon, 16 Nov 2026 03:06:05 GMT</pubDate>
            <atom:updated>2026-11-16T03:06:05.105Z</atom:updated>
            <content:encoded><![CDATA[<h4>The Great AI Leapfrog Illusion: Why AI Cannot Replace Enterprise EHS Compliance Software</h4><p><strong>The dangerous assumption that corporations can bypass decades of software development, data governance, and regulatory expertise</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FmUjb0FrGLc_mP8V9GT5JQ.png" /></figure><h3>Executive Summary</h3><p>A troubling new trend is emerging in corporate environmental, health, and safety (EHS) management. Some organizations are beginning to question whether they need to invest in enterprise EHS software at all. Their reasoning sounds deceptively simple: Why purchase specialized software when artificial intelligence can generate applications, interpret regulations, analyze environmental data, and automate compliance workflows?</p><p>In their view, AI offers an opportunity to leapfrog decades of enterprise software development and move directly into a new era of intelligent automation.</p><p>The idea has an appealing historical precedent.</p><p>In parts of the developing world, countries that never built extensive fixed-line telephone networks moved directly to mobile communications. They avoided the cost of installing millions of miles of copper infrastructure and adopted a newer, more flexible technology.</p><p>Could corporations do something similar with EHS compliance? Could they bypass traditional environmental information management systems and rely instead on AI agents only to perform the same functions?</p><p><strong>My prediction is that this experiment will fail for most large enterprises attempting to replace their EHS systems with AI alone. Not because AI is incapable, but because the proposed shortcut misunderstands the fundamental nature of environmental compliance.</strong></p><p>Mobile networks replaced an older communications infrastructure with a newer one. They did not eliminate the need for communications infrastructure.</p><p>Likewise, AI can transform enterprise EHS software, but it cannot eliminate the need for the <strong>underlying data infrastructure</strong>, validated information, regulatory logic, transactional systems, and governance controls on which compliance depends.</p><p>Indeed, AI makes those foundations more important than ever.</p><p>And here is the irony: Many companies now considering an AI-first approach have never established effective control over their environmental data. Their information remains scattered across consulting firms, local offices, laboratory systems, spreadsheets, PDF reports, and disconnected applications.</p><p>They are attempting to build an intelligent compliance system without first knowing where their data resides, whether it is complete, or whether it can be trusted.</p><p>That is not technological leapfrogging. It is attempting to skip the foundation of a building and start construction with the roof.</p><h3>1. The Mobile Phone Analogy: Why It Breaks Down for EHS</h3><p>The leapfrogging analogy deserves closer examination because it reveals the central misconception behind the AI-only strategy.</p><p>Countries that bypassed fixed-line telecommunications did not eliminate the fundamental requirements of a communications network. They adopted a different architecture that still required towers, spectrum allocation, switching equipment, network protocols, billing systems, security, and ongoing maintenance.</p><p>The technology changed. The underlying engineering requirements did not disappear.</p><p>The same principle applies to enterprise EHS compliance.</p><p>A corporation operating hundreds of industrial facilities across multiple jurisdictions must manage complex, interconnected information and obligations, including:</p><ul><li>Environmental permits, regulatory limits, and reporting deadlines.</li><li>Laboratory analytical results, sampling locations, methods, and detection limits.</li><li>Waste characterization, accumulation, transportation, and disposal.</li><li>Air emissions calculations and equipment inventories.</li><li>Water quality, discharge monitoring, and regulatory reporting.</li><li>Chemical inventories, safety incidents, inspections, and corrective actions.</li><li>Historical monitoring records, quality assurance, and defensible audit trails.</li></ul><p>These are not simply documents that an AI model must read. They represent a continuously evolving system of regulated transactions, scientific measurements, organizational responsibilities, and legally consequential decisions.</p><p>AI can interpret, summarize, classify, predict, and recommend. It can also generate code and execute authorized workflows.</p><p>But it still needs reliable systems in which to store results, enforce constraints, maintain state, manage approvals, and preserve evidence.</p><p>A language model does not become a regulatory system of record merely because it can answer questions about environmental regulations.</p><p>Nor does an AI-generated application become enterprise-grade because it can demonstrate a few compliance workflows.</p><p><strong>The ability to generate software is not equivalent to the ability to operate and maintain a regulated enterprise system.</strong></p><h3>2. The Problem I Identified Nearly Two Decades Ago Has Never Been Fully Solved</h3><p>In a paper published by Locus Technologies in 2008, <em>Environmental Information Management: Ten Things Your Consultant Won’t Tell You</em>, I raised a fundamental question:</p><p>Why do corporations maintain centralized, professionally managed systems for their financial, human resources, and customer data while leaving environmental information fragmented across external consultants?</p><p>The paper identified an industry-wide structural problem.</p><p>Companies might engage five, ten, or twenty consulting firms to manage environmental programs across their facilities. Each consulting firm could maintain its own spreadsheets, databases, naming conventions, and quality control procedures. Different offices within the same consulting organization might follow different practices.</p><p>The corporation technically owned its environmental information but did not necessarily possess effective operational control over it.</p><p>Retrieving information often required contacting a consultant, locating the responsible project manager, searching local files, reconciling inconsistent formats, and paying additional professional fees.</p><p>I argued that this arrangement created unnecessary costs, weakened corporate oversight, and made it difficult to use environmental information strategically.</p><p>The paper also challenged the practice of relying on consulting firms to develop and maintain proprietary environmental software.</p><p>Consultants are essential for environmental science, engineering, remediation, regulatory interpretation, and field operations. But developing secure, scalable, continuously maintained enterprise software is a different discipline.</p><p>A consulting firm primarily earns revenue by providing professional services. A software company must invest continuously in architecture, development, testing, cybersecurity, documentation, support, and product improvement.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1016/1*ruiJMFLKAMDUCny8_yoEgA.png" /><figcaption>A 2008 article published by Locus Technologies warned about fragmented environmental data and the lack of corporate control over critical information. Nearly two decades later, these problems remain largely unresolved — and now pose significant obstacles to successful enterprise-wide AI deployment.</figcaption></figure><p>These business models can coexist, but they are not interchangeable.</p><p>The distinction becomes even more important with AI.</p><p>If a corporation asks its consultants to build custom AI applications around environmental data that remain in consultant-controlled systems, it risks recreating the same dependency I warned about nearly twenty years ago.</p><p>Only now, instead of isolated spreadsheets and databases, the corporation may accumulate isolated AI agents, custom integrations, undocumented prompts, and applications whose long-term maintenance depends on the consulting firm that built them.</p><p><strong>We risk replacing spreadsheet sprawl with AI-agent sprawl without solving the underlying data ownership problem.</strong></p><h3>3. AI Runs on Data. EHS Data Are Not Ready by Default.</h3><p>The most fundamental weakness of an AI-only EHS strategy is assuming useful environmental data are readily available.</p><p>In many corporations, they are not.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GQWeebclOW_S3H55pkKaDQ.png" /></figure><p>Environmental data differ from ordinary business documents in several important ways.</p><p>Consider a single groundwater laboratory result.</p><p>A reported concentration of 5.2 micrograms per liter (5.2 µg/L) has little meaning without additional information:</p><ul><li>Which chemical was measured?</li><li>Which monitoring well and sampling interval produced the sample?</li><li>When was the sample collected and analyzed?</li><li>Which laboratory and analytical method were used?</li><li>What were the reporting and detection limits?</li><li>Was the result qualified, estimated, rejected, or validated?</li><li>Was it a dissolved or total concentration?</li><li>What units, analytical basis, and regulatory thresholds apply?</li><li>Was the sample part of routine monitoring, a duplicate, or a quality control investigation?</li><li>How does it relate to previous measurements at the same location?</li></ul><p>An AI model that encounters the number 5.2 without this context cannot reliably determine its environmental significance.</p><p>Even when supporting metadata exist, they may be distributed among laboratory electronic data deliverables, consultant spreadsheets, scanned reports, and historical databases.</p><p>The model may retrieve an apparently relevant value while missing a qualification that makes the result unsuitable for the intended analysis.</p><p>It may confuse reporting limits with measured concentrations, combine incompatible units, or compare results obtained using different analytical methods without recognizing their limitations.</p><p>These are not minor formatting errors.</p><p>They can change the scientific interpretation of environmental conditions and produce incorrect compliance conclusions.</p><p>Environmental data require more than storage. They require structured relationships, standardized terminology, validation, provenance, and controlled transformations.</p><p>The same requirements apply to air emissions, hazardous waste, refrigerant management, and industrial wastewater compliance.</p><p>AI cannot reliably compensate for missing information that was never collected, inconsistent metadata that was never reconciled, or historical records that are no longer accessible.