Why We Invested
Why we invested in DiLT Analytics: explainable AI for the buildings that run Europe
Part of our Why We Invested series, where Noctua Science Ventures shares the reasoning behind our portfolio bets with the deep tech community and our investors.
Buildings are one of the largest, most stubborn energy problems in Europe — and one of the least digitised. According to the European Commission, buildings account for roughly 40% of the EU's energy consumption and 36% of its energy-related emissions. Yet the operational intelligence needed to run them efficiently still lives in PDFs, spreadsheets, disconnected sensor streams, and the heads of a shrinking pool of experienced technicians.
We invested in DiLT Analytics because its product, FaciliMind, tackles that gap at the root: it turns raw building data into automated, explainable fault diagnoses that facility teams can actually act on. This is our thesis, the market we see, and the technical edge that convinced us.
The problem: technical building knowledge doesn't scale
Modern buildings are heterogeneous by nature — a patchwork of legacy protocols (BACnet, ModBus, M-Bus, MQTT) layered over decades of retrofits. Turning that mess into a machine-readable model is done largely by hand today: slow, expensive, error-prone, and impossible to scale across a portfolio.
At the same time, three forces are pushing operators toward continuous technical monitoring, not away from it:
- Regulation. The revised Energy Performance of Buildings Directive (EPBD) entered into force in 2024, with member states required to transpose it into national law by May 2026. ESG and taxonomy reporting are turning building performance into a board-level metric.
- Cost. Energy prices remain structurally elevated, and operating-cost pressure is relentless for institutional real-estate portfolios.
- Labour. A deepening skilled-labour shortage in facility management means the expert knowledge that keeps buildings efficient is getting scarcer, not more abundant.
The result: continuous, software-driven fault detection is shifting from nice-to-have to must-have for anyone operating a serious building portfolio.
The technical edge: automated diagnosis, not just anomaly flags
Most "AI for buildings" tools stop at anomaly detection — they flag that something is off and leave a human to figure out what and why. FaciliMind goes further, and the way it does so is the core of why we invested.
An automated semantic digital twin
FaciliMind ingests raw building-management-system data and automatically generates a knowledge graph — a semantic digital twin — of the building's systems. This is the step competitors typically do manually, and it is the single biggest bottleneck in the category. Automating it collapses onboarding effort dramatically and is what makes portfolio-scale rollout realistic.
Explainable by design
DiLT's approach is best summarised as "the AI builds the rules, and operations then run rule-based." Instead of a black box, technical buyers get transparent, causal fault diagnoses — each with a probable root cause, a concrete recommendation, and an estimated ROI per repair. In our diligence, this explainability came up again and again as the reason technical buyers preferred DiLT over black-box alternatives. Engineers trust what they can inspect.
Analytics-only positioning
FaciliMind reads and diagnoses; it does not write commands back to the building-management system. That keeps it clean from a safety and security standpoint and lowers the barrier