In AI for Building Decarbonization, Faster Analysis Makes Human Judgment More Valuable

Subscribe to Stok Insights
In AI for Building Decarbonization, Faster Analysis Makes Human Judgment More Valuable

If you’re a sustainability or real estate leader still working out how to best use AI to decarbonize your assets and portfolio, don’t worry, you’re not alone. In fact, at our Greenbuild 2026 panel on AI and decarbonization, fewer than 10% of attendees said their organization is actively using AI for decarbonization. Another 40% are piloting AI tools for decarb, and the remaining 50% haven’t started at all. In the same poll, trust and security, along with a lack of clean, structured data, came up as the biggest barriers to adoption.

This matches what we see in our own client work. While many effectively use AI for process efficiency, fewer companies have taken it directly into decarbonization.

AI is uniquely positioned to accelerate this field because building decarbonization runs on data. AI can structure building data so people can easily reach it, a big step on its own, and more importantly it can analyze hundreds of decarbonization scenarios against an owner’s unique priorities. There are far more buildings in the world than there are licensed engineers to audit them, so without these advancements we’d never reach most of the building stock. As energy engineers, we see AI as a way to impact far more buildings than we could ever reach by hand.

But technology alone doesn’t deliver results. The value comes when AI’s speed is paired with human judgment, context, creativity, and the accountability to execute.

WHERE IS AI ADDING VALUE TO BUILDING DECARBONIZATION?

AI excels in speed, pattern recognition, and scenario modeling, to name a few. For sustainability and real estate leaders, that means it can:

  • Analyze large volumes of utility, building management system (BMS), and equipment data faster than a person can.
  • Surface opportunities, outliers, and likely errors that would otherwise go unnoticed.
  • Handle routine analysis and run and compare carbon and cost scenarios at the asset and portfolio level.
Header "Where can AI add value across decarbonization workflows?" followed by 3 text boxes: (1) Analyze, enrich, and structure data, (2) Identify patterns, outliers, and opportunities, and (3) Accelerate routine analysis and scenario modeling.
Where AI can add value across decarbonization workflows.

WHERE IS HUMAN JUDGMENT AND EXPERTISE STILL CRITICAL?

The faster the analysis gets done and the more options it generates, the more valuable human judgment becomes. Both AI output and human insight become more powerful when they work together. Four places where it really matters:

#1: Fix the data that AI relies on

AI tools are only as reliable as the data they run on, yet many companies are buying AI tools before they trust their own building data. Strong portfolio-level data management is a competitive advantage for owners and occupiers, and the rise of AI only strengthens the case. A consultant can find the gaps, strengthen the inputs, and decide which data is credible enough to act on.

Different kinds of AI also need different data. Large language models (LLMs) handle unstructured documents such as PDFs well. The machine-learning models behind fault detection and energy forecasting need clean, consistent time-series data. On a greenhouse gas inventory for a global investment management firm, LLMs pulled data out of PDFs quickly but misread the layout of utility statements, mixing up line items shown for cost transparency with the values that mattered: total consumption and total spend. The tools saved real time, and an expert caught the error before anyone acted on it. AI should save you time but not replace expert review.

Even well-structured data needs interpretation. One BMS can hold five trends with nearly identical point names, and utility bills are infrequent, limited in scope, and sometimes inaccurate. If you haven’t documented how your physical assets, meters, and data streams connect, in a form an AI tool can easily read, expect it to misclassify data or invent answers.

#2: Put outputs in context

Experience is data too. Years of completed projects give an advisor benchmarks and lessons learned that most models don’t have. AI knows what it was trained on, and chances are it wasn’t trained on your organization, your portfolio, your people, your priorities, or your constraints. That context is what enables custom, creative recommendations that fit each organization’s unique needs.

“AI can give us a strong first pass in many cases, but often the final call must reflect our investment priorities, our tenants, and the long-term values of our diverse business. Those priorities shift and may exist in live conversations, not ready-to-digest AI content. It takes people who know the business, and partners who know the technical details, to consider these shifts and adjust.”
—Alex Spilger, Vice President, Real Estate Sustainability, The Sobrato Organization

#3: Challenge and validate before anything reaches a capital plan

Many AI outputs are directional rather than decision-ready. They still need an expert to ask what the model got wrong, what data was missing, and which assumptions may have driven big swings in the results, and then remain accountable for the result. This human expertise is what helps determine whether those insights are accurate, relevant, actionable, and ready for capital planning.

Before it gets there, three things should be confirmed: the data behind it can be traced to its source; it accounts for real limits such as electrical capacity and tenant disruption; and it aligns with the capital cycle, so the project lands when equipment is due for replacement.

#4: Tie the tools together and act

Many organizations are dealing with tech-stack fatigue. Someone neutral, who understands the goals, has to tie tools into one cohesive program.

More robust dashboards with real-time data won’t automatically help a facilities manager who’s already buried in tickets. What will is a short list of recommended next actions, prioritized by cost and carbon, and an expert to implement them at the asset level, bridging the gap between data and delivery.

TWO EXAMPLES OF AI IN BUILDING DECARBONIZATION

#1: Carbon and cost reduction scenario modeling

The Sobrato Organization used an AI tool to prioritize carbon and cost reduction opportunities at two of its non-profit, equitable housing assets. The tool analyzed hundreds of potential solutions and identified several large capital projects. Sobrato is now moving forward with two large solar projects while also considering an electrification project. The speed allowed the team to iterate and run far more scenarios. The judgment was in choosing solutions that would maximize value both for the owner and the tenant.

#2: Simplified building management

Stok’s extensible AI-enabled platform allows portfolio and property managers to quickly understand their building or portfolio, including energy use, systems information, and live BMS trends. They can query the platform to find real-time operational information about their portfolio as easily as using a search engine and, more importantly, Stok can build bespoke AI tools within the platform for these clients to significantly reduce the time and effort to produce analysis like portfolio decarbonization planning and climate risk analysis, which used to be very expensive to perform. We’re currently using this platform to automate retro-commissioning (RCx) and fault detection across a large global real estate portfolio for a corporate technology client.

WHAT CAN SUSTAINABILITY AND REAL ESTATE LEADERS DO RIGHT NOW?

For those unsure where to start or how to grow, jumping into a long-term contract with an AI tool in an era where technology and AI capabilities are changing every week may not be the best answer. The most valuable thing many companies can do right now is begin with small-scale pilots with multiple AI tools to build internal fluency and identify high-value use cases before a larger investment in the most aligned tool(s). Consultants can work alongside this AI experimentation to evaluate tool effectiveness and enable execution, bridging insight and delivery.

We’re here to support your company’s AI-powered decarbonization efforts. Reach out to our team to discuss where your company is today and how to use AI-enabled tools more strategically for faster progress toward your climate goals. Also, we won’t pretend we have all the answers. What questions are you grappling with about the role of AI in our shared work? Let’s figure it out together.

Share this post