</p><p>The familiar principle of <em>garbage in, garbage out</em> has acquired a more dangerous dimension.</p><p>Traditional software often exposes bad data through errors, failed calculations, or inconsistent reports.</p><p>Generative AI may produce a polished, convincing explanation from incomplete or incorrect inputs.</p><p><strong>A confident AI answer is not evidence of a correct environmental conclusion.</strong></p><h3>4. Why Decades of Enterprise Software Development Cannot Be Replaced by a Prompt</h3><p>Enterprise EHS software is frequently misunderstood as a collection of forms, screens, reports, and dashboards.</p><p>That misunderstanding makes AI-generated replacements seem more feasible than they actually are.</p><p>The visible user interface is only a small portion of the system.</p><p>Behind it lies a substantial body of engineering and domain-specific functionality developed and refined over many years.</p><p>A production-grade EHS platform must provide several distinct layers of capability.</p><h4>Data architecture</h4><p>The system must support structured entities, relationships, controlled vocabularies, historical records, unit conversions, geographic references, and changing regulatory requirements.</p><h4>Data validation</h4><p>The system must enforce data types, mandatory fields, permissible values, scientific quality control rules, referential integrity, and domain-specific validation.</p><h4>Transaction management</h4><p>Compliance activities must create persistent, traceable records. The system must manage changes, concurrent users, approvals, corrections, and transaction histories.</p><h4>Regulatory workflows</h4><p>Permits, inspections, corrective actions, reporting deadlines, and approvals require deterministic rules and documented responsibilities.</p><h4>Security and governance</h4><p>Large enterprises require identity management, role-based access, tenant isolation, encryption, audit logging, backups, disaster recovery, and controlled third-party access.</p><h4>Integration and interoperability</h4><p>EHS platforms must exchange information with laboratories, enterprise resource planning systems, geographic information systems, sensors, regulators, field devices, and external service providers.</p><h4>Operational continuity</h4><p>The system must remain maintainable as regulations change, personnel leave, facilities are acquired or divested, and new technologies emerge.</p><p>AI can help develop and operate all these capabilities. In some cases, it can dramatically accelerate software engineering.</p><p>But AI does not make the capabilities optional.</p><p>An enterprise attempting to replace commercial EHS software with AI-generated applications must still design, test, secure, maintain, and validate the resulting system.</p><p>It has not eliminated the software investment.</p><p>It has transferred the responsibility for that investment to itself or its consultants.</p><p>And unless the organization has the engineering resources, regulatory expertise, and long-term commitment to maintain the resulting platform, it may end up with a more expensive and less reliable solution.</p><h3>5. Geoffrey Moore’s AI Playbook: The Missing Foundation in EHS</h3><p>In his November 9, 2026 article, <em>For Enterprise CIOs: A Playbook for Navigating the AI Bubble</em>, Geoffrey Moore identifies two major challenges confronting enterprise AI initiatives.</p><p>First, organizations must invest in new approaches to data management and supporting infrastructure.</p><p>Second, too few AI pilots and proofs of concept are progressing into meaningful production deployments.</p><p>These observations are particularly relevant to EHS.</p><p>Moore proposes that organizations identify where AI can create value, establish clear business ownership, define expected returns, measure progress, and solve data-readiness problems tied to specific use cases.</p><p>I agree with the discipline behind this approach.</p><p>However, EHS introduces an important qualification.</p><p><strong>Data readiness can be addressed incrementally, but enterprise data governance cannot remain an afterthought.</strong></p><p>For example, a corporation may initially deploy AI to automate quarterly groundwater reports at ten facilities.</p><p>It would be reasonable to begin by preparing the data needed for those facilities rather than attempting to clean every environmental record in the corporation.</p><p>But the implementation should use an enterprise data model, consistent validation rules, common identifiers, governed access, and an architecture that can accommodate the next hundred facilities.</p><p>Otherwise, the organization simply creates another isolated application.</p><p>Moore’s recommendation to focus on specific use cases should not be interpreted as permission to create separate data architectures for every AI project.</p><p>The distinction is critical.</p><p>A use-case-driven implementation strategy is sound.</p><p>A use-case-specific collection of disconnected data systems is not.</p><p>For regulated environmental programs, the enterprise must establish a common foundation even when individual workflows are modernized incrementally.</p><h3>6. The Economics of Environmental Data: Expensive to Collect, Cheap to Preserve</h3><p>The economics of environmental information make the case for centralized data management even more compelling.</p><p>Across its customers, Locus Technologies manages more than 500 million laboratory analytical data points.</p><p>These records represent an enormous investment.</p><p>Each analytical result is associated with an operational chain that may include sampling program design, field mobilization, sample collection, laboratory analysis, data validation, regulatory review, reporting, and long-term program management.</p><p>Based on Locus’ analysis of customer programs, the average fully burdened cost associated with collecting and managing an analytical data point can approach $197.</p><p>Applying that estimate to 500 million analytical records produces an illustrative cumulative investment of:</p><p><strong>500,000,000 × $197 = $98.5 billion</strong></p><p>This is not the cost of the laboratory test alone. It estimates the broader environmental program expenditures associated with generating and using the data. Actual costs vary significantly by program, location, analyte, and sampling methodology.</p><p>Nevertheless, the scale is revealing.</p><p>Corporations collectively invest enormous sums in collecting environmental information, yet some continue to manage the resulting records through fragmented spreadsheets and consultant-controlled systems.</p><p>Locus’ experience indicates that a structured analytical record can be managed in an enterprise environmental information system for less than two cents per year.</p><blockquote>At that illustrative rate, managing 500 million records would cost less than $10 million annually for the record-management component, excluding other program, implementation, and operational costs. That is what <a href="https://www.locustec.com/a/analytical-lab-data-eim/">Locus EIM</a> does very effectively.</blockquote><p>The contrast is striking.</p><p>Organizations spend substantial sums generating environmental data but often resist investing a tiny fraction of that amount in making the information accessible, reliable, and reusable.</p><p>This is a fundamental misallocation of resources.</p><p>And AI dramatically increases the potential return on correcting it.</p><p>A complete, validated environmental dataset can support historical trend analysis, anomaly detection, sampling optimization, predictive modeling, automated reporting, and identification of unnecessary expenditures.</p><p>A fragmented dataset cannot deliver the same benefits without substantial preparation and reconciliation.</p><p><strong>The real opportunity is not to eliminate environmental software. It is to unlock the value of decades of environmental investments through software-enabled AI.</strong></p><h3>7. Why Historical Data Must Remain Accessible to AI</h3><p>One of the most valuable characteristics of environmental data is its historical continuity.</p><p>Environmental processes often evolve over years or decades.</p><p>Groundwater contamination may migrate slowly. Chemical concentrations may respond to seasonal changes, remediation activities, changes in hydraulic gradients, or shifts in industrial operations.</p><p>A short monitoring record may suggest stability while a longer record reveals a recurring pattern.</p><p>Historical data are therefore not simply old information.</p><p>They are part of the scientific evidence needed to understand environmental behavior. This creates an important distinction between environmental information and many transactional business datasets.</p><p>A financial transaction from twenty years ago may have limited relevance to today’s operational decisions.</p><p>A groundwater measurement from twenty years ago may be essential to understanding whether a contaminant plume is stable, shrinking, or expanding. It may provide a basis for asking regulators <a href="https://www.locustec.com/blog/big-data-reduces-compliance-burden-monitoring-optimization/">to reduce monitoring frequency or deploy predictive analytics agents</a>.</p><p>Organizations that remove historical records from accessible analytical environments can undermine AI&#39;s value.</p><p>This does not mean every historical record must remain in the most expensive operational storage tier. It means historical information should remain complete, discoverable, validated, and readily accessible for authorized analysis.</p><p>Archiving that preserves these properties can be compatible with AI. Archiving that effectively removes records from analysis is not.</p><p>AI models can identify long-term relationships only when the relevant historical evidence is available. For environmental compliance, preserving data continuity is not merely a storage decision.</p><p>It is a scientific and risk-management requirement.</p><h3>8. The Real AI Opportunity: Reduce the Cost of Compliance, Not the Integrity of Compliance</h3><p>The most valuable applications of AI in EHS are not attempts to recreate software functionality that already exists. They are opportunities to improve the decisions and workflows supported by enterprise software.</p><p>Consider long-term environmental monitoring.</p><p>Thousands of contaminated sites remain in monitoring programs that require monthly, quarterly, or annual sampling.</p><p>Many such programs were established years ago, sometimes under very different site conditions.</p><p>At facilities where contaminant sources have been removed, and no new contamination is being introduced, monitoring requirements may deserve periodic scientific reassessment.</p><p>A well-designed <a href="https://www.locustec.com/blog/big-data-reduces-compliance-burden-monitoring-optimization/">AI sampling optimization</a> agent can analyze historical concentrations, spatial relationships, trends, variability, and monitoring objectives to identify locations or analytes that may qualify for reduced monitoring. The objective is not to bypass regulators or compromise environmental protection.</p><p>It is to develop a scientifically defensible basis for determining whether the existing monitoring program remains appropriate.</p><p>AI can also help identify locations where additional monitoring is warranted.</p><p>This is fundamentally different from asking a chatbot to decide whether sampling is necessary.</p><p>The AI agent needs access to validated historical measurements, well construction information, sampling events, hydrogeological context, regulatory requirements, and documented decision criteria.</p><p>It must produce recommendations that qualified professionals can review and, where required, regulators can approve.</p><p>The potential savings can be substantial because sampling programs involve recurring field labor, laboratory analysis, data validation, and reporting expenses.</p><p>Similar opportunities exist in automated compliance reporting.</p><p>An AI reporting agent can review validated data, identify exceptions, assemble narrative explanations, and prepare draft reports.</p><p>But the underlying system must still provide authoritative data, approved calculations, permit limits, version control, and an auditable approval process.</p><p>The proper objective is to automate repetitive work while preserving scientific and regulatory accountability.</p><p>AI should reduce the cost of compliance, not reduce confidence in compliance.</p><h3>9. The Consultant Dependency Problem Could Become Worse Under AI</h3><p>My original paper argued that companies should retain control of their environmental data rather than letting it become fragmented across consulting organizations.</p><p>That warning deserves renewed attention.</p><p>Imagine a multinational industrial company with 300 facilities and eight environmental consulting firms.</p><p>Each consultant maintains some combination of spreadsheets, local databases, laboratory files, and regulatory reports.</p><p>The corporation decides not to purchase an enterprise EHS platform. Instead, it asks its consultants to develop AI agents that can answer environmental questions and prepare compliance reports.</p><p>Initially, the demonstrations may be impressive.</p><p>One agent summarizes groundwater trends. Another drafts discharge monitoring reports. A third extracts permit conditions from PDFs.</p><p>But what happens when the corporation wants a consistent global answer to a question such as:</p><p>Which facilities have recurring exceedances of the same contaminant, how have those concentrations changed over the past decade, and what is our associated regulatory exposure?</p><p>The answer requires more than language generation.</p><p>It requires consistent chemical identifiers, comparable units, validated analytical records, location and facility hierarchies, historical permit limits, regulatory applicability, and reliable relationships among datasets.</p><p>If these elements have never been standardized, the AI agents must reconstruct them.</p><p>If different consultants use incompatible definitions, the agents may return inconsistent results.</p><p>If historical records are missing, the agents cannot recover the underlying measurements through inference.</p><p>And if each consulting firm owns the custom integrations and AI workflows, the corporation may become even more dependent on those firms.</p><p>The company has not achieved independence from software vendors.</p><p>It has exchanged a transparent software subscription for a collection of potentially opaque consulting dependencies.</p><p>This raises an uncomfortable question:</p><p><strong>Why would a corporation entrust the development and long-term operation of its AI compliance infrastructure to organizations whose primary business is selling consulting hours rather than maintaining enterprise software?</strong></p><p>Consultants should play a central role in applying scientific judgment, interpreting complex site conditions, and validating decisions.</p><p>But the corporation should own its data strategy, control access to its records, and ensure that its enterprise systems remain independent of any individual consultant.</p><h3>10. The Architecture That Will Succeed: Clean Data+ Enterprise Platform + AI Agents + Models</h3><p>The future of EHS compliance is not a contest between traditional software and artificial intelligence.</p><p>It is the integration of enterprise software with increasingly capable AI models and specialized agents.</p><p>A sustainable architecture has three complementary layers.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zOyxGo5UyCxnAsq-5ZnxtA.png" /></figure><h4>Layer 1: The enterprise platform</h4><p>The platform is the system of record.</p><p>It maintains validated environmental data, structured relationships, regulatory obligations, transactions, access controls, audit trails, integrations, and historical information.</p><p>It provides the operational foundation the organization can depend on.</p><h4>Layer 2: Specialized AI agents</h4><p>Agents interact with the enterprise platform and execute specific workflows.</p><p>Examples include:</p><ul><li>Sampling optimization agents that identify scientifically defensible monitoring efficiencies.</li><li>Reporting agents that prepare compliance reports from validated data.</li><li>Data validation agents that identify inconsistent laboratory submissions.</li><li>Regulatory agents that help assess changes in applicable requirements.</li><li>Incident classification agents that interpret field observations and images.</li><li>Analytical agents that identify anomalies, trends, and potential risks.</li></ul><p>These agents should operate within defined permissions, use governed data sources, record their actions, and route consequential decisions through appropriate human review.</p><h4>Layer 3: AI models</h4><p>Large language models and specialized analytical models provide reasoning, classification, prediction, summarization, and natural-language interaction.</p><p>Different models may be selected for different tasks.</p><p>Some applications require language reasoning. Others require statistical methods, optimization algorithms, geospatial analysis, or deterministic calculations.</p><p>The enterprise should not confuse a general-purpose language model with the complete analytical machinery required for environmental science.</p><p>The most effective architecture combines these technologies rather than relying on a single model to perform every function.</p><p>The relationships are straightforward:</p><p><strong>The platform provides trusted data and operational controls. Agents execute governed workflows. Models provide intelligence.</strong></p><p>Removing the platform does not simplify this architecture.</p><p>It destabilizes it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hCVwTsCLV43P0L8mwKsRWQ.png" /></figure><h3>10. What Fortune 100 Companies Have Told Us: Five Warning Signs for Enterprise AI</h3><p>Over nearly three decades of working with some of the world’s largest corporations, <a href="https://www.locustec.com/">Locus Technologies</a> has encountered a remarkable range of environmental information management practices.</p><p>Some organizations have made extraordinary progress. Others continue to operate with structures and assumptions that were outdated twenty years ago.</p><p>Several conversations with executives at Fortune 10 and Fortune 100 companies illustrate just how far the industry still has to go before AI can deliver its promised benefits.</p><p>These examples are based on actual discussions with Locus customers and former customers. The organizations are intentionally not identified.</p><h4>Example 1: $19 Million Annually for Laboratory Testing and Monitoring</h4><p>One of Locus’ Fortune 10 customers told us it spends about $19 million annually on laboratory testing and environmental monitoring at contaminated sites.</p><p>That is $19 million every year, primarily to generate scientific measurements needed to understand environmental conditions and fulfill monitoring obligations. Over ten years, assuming constant annual expenditures, that represents $190 million.</p><p>The analytical measurements produced by such a program are not merely an annual compliance expense. They are a continuously growing scientific asset.</p><p>Yet how many corporations treat this information as an investment that should generate returns beyond the immediate regulatory requirement?</p><p>With a centralized, validated historical dataset, AI could potentially identify statistically redundant sampling, evaluate monitoring frequency, detect emerging contamination patterns, and automate portions of the reporting process.</p><p>Even a 10% reduction in recurring program expenditures, where scientifically justified and regulator-approved, would save $1.9 million annually.</p><p>A 20% reduction would represent $3.8 million annually.</p><p>These are illustrative scenarios, not guaranteed savings. But they demonstrate why the economics of environmental AI deserve serious attention.</p><p>The central question is whether the corporation possesses the historical data, scientific context, and enterprise infrastructure necessary to conduct such analyses.</p><p>Without those foundations, AI cannot reliably determine which monitoring activities are necessary and which may have become redundant.</p><p>The company may continue spending millions of dollars annually collecting environmental measurements while failing to capitalize on the enormous value of its accumulated historical data. Centralizing laboratory results from multiple laboratories in a single enterprise system creates another significant financial advantage: the ability to track and compare laboratory performance, analytical costs, turnaround times, and testing volumes across the entire organization. With enterprise-wide visibility, corporations can consolidate purchasing, negotiate volume-based pricing, eliminate redundant testing, and hold laboratories accountable for cost and quality. These economies of scale are difficult, if not impossible, to achieve when laboratory data and purchasing decisions remain fragmented across facilities, business units, and consulting firms. <strong>The same centralized data foundation that enables AI-driven monitoring optimization also provides the purchasing intelligence needed to reduce laboratory expenditures before AI is even deployed.</strong></p><h4>Example 2: More Than 120 EHS Software Systems in One Corporation</h4><p>Another Fortune 100 customer told us that its organization was operating more than 120 different vendor EHS software systems, many with overlapping functionality.</p><p>Think about that for a moment.</p><p>One corporation. More than 120 software systems. Multiple business units, inconsistent data structures, duplicated functionality, different vendors, and potentially incompatible interfaces.</p><p>How does such an organization expect to deploy AI coherently across the enterprise?</p><p>An AI agent seeking a consolidated picture of environmental compliance would need to identify which systems contain authoritative information, reconcile conflicting definitions, navigate access restrictions, and determine whether apparently similar records represent the same activities.</p><p>For example, one system may identify a facility by a numerical code, another by a site name, and a third by a regional designation. Different systems may maintain conflicting inventories, regulatory limits, or compliance statuses.</p><p>AI can help identify and reconcile these differences. But it cannot simply assume that inconsistent records describe the same entity. Nor can it reliably resolve conflicting information without established rules for determining which source is authoritative.</p><p>A corporation with 120 EHS systems does not necessarily need 120 separate AI agents. It needs an enterprise information strategy that rationalizes redundant applications, establishes common data standards, and enables agents to interact with authoritative information.</p><p>Otherwise, AI may simply become a 121st layer of complexity.</p><p><strong>AI should be an opportunity to consolidate an enterprise software portfolio, not another justification for perpetuating its fragmentation.</strong></p><h4>Example 3: More Than 20 Consultants, Constant Turnover, and No Corporate Data Ownership Strategy</h4><p>A major corporation told us that it uses more than twenty environmental consulting firms at any given time and changes them frequently.</p><p>That is not inherently problematic. Large enterprises often require multiple consulting firms with different geographical and technical specialties.</p><p>The troubling part came when we asked about environmental data ownership and management.</p><p>The company indicated that it did not need to maintain its own centralized environmental information because it expected its consultants to manage the data on its behalf.</p><p>This was precisely the problem I described in my 2008 paper, <em>Ten Things Your Consultant Won’t Tell You About Information Management</em>.</p><p>Nearly two decades later, the assumption persists.</p><p>Consider what happens when this corporation decides to deploy AI.</p><p>Which consultant has the complete history of a facility? Which one maintains the laboratory data? What happens to the data when the consultant is replaced? Are sampling locations, analytical methods, quality flags, and regulatory thresholds standardized across all twenty firms?</p><p>Who determines which version of a dataset is authoritative?</p><p>Who is responsible for integrating historical records when a new consultant takes over?</p><p>And most importantly, who is responsible for ensuring an AI agent has access to the complete, validated information needed to reach a defensible conclusion?</p><p>The corporation may own its data contractually, but effective data ownership requires more than a clause in a consulting agreement.</p><p>It requires operational custody, accessibility, portability, governance, and continuity.</p><p>If twenty consulting firms independently manage environmental information, the company has effectively outsourced the foundations of its AI strategy to twenty separate organizations.</p><p>This is not enterprise data management.</p><p>It is institutionalized fragmentation.</p><p>The resulting AI program will be limited by the weakest links in that fragmented information chain.</p><h4>Example 4: A Company That Feared Centralized Data Could Be Subpoenaed</h4><p>Perhaps the most disturbing example came from a major publicly traded corporation that was once a Locus customer.</p><p>When the company decided not to renew its contract, one of its managers explained the reasoning.</p><p>The manager was concerned that consolidating all environmental information in a single system could make it easier for the U.S. Environmental Protection Agency to subpoena the software provider and obtain information about the corporation’s facilities and environmental problems.</p><p>He went further, suggesting that the corporation would be better off destroying its historical EHS data.</p><p>It was a remarkable statement.</p><p>The manager’s tenure at the corporation did not last much longer.</p><p>More importantly, the underlying philosophy illustrates a profound misunderstanding of environmental information governance.</p><p>Centralizing information does not create environmental liabilities. Those liabilities arise from the organization’s actual operations, conditions, and legal obligations.</p><p>And fragmenting or destroying records does not eliminate the underlying environmental problems. Depending on applicable retention obligations, litigation holds, regulatory requirements, and other legal duties, destroying records can itself create serious legal exposure.</p><p>Centralized systems can also make it easier to implement consistent retention schedules, enforce security controls, document access, and respond appropriately to legitimate legal requests.</p><p>The correct response to concerns about disclosure is sound legal governance, not deliberate information destruction.</p><p>From an AI perspective, the consequences of destroying historical records are especially severe.</p><p>An AI model cannot reconstruct decades of actual monitoring measurements that no longer exist. It cannot reliably establish historical contamination trends from missing data. Nor can it determine whether remediation efforts succeeded if the supporting evidence has been discarded.</p><p>Environmental data have a long scientific life.</p><p>Their value often increases as the historical record becomes more complete.</p><p>An organization that deliberately eliminates its historical information is not simply weakening its compliance posture. It is destroying the scientific foundation on which future AI applications would depend.</p><p><strong>You cannot use AI to learn from environmental history that you have chosen to erase.</strong></p><h4>Example 5: The Enterprise That Cannot Communicate Across Business Units</h4><p>Another problem, equally consequential but less visible, is the organizational structure of large corporations.</p><p>Many multinational enterprises operate through highly autonomous business units, regional organizations, acquired subsidiaries, and functional departments.</p><p>Each may have its own environmental personnel, consulting firms, software vendors, budgets, and reporting procedures.</p><p>The result is a collection of environmental information islands.</p><p>One business unit may have invested in a sophisticated data management platform while another relies on spreadsheets. A third may have outsourced its environmental information entirely to consultants.</p><p>Even when the corporation has a central EHS leadership function, that function may have limited authority over software purchasing, data standards, and technology decisions made within individual business units.</p><p>Now consider what happens when the CIO announces an enterprise-wide AI initiative.</p><p>The technology team may be able to license an advanced model and establish a cloud AI environment relatively quickly.</p><p>But who has the authority to require business units to share their data?</p><p>Who can require standardized facility identifiers, analytical naming conventions, or regulatory workflows?</p><p>Who owns the budget for integrating systems that belong to different divisions?</p><p>Who resolves disagreements about which data source should be authoritative?</p><p>AI cannot resolve these organizational questions on its own.</p><p>In fact, AI frequently exposes governance failures that corporations have tolerated for years.</p><p>Implementing enterprise AI is therefore not just a technology project.</p><p>It is an organizational transformation that requires executive authority, shared data governance, clearly assigned accountability, and cross-business-unit cooperation.</p><p>Without those elements, individual departments may deploy successful AI pilots while the corporation fails to achieve meaningful enterprise-scale benefits.</p><h3>11. Organizational Fragmentation: The Hidden Barrier to Enterprise AI</h3><p>Geoffrey Moore correctly emphasizes executive sponsorship, program ownership, measurable business outcomes, and end-user adoption.</p><p>For EHS, I would add another essential requirement: enterprise-wide data and governance authority.</p><p>The problem is not simply that large corporations possess too many disconnected software systems. Their organizations are often structured in ways that perpetuate this condition.</p><p>Business units purchase applications independently. Regional environmental teams develop their own procedures. Consulting firms introduce preferred tools. Individual facilities maintain local databases that satisfy immediate requirements but provide little value to the broader enterprise.</p><p>Over time, the organization accumulates incompatible systems, inconsistent definitions, and overlapping contracts.</p><p>No single executive may have complete visibility into the resulting architecture. AI magnifies the consequences of this structure because the technology derives much of its potential value from identifying relationships across organizational and information boundaries.</p><p>Consider a corporation operating chemical plants, refineries, manufacturing facilities, and distribution centers. One division may possess detailed historical water quality records. Another may maintain waste characterization information. A third may track chemical inventories and emissions.</p><p>AI could potentially identify correlations among chemical usage, waste generation, environmental releases, and regulatory risks across the corporation.</p><p>But only if the data can be connected through consistent identifiers, definitions, access permissions, and governed analytical processes.</p><p>If each business unit operates independently, the analytical opportunity may never materialize.</p><p>The corporation may possess all the necessary information collectively while lacking the organizational ability to use it collectively.</p><p>This is a failure of enterprise architecture and governance, not a limitation of AI.</p><h4>The Need for an Enterprise Environmental Data Authority</h4><p>For large corporations, I would recommend establishing clear corporate responsibility for environmental information architecture.</p><p>This does not necessarily require centralizing every operational decision or moving every data record into a single physical database.</p><p>Modern architectures can support distributed data ownership through controlled APIs, federated access, and shared governance.</p><p>But the enterprise must have common standards and enforceable responsibilities.</p><p>At a minimum, this requires:</p><ol><li><strong>A corporate environmental data owner:</strong> An executive accountable for enterprise-wide data governance, accessibility, quality, and lifecycle management.</li><li><strong>An authoritative inventory of EHS applications:</strong> A documented understanding of which systems exist, what functions they perform, where their data reside, and which applications duplicate one another.</li><li><strong>Common data standards:</strong> Consistent definitions for facilities, locations, substances, analytical results, units, regulatory obligations, and other critical entities.</li><li><strong>Enterprise integration architecture:</strong> Governed interfaces that allow authorized applications and AI agents to access reliable information across organizational boundaries.</li><li><strong>Shared security and audit requirements:</strong> Consistent access controls, provenance, transaction history, and safeguards for sensitive environmental information.</li><li><strong>Joint accountability between IT and EHS:</strong> Environmental professionals must define scientific and regulatory requirements, while IT ensures scalable, maintainable, and secure implementation.</li><li><strong>A measurable application rationalization strategy:</strong> Redundant systems should be evaluated for retirement or consolidation as part of AI modernization rather than preserved indefinitely.</li></ol><p>These are enterprise management responsibilities. They cannot simply be delegated to external consultants or an AI model provider.</p><h4>Why Local AI Success Does Not Guarantee Enterprise Success</h4><p>A business unit may successfully build an AI agent that automates monthly compliance reporting.</p><p>Another may deploy an agent to analyze laboratory results.</p><p>A third may introduce AI-assisted permit management.</p><p>Each initiative may produce measurable local benefits.</p><p>But if the agents rely on incompatible data, inconsistent rules, and separate security structures, they will not necessarily produce enterprise-wide intelligence.</p><p>The corporation may end up with dozens of successful demonstrations and very little genuine transformation.</p><p>This is one reason Geoffrey Moore’s observation about the gap between AI pilots and production deployments deserves special attention.</p><p>The barriers to scale are often not limitations of the models themselves.</p><p>They are inherited weaknesses in enterprise data architecture, organizational structure, and governance.</p><p><strong>Companies cannot expect enterprise-wide AI performance from an organization that has never operated its environmental information as an enterprise asset.</strong></p><h3>12. Do Not Outsource Your AI Strategy to Research Analysts: The Credibility Crisis in Technology Rankings</h3><p>Another group whose advice CIOs and environmental executives should approach with considerable skepticism when planning their AI strategies is technology research analysts.</p><p>For decades, research and advisory firms have positioned themselves as independent authorities on enterprise software. Their reports, rankings, quadrants, waves, and market assessments have influenced billions of dollars in technology purchasing decisions.</p><p>Yet the commercial relationships between research firms and the software vendors they evaluate deserve much more scrutiny than they typically receive.</p><h4>The Problem with the Pay-to-Play Perception</h4><p>The traditional technology analyst business model creates an inherent tension.</p><p>Research organizations often generate revenue by selling subscriptions, advisory services, research access, conferences, and other commercial offerings to the very technology vendors operating in the markets they evaluate.</p><p>At the same time, they market their rankings and assessments as independent guidance for enterprise buyers.</p><p>This does not establish that vendors can purchase favorable rankings. But it raises a legitimate question about incentives, access, transparency, and the appearance of independence.</p><p>How can a CIO determine whether a vendor’s favorable position reflects superior technology, better execution, stronger customer outcomes, more effective analyst engagement, or some combination of these factors?</p><p>The answer should never be to accept a quadrant position as sufficient evidence.</p><p>In my experience, the EHS software industry is particularly vulnerable to this problem.</p><p>Large software vendors and private-equity-backed consolidators often have substantial marketing resources, dedicated analyst relations teams, and the budgets necessary to maintain continuous engagement with research organizations.</p><p>Smaller, technically sophisticated software developers may invest more heavily in engineering, scientific functionality, and customer support than in analyst relations.</p><p>Yet they can be less visible in market assessments that enterprise procurement teams rely upon.</p><p><strong>A company’s position in a research quadrant should never be confused with a scientifically validated measurement of its software’s technical superiority.</strong></p><h4>AI Is Challenging the Traditional Research Analyst Business Model</h4><p>Artificial intelligence is now challenging research firms themselves.</p><p>Historically, a significant portion of technology research firms&#39; value proposition was their ability to gather information, organize markets, summarize vendor capabilities, interpret trends, and produce comparative assessments.</p><p>AI can increasingly assist with many of these activities.</p><p>Enterprise buyers can use AI-supported research to examine technical documentation, compare product architectures, analyze published capabilities, evaluate security certifications, and prepare detailed requirements for direct vendor demonstrations.</p><p>This does not eliminate the value of experienced human analysts. Independent testing, firsthand implementation experience, proprietary market data, and informed judgment remain valuable.</p><p>But it does challenge the economics of selling expensive reports that primarily repackage vendor-supplied information into proprietary rankings.</p><p>The research industry must now demonstrate that its analysis provides value beyond what customers can obtain through direct evidence, publicly available technical information, and AI-assisted evaluation.</p><p>Ironically, some of the organizations advising enterprises on AI disruption face disruption in their own business models.</p><p>The appropriate response is greater analytical rigor and transparency, not louder predictions about AI&#39;s limitations.</p><h4>The Danger of Following Yesterday’s Leaders Into Tomorrow’s Architecture</h4><p>AI is fundamentally changing how enterprise software is built and used.</p><p>Traditional analyst evaluations may place considerable weight on company size, market presence, sales execution, breadth of product portfolios, and historical vendor positioning.</p><p>Those factors may matter when evaluating a supplier’s financial stability and ability to support large deployments.</p><p>But they do not necessarily establish whether a vendor has the best architecture for AI-enabled EHS compliance.</p><p>For example, the most important technical questions should include:</p><ul><li>Does the vendor maintain a unified, validated environmental data foundation?</li><li>Can AI agents operate on structured information rather than relying primarily on unverified documents?</li><li>Are regulatory calculations deterministic, tested, and auditable?</li><li>Can the platform preserve and analyze decades of historical environmental measurements?</li><li>Does the system maintain provenance, permissions, and human approval controls?</li><li>Can the platform integrate data across business units, consultants, and facilities?</li><li>Can the vendor demonstrate measurable reductions in compliance costs without compromising scientific integrity?</li></ul><p>These capabilities cannot be reliably determined from a two-dimensional market-positioning graphic.</p><p>They require technical evaluation, direct demonstrations, representative customer data, independent validation, and evidence from actual production deployments.</p><p>An analyst’s favorable assessment may help a CIO construct a shortlist. It should not substitute for technical due diligence.</p><h4>A Warning for CIOs: Do Not Replace Vendor Hype with Analyst Hype</h4><p>Some research analysts have been vocal critics of enterprise AI, emphasizing uncertain returns, unreliable models, and failed proofs of concept.</p><p>These concerns deserve serious consideration.</p><p>But the advice must be evaluated on its technical merits, regardless of the analyst’s reputation or the research firm&#39;s commercial standing.</p><p>A credible analyst should distinguish between AI used as an ungoverned replacement for enterprise software and AI integrated into mature, validated enterprise systems.</p><p>These are fundamentally different architectures with different risk profiles and economic outcomes.</p><p>Blanket skepticism about AI can be just as misleading as exaggerated claims that AI will replace every enterprise application.</p><p>The question should not be whether AI is good or bad for EHS software.</p><p>The question is which architectures, data foundations, controls, and operating models allow AI to produce reliable and measurable outcomes.</p><h4>A Better Way to Evaluate AI and EHS Software Vendors</h4><p>CIOs should demand evidence that can be independently verified.</p><p>Ask vendors to demonstrate their software using actual enterprise scenarios, not carefully scripted sales presentations.</p><p>Require data validation tests, documented integrations, performance benchmarks, security evidence, regulatory workflow demonstrations, and customer references from comparable production environments.</p><p>Measure implementation time, total cost of ownership, historical data accessibility, reporting accuracy, and the ability to reduce recurring compliance expenditures.</p><p>Ask how an AI agent reaches its conclusions.</p><h3>13. Implications for the CIO: Five Questions Before Launching an EHS AI Strategy</h3><p>The experiences described above suggest that every CIO and chief environmental officer should ask five questions before committing substantial resources to an enterprise AI initiative:</p><p><strong>1. Where are our environmental data?</strong> Can we identify and access complete records across every business unit, facility, and external consultant?</p><p><strong>2. How many EHS systems are we operating?</strong> Do we know which applications overlap, which contain authoritative records, and which can be consolidated?</p><p><strong>3. Can we trust the data?</strong> Are analytical measurements, permit conditions, regulatory obligations, and historical records standardized, validated, and traceable?</p><p><strong>4. Who has authority across organizational boundaries?</strong> Can corporate leadership enforce common data standards and integration requirements across autonomous divisions?</p><p><strong>5. What infrastructure will our AI agents depend on?</strong> Are we building on a professionally maintained enterprise platform, or creating another collection of disconnected applications and consulting dependencies?</p><p>If corporate leadership cannot answer these questions, the immediate priority should not be acquiring more AI tools.</p><p>It should be establishing the information and governance foundation that will allow those tools to deliver meaningful value.</p><p>The ultimate conclusion is unavoidable:</p><p><strong>The greatest obstacle to enterprise AI in EHS may not be technology at all. It may be decades of poor environmental information management, fragmented corporate structures, and misplaced confidence that someone else is taking care of the data.</strong></p><h3>14. A Technical Comparison: AI-Only Versus Platform-Centered AI</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3fKAEeJOYXZUd0u4JRuecg.png" /><figcaption>AI can accelerate environmental compliance, but it cannot eliminate the need for enterprise data governance, regulatory controls, and auditable systems of record.</figcaption></figure><p>The comparison does not suggest that commercial software is automatically secure, compliant, or superior. Those capabilities must be demonstrated and independently evaluated.</p><p>Nor does it mean that an organization cannot develop its own enterprise platform. A sufficiently capable company can build and maintain the necessary infrastructure internally. But once it does, it has effectively become an enterprise software developer. In my 30 years of experience, most internal development efforts fail and go significantly over budget.</p><p>The distinction is not between buying software and building software. It is between recognizing the engineering requirements and pretending they no longer exist.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6vVGK5OMT1pjwENibhFVDw.png" /></figure><h3>15. A CIO Playbook for AI-Enabled EHS Modernization</h3><p>Building on Geoffrey Moore’s recommendations, I would propose seven priorities for CIOs and environmental executives.</p><ol><li>Establish ownership and control of environmental data. Inventory where environmental information resides, identify external custodians, and ensure that the corporation has contractual and technical access to complete records and metadata.</li><li>Establish an enterprise data foundation. Standardize critical entities, identifiers, measurement units, validation rules, and data governance requirements. Consolidation may be physical or federated, as long as the enterprise maintains consistent control and authoritative definitions.</li><li>Identify high-value AI use cases. Prioritize workflows with measurable opportunities to reduce latency, improve data quality, lower recurring costs, or reduce regulatory risk.</li><li>Deploy AI against trusted systems of record. Require agents to use approved data sources and controlled interfaces rather than relying on unverified document retrieval or ad hoc spreadsheet analysis.</li><li>Preserve human accountability. Maintain clear responsibility for regulatory interpretations, scientific conclusions, permit submissions, and material compliance decisions.</li><li>Measure return on investment. Establish baseline costs, processing times, error rates, and compliance outcomes before deployment. Measure actual improvements rather than relying on demonstrations or projected productivity gains.</li><li>Design for long-term independence. Ensure that the corporation can change consultants, AI model providers, and implementation partners without losing its environmental data, business rules, or operational history.</li></ol><p>This last principle is particularly important. The organization should avoid replacing dependency on a software vendor with even greater dependency on a consulting firm or a proprietary AI implementation.</p><h3>16. The Next Competitive Divide Will Be Between Companies That Control Their Data and Those That Do Not</h3><p>The emerging divide in environmental compliance will not be between companies that use AI and companies that do not.</p><p>AI capabilities will become increasingly accessible.</p><p>The meaningful distinction will be between organizations that possess trustworthy, well-organized, historically complete environmental information and those that must reconstruct their data every time they want to use it.</p><p>Companies that have invested in enterprise environmental information management will be able to deploy increasingly sophisticated AI agents against established data foundations.</p><p>They can automate reporting, optimize monitoring, identify anomalies, improve compliance oversight, and evaluate environmental risks across their portfolios.</p><p>Companies that have neglected data management will face a more difficult path.</p><p>Before realizing AI’s potential, they will need to locate their data, recover it from consultants, reconcile incompatible formats, establish validation procedures, reconstruct missing metadata, and develop governance mechanisms.</p><p>Companies that thought they were saving money by avoiding enterprise software may discover they must pay for the missing infrastructure after all. Worse, they may have to pay their consultants to reconstruct the same information they originally paid those consultants to collect.</p><p>This is the hidden cost of decades of fragmented environmental data management.</p><h3>Conclusion: You Cannot Leapfrog the Foundation</h3><p>Nearly twenty years ago, I argued that corporations needed to own their environmental information and manage it with the same discipline they applied to financial and operational data.</p><p>That argument was about cost, efficiency, transparency, and environmental accountability.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IU99FCjtsvnDeSHU3wCB-Q.png" /></figure><p>Today, it is also about artificial intelligence.</p><p>AI has extraordinary potential to transform EHS compliance. It can help organizations automate repetitive work, improve scientific analysis, optimize environmental monitoring, and reduce the cost of maintaining regulatory compliance.</p><p>But these opportunities depend on the quality, completeness, accessibility, and governance of the underlying information.</p><p>The mobile telecommunications revolution succeeded because countries replaced an obsolete infrastructure with a more advanced one. They did not replace infrastructure with wishful thinking.</p><p>Corporations attempting to bypass enterprise EHS software through AI are making precisely that mistake.</p><p>They are confusing intelligence with infrastructure, language generation with data management, and rapid prototyping with sustainable enterprise operations.</p><p>The lesson for corporations with EHS compliance requirements and their CIOs is simple:</p><p><strong>AI can help leapfrog outdated workflows. It cannot leapfrog the need for trusted data, scientific integrity, and enterprise governance.</strong></p><p>The winners in the next generation of environmental compliance will not be those who eliminate software.</p><p>They will be those who combine decades of validated environmental information, professionally engineered enterprise platforms, and AI agents into a unified operating model.</p><p><strong>The future of EHS is not AI instead of software. It is AI powered by the right software and, above all, by the right data.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8356466db818" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[2035: The Day the River Started Talking]]></title>
            <link>https://nenoduplan.medium.com/2035-the-day-the-river-started-talking-478dc6d29963?source=rss-57827543a31a------2</link>
            <guid isPermaLink="false">https://medium.com/p/478dc6d29963</guid>
            <category><![CDATA[ehs-software]]></category>
            <category><![CDATA[ghg-emissions]]></category>
            <category><![CDATA[water]]></category>
            <category><![CDATA[esg]]></category>
            <category><![CDATA[compliance]]></category>
            <dc:creator><![CDATA[Neno Duplan]]></dc:creator>
            <pubDate>Thu, 19 Feb 2026 15:01:03 GMT</pubDate>
            <atom:updated>2026-07-11T21:49:42.970Z</atom:updated>
            <cc:license>http://creativecommons.org/licenses/by/4.0/</cc:license>
            <content:encoded><![CDATA[<p>How Real-Time Environmental Intelligence Will Redefine <a href="https://www.locustec.com/why-most-ehs-esg-software-cant-deliver-real-ai/">EHS Compliance and ESG Accountability</a></p><p>Thanks to advances in technology, we are learning to <a href="https://www.projectceti.org/">decode whale song</a>. We are beginning to <a href="https://birdnet.cornell.edu/">recognize bird calls</a>. And we are starting to <a href="https://blog.google/innovation-and-ai/products/dolphingemma/">understand dolphin whistles</a>. We are on the verge of <a href="https://www.earth.com/news/ai-helps-humans-have-20-minute-conversation-with-humpback-whale-named-twain/">the first true human-nonhuman conversation</a> in history.</p><p>For the first time ever, advances in sensing, data infrastructure, and AI make it possible to <em>listen</em> and <em>respond</em> to the Earth’s voices and signals.</p><p>The next frontier is decoding the <a href="https://www.locustec.com/">entire interconnected ecosystem</a>. A river expresses stress long before it turns brown. A watershed signals instability long before it floods.</p><p>Today, we can stream the life of that watershed — flow rates, conductivity, turbidity, temperature, pH — into a continuous system.</p><p>Yet as of 2026, most organizations still trap that stream inside forms, spreadsheets, compliance reports, and static dashboards.</p><p>We are listening.</p><p>But we are not yet conversing.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CGGwGkAjZuwod1sgwBw2-A.png" /><figcaption>Compliance turns into signal monitoring</figcaption></figure><p>I’m not going to make any product predictions in this essay. Instead, I’m going to share my worldview forecast. A look at what EHS compliance will look like by 2035 once the web undergoes its next great shift, and which old habits will not survive it.</p><p>The most important shift that will enable us to <em>converse</em> with our living environment, from the river to the reef, is architectural, not regulatory.</p><p>The old web was about publishing: a place for statements, documents, and static claims. The new web is much more alive and event-driven. Signals flow through AI-native systems that exchange verifiable facts in real time. A single-source-of-truth becomes a continuous stream of truth. Once you accept that, the entire logic of today’s “reporting” workflows feels archaic. Filing a quarterly compliance report is like mailing your heartbeat to a hospital every three months, hoping no one needs it sooner.</p><p>In the world we live in now, environmental truth usually arrives late. After the samples travel from the field to the lab. After spreadsheets get reconciled. After debates about whether an outlier “counts.” After the calendar says it’s time to submit. By the time the story becomes official, the river has already carried the truth downstream.</p><p>But imagine a different world.</p><p>A world where water quality isn’t something we <em>report</em>, but something we <em>maintain</em>. Like a patient on life support. Continuous telemetry. Always on. Forever.</p><p>By 2035, don’t expect major upgrades to today’s EHS compliance or ESG software. Expect compliance to operate on an entirely new architecture and web.</p><h3><strong>Reporting has been a workaround until now</strong></h3><p>We still run EHS and ESG like paperwork dressed up as software. We buy platforms, bolt on features, and believe the future will be “the same process, but faster.” We debate definitions, chase assurance, and polish disclosures for regulators, investors, and boards.</p><p>Reporting, as designed for the old web, has been a workaround.</p><p>The<a href="https://www.locustec.com/what-software-can-replace-outdated-fragmented-ehs-systems/"> compliance stack</a> was built for a world where most measurements were manual, sensors were rare, identities were inconsistent, and audit trails were thin. <a href="https://www.locustec.com/read-ehs-market-reports-without-being-misled/">Analyst teams </a>had to assemble evidence, reconcile versions, and ship reports on a schedule. Reporting made sense because the underlying system (i.e., reality) could not speak for itself. <a href="https://www.locustec.com/press-release/locusai-report-agent-for-eim/">But with advancements in automation and AI, that is quickly changing.</a></p><p>AI (models and LLMs) is only part of the shift. The underlying data and architecture to access it represent the biggest opportunity for the future we envision. AI performs poorly on PDFs, exports, and disconnected systems with mismatched definitions and timelines. AI performs well when data is standardized, continuously updated, tied to stable identifiers, and recorded with integrity.</p><p>The next decade will not be won by a<a href="https://www.locustec.com/locus-platform-unified-foundation-ehs-compliance-sustainability/"> smarter dashboard</a>. The next decade will be won by an improved infrastructure underneath. A web-native, multitenant database that makes compliance data continuous, auditable, and ready for automation.</p><h3><strong>Multitenant is not a pricing model</strong></h3><p>Multitenant architecture reflects a shift in worldview. Cost efficiency may follow, but the real change is philosophical.</p><p>Multitenant systems prevent each company from inventing its own version of an “incident,” a “permit limit,” an “exceedance,” “scope boundary,” or “effluent.” Multitenancy enables shared semantics; a discharge limit means the same thing upstream and downstream, across facilities, regulators, and ecosystems.</p><p>Therefore, a strong foundation of standardized inputs, along with a multitenant platform and tooling, enables AI to reach its full potential. Companies that focus on making the world more legible for AI will have the biggest opportunities to effect positive change for the environment and shareholders.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UwX3L9BHV4MtEfs6O7KUvw.png" /><figcaption>EHS Compliance Opportunities using a unified multitenant SaaS platform</figcaption></figure><h3><strong>By 2035, compliance is infrastructure</strong></h3><p>The most uncomfortable prediction is also the most likely:</p><p>By 2035, EHS compliance and ESG will not be about the software you buy. They will be about the infrastructure you connect to.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bQrjlnata3FA_We5f6JI8g.png" /><figcaption>Yesterday’s compliance will turn into tomorrow’s competitive advantage. No more PDFs report submitted into a black hole. Compliance in 2035 will be an always-on signal.</figcaption></figure><p>There will be no bespoke platform per company. No modular universe of air, water, carbon, and supplier risk stitched together behind the firewall. Those ideas belong to an era when every organization pretended it ran its own planet.</p><p>In 2035, environmental truth is no longer a company-level possession.</p><p>Organizations will connect to a globally integrated, permissioned data fabric, harmonized at the planetary level, that links emissions sources, permits, thresholds, and ecosystems into a single living system. Not a single government database like the EPA IBM mainframe from the seventies. Not a vendor monopoly. Something closer to DNS or GPS: shared infrastructure that nobody owns, the way you own a product, but everybody depends on it, the way you depend on roads or the internet.</p><p>And because trust alone cannot sustain shared truth, the fabric will carry its own memory of integrity. Call it blockchain. Call it a cryptographic audit. Call it tamper-evident provenance. The name matters less than the behavior: data carries its chain of custody the way a river carries sediment or contaminants, where it came from, what touched it, and whether it was altered.</p><p>Verification stops being an annual ritual and becomes a default condition.</p><p>In that world, regulators don’t need your report.</p><p>They need your signal.</p><p>Compliance stops being a calendar event and becomes a continuous state — an always-on relationship between limits and reality. When nothing exceeds, nothing happens. When something drifts, the system warns. When something crosses the line, the system acts.</p><p>The report is the stream.<br>The audit is the provenance.<br>The submission is meaningless.</p><h3><strong>Water arrives first</strong></h3><p>Water is where this future lands first because water doesn’t tolerate fiction. Unlike air, water keeps receipts. It’s the universal solvent: it touches everything, cleans everything, and carries away whatever we fail to control. And once something is in the water, “removal” isn’t a metaphor. It’s a cost. It demands infrastructure and energy — energy we don’t have to waste. When that energy isn’t renewable, it becomes carbon.</p><p>Air emissions are one-way. If you don’t capture them before release, you don’t get a second chance. They disperse, dilute, and do their work somewhere else — immediately or later, invisibly and cumulatively. Water is different. Contamination moves through pathways. It leaves a trail. Most of the time, you can trace it back, find it in transit, and intercept it — contain, treat, or remove it before it spreads the damage.</p><p>Carbon is planetary and abstract — so abstract we’ve built an entire narrative industry around it. Almost like a religion. Water is local, embodied, finite. It runs through pipes, fields, factories, and cities. It shows up in the lungs and kidneys need it. When water goes wrong, you don’t get to debate it. It is not political. Not Republican. Not Democrat.</p><p>That’s why the water system of 2035 will look like healthcare, not in the bureaucracy sense, but in the life-support sense. Watersheds will be monitored like ICU patients: continuous telemetry, redundant sensors, automatic validation, anomaly detection, and provenance embedded in every measurement. No waiting for someone to notice. No waiting for someone to file. No waiting for someone to confess.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kDipJ9SEyDxSMp486l5QiA.png" /><figcaption>Prediction on water monitoring and compliance in 2035</figcaption></figure><p>And because the web finally does what it was always meant to do — move reliable signals between systems — the response won’t be a memo. It will be control: throttling discharges, rerouting intakes, dispatching crews, alerting downstream operators before the plume arrives.</p><p>The goal won’t be “document water quality.”<br>It will be “stabilize water quality.”</p><p>That is the moral pivot of the next era: from storytelling to stewardship.</p><h3><strong>The new compliance workforce</strong></h3><p>In 2035, compliance professionals will no longer write narratives. They will be supervising agents.</p><p>AI agents will reconcile sensor data with production volumes. They will detect drift, tampering, and suspicious gaps. They will forecast exceedances before they occur, simulate operational changes, and propose corrective actions. Human judgment remains — but above the mechanics: approving actions, resolving edge cases, debating policy where values matter.</p><p>The vendors who thrive will not be the ones who make reporting prettier. They will be the ones who make truth more continuous, more verifiable, and more actionable. They will operate standards, identity, provenance, and optimization engines that keep systems within limits with minimal waste.</p><p>The rest will sell artifacts to a world that no longer needs artifacts.</p><h3><strong>Infrastructure outlives politics</strong></h3><p>Shared systems become civilization-critical. Civilization-critical systems become resilient. Resilient systems replicate.</p><p>A <a href="https://www.locustec.com/artificial-intelligence-ai-and-environmental-compliance-revisited-pt-4/">planetary environmental ledger</a>, if it truly becomes the nervous system of compliance, water, and survival, will become part of humanity’s long memory. If future generations leave Earth, they won’t just carry culture and science. They’ll carry the record of what happens when truth is treated like paperwork.</p><p>A Noah’s Ark of data. Not as punishment. As prevention.</p><p>We are not there yet. We are still in adolescence: definitions, boundaries, assurance debates, spreadsheets dressed up as platforms, reporting cycles that pretend reality respects deadlines.</p><p>But the arc is clear.</p><p>The web is becoming a nervous system.<br> Multitenant, web-native databases are becoming the shared language.<br> Tamper-evident provenance is becoming the default trust.<br> AI agents are becoming operators.</p><p>And reporting — as ritual, as narrative, as a substitute for continuous truth — will soon feel as outdated as mailing your heartbeat.</p><p>In 2035, the planet won’t wait for your report.</p><p>It will read your telemetry.</p><p>And when the river speaks, it won’t file a report.</p><p>It will speak in data and signals. The question is whether we will have built the systems capable of understanding and responding to it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mJt6cBglzRL8oFwMvI_opw.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=478dc6d29963" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How Roll-ups Continue to Destroy EHS Compliance, Sustainability, and Water Software Companies]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://nenoduplan.medium.com/how-roll-ups-continue-to-destroy-ehs-compliance-sustainability-and-water-software-companies-c7102424a8fa?source=rss-57827543a31a------2"><img src="https://cdn-images-1.medium.com/max/800/1*r2yTxAHQocJSFlKPWUbWMw.jpeg" width="800"></a></p><p class="medium-feed-snippet">Dante&#x2019;s EHS Inferno: Surviving the Fiery Aftermath of Private Equity Roll-ups.</p><p class="medium-feed-link"><a href="https://nenoduplan.medium.com/how-roll-ups-continue-to-destroy-ehs-compliance-sustainability-and-water-software-companies-c7102424a8fa?source=rss-57827543a31a------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://nenoduplan.medium.com/how-roll-ups-continue-to-destroy-ehs-compliance-sustainability-and-water-software-companies-c7102424a8fa?source=rss-57827543a31a------2</link>
            <guid isPermaLink="false">https://medium.com/p/c7102424a8fa</guid>
            <category><![CDATA[ehs-software]]></category>
            <category><![CDATA[rollup]]></category>
            <category><![CDATA[climate-change]]></category>
            <category><![CDATA[esg-investing]]></category>
            <category><![CDATA[sustainability]]></category>
            <dc:creator><![CDATA[Neno Duplan]]></dc:creator>
            <pubDate>Mon, 31 Mar 2025 06:37:55 GMT</pubDate>
            <atom:updated>2025-04-01T16:15:37.132Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[From Fairchild Semiconductor to Locus Technologies]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://nenoduplan.medium.com/from-fairchild-semiconductor-to-locus-technologies-c68376ab0f5f?source=rss-57827543a31a------2"><img src="https://cdn-images-1.medium.com/max/1456/0*tDjK82r6aPxhTb4f.png" width="1456"></a></p><p class="medium-feed-snippet">This story is about lineage, particularly the linkage between Fairchild (Fairchild) Semiconductor, its successors and offshoots, and Locus&#x2026;</p><p class="medium-feed-link"><a href="https://nenoduplan.medium.com/from-fairchild-semiconductor-to-locus-technologies-c68376ab0f5f?source=rss-57827543a31a------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://nenoduplan.medium.com/from-fairchild-semiconductor-to-locus-technologies-c68376ab0f5f?source=rss-57827543a31a------2</link>
            <guid isPermaLink="false">https://medium.com/p/c68376ab0f5f</guid>
            <category><![CDATA[environment]]></category>
            <category><![CDATA[silicon-valley]]></category>
            <category><![CDATA[bootstrapping]]></category>
            <category><![CDATA[ehs-software]]></category>
            <category><![CDATA[esg-software]]></category>
            <dc:creator><![CDATA[Neno Duplan]]></dc:creator>
            <pubDate>Sun, 20 Aug 2023 21:24:00 GMT</pubDate>
            <atom:updated>2023-08-20T21:24:00.976Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Beware of Shadow ESG Software]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://nenoduplan.medium.com/beware-of-shadow-esg-software-44358882384d?source=rss-57827543a31a------2"><img src="https://cdn-images-1.medium.com/max/1782/1*hcHf6y7Nqqkr5pO2dngcMg.png" width="1782"></a></p><p class="medium-feed-snippet">Not this type of circular economy</p><p class="medium-feed-link"><a href="https://nenoduplan.medium.com/beware-of-shadow-esg-software-44358882384d?source=rss-57827543a31a------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://nenoduplan.medium.com/beware-of-shadow-esg-software-44358882384d?source=rss-57827543a31a------2</link>
            <guid isPermaLink="false">https://medium.com/p/44358882384d</guid>
            <category><![CDATA[esg-software]]></category>
            <category><![CDATA[greenwashing]]></category>
            <category><![CDATA[sustainability]]></category>
            <category><![CDATA[esg-reporting]]></category>
            <category><![CDATA[circulareconomy]]></category>
            <dc:creator><![CDATA[Neno Duplan]]></dc:creator>
            <pubDate>Fri, 19 May 2023 01:28:09 GMT</pubDate>
            <atom:updated>2023-05-19T01:28:09.434Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Ancient Roman Solution for Water Contamination: Pecunia non olet]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://nenoduplan.medium.com/ancient-roman-solution-for-water-contamination-pecunia-non-olet-43f1146bba1d?source=rss-57827543a31a------2"><img src="https://cdn-images-1.medium.com/max/1089/0*ZgoEBqkhwyVaLxXo" width="1089"></a></p><p class="medium-feed-snippet">After news that Roman concrete outperforms today&#x2019;s modern concrete and generates fewer carbon emissions, ancient Rome&#x2019;s solution to reduce&#x2026;</p><p class="medium-feed-link"><a href="https://nenoduplan.medium.com/ancient-roman-solution-for-water-contamination-pecunia-non-olet-43f1146bba1d?source=rss-57827543a31a------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://nenoduplan.medium.com/ancient-roman-solution-for-water-contamination-pecunia-non-olet-43f1146bba1d?source=rss-57827543a31a------2</link>
            <guid isPermaLink="false">https://medium.com/p/43f1146bba1d</guid>
            <category><![CDATA[energy]]></category>
            <category><![CDATA[climate-change]]></category>
            <category><![CDATA[water]]></category>
            <category><![CDATA[water-contamination]]></category>
            <category><![CDATA[sustainability]]></category>
            <dc:creator><![CDATA[Neno Duplan]]></dc:creator>
            <pubDate>Mon, 23 Jan 2023 02:57:55 GMT</pubDate>
            <atom:updated>2023-01-23T02:57:55.088Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Who Controls Your Emissions Data]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://nenoduplan.medium.com/who-controls-your-emissions-data-903e273143c?source=rss-57827543a31a------2"><img src="https://cdn-images-1.medium.com/max/1024/0*88aYfZDGfNn66QWX" width="1024"></a></p><p class="medium-feed-snippet">Beati Possidentes</p><p class="medium-feed-link"><a href="https://nenoduplan.medium.com/who-controls-your-emissions-data-903e273143c?source=rss-57827543a31a------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://nenoduplan.medium.com/who-controls-your-emissions-data-903e273143c?source=rss-57827543a31a------2</link>
            <guid isPermaLink="false">https://medium.com/p/903e273143c</guid>
            <category><![CDATA[esg]]></category>
            <category><![CDATA[emissions]]></category>
            <category><![CDATA[water]]></category>
            <category><![CDATA[ehs-software]]></category>
            <category><![CDATA[ghg-emissions]]></category>
            <dc:creator><![CDATA[Neno Duplan]]></dc:creator>
            <pubDate>Fri, 16 Dec 2022 22:08:02 GMT</pubDate>
            <atom:updated>2022-12-16T22:08:02.574Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Two Tiny Molecules Will Determine the Future of Mankind]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://nenoduplan.medium.com/two-tiny-molecules-will-determine-the-future-of-mankind-521e906f7567?source=rss-57827543a31a------2"><img src="https://cdn-images-1.medium.com/max/1100/0*8OBevJAy-KEssmAD" width="1100"></a></p><p class="medium-feed-snippet">Water and carbon dioxide are some of the smallest molecules.</p><p class="medium-feed-link"><a href="https://nenoduplan.medium.com/two-tiny-molecules-will-determine-the-future-of-mankind-521e906f7567?source=rss-57827543a31a------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://nenoduplan.medium.com/two-tiny-molecules-will-determine-the-future-of-mankind-521e906f7567?source=rss-57827543a31a------2</link>
            <guid isPermaLink="false">https://medium.com/p/521e906f7567</guid>
            <category><![CDATA[carbon-emissions]]></category>
            <category><![CDATA[esg]]></category>
            <category><![CDATA[climate-change]]></category>
            <category><![CDATA[water-quality]]></category>
            <category><![CDATA[water]]></category>
            <dc:creator><![CDATA[Neno Duplan]]></dc:creator>
            <pubDate>Thu, 01 Sep 2022 00:18:36 GMT</pubDate>
            <atom:updated>2025-04-18T02:30:02.454Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[WEB3 for ESG and Climate Change Management]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://nenoduplan.medium.com/web3-for-esg-and-climate-change-management-62aaf94614ba?source=rss-57827543a31a------2"><img src="https://cdn-images-1.medium.com/max/1461/1*3cAcRGj-kkALO03hOmCk2Q.png" width="1461"></a></p><p class="medium-feed-snippet">To make better progress on global warming, regulators need to find ways to reward positive and proactive behaviors instead of leveraging&#x2026;</p><p class="medium-feed-link"><a href="https://nenoduplan.medium.com/web3-for-esg-and-climate-change-management-62aaf94614ba?source=rss-57827543a31a------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://nenoduplan.medium.com/web3-for-esg-and-climate-change-management-62aaf94614ba?source=rss-57827543a31a------2</link>
            <guid isPermaLink="false">https://medium.com/p/62aaf94614ba</guid>
            <dc:creator><![CDATA[Neno Duplan]]></dc:creator>
            <pubDate>Mon, 04 Jul 2022 23:43:47 GMT</pubDate>
            <atom:updated>2022-07-04T23:43:47.601Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Hotel California]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://nenoduplan.medium.com/hotel-california-b58c1cdffc6b?source=rss-57827543a31a------2"><img src="https://cdn-images-1.medium.com/max/2600/1*ssJhixPs3RtOpny5oJC6Ew.jpeg" width="4032"></a></p><p class="medium-feed-snippet">During frequent and extended commercial breaks of the Olympics broadcast, I noticed a repeated commercial for a company I did not&#x2026;</p><p class="medium-feed-link"><a href="https://nenoduplan.medium.com/hotel-california-b58c1cdffc6b?source=rss-57827543a31a------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://nenoduplan.medium.com/hotel-california-b58c1cdffc6b?source=rss-57827543a31a------2</link>
            <guid isPermaLink="false">https://medium.com/p/b58c1cdffc6b</guid>
            <category><![CDATA[esg-investing]]></category>
            <category><![CDATA[esg-reporting]]></category>
            <category><![CDATA[ehs-software]]></category>
            <category><![CDATA[data-lake]]></category>
            <category><![CDATA[sustainability]]></category>
            <dc:creator><![CDATA[Neno Duplan]]></dc:creator>
            <pubDate>Tue, 17 Aug 2021 23:54:57 GMT</pubDate>
            <atom:updated>2021-08-17T23:54:57.791Z</atom:updated>
        </item>
    </channel>
</rss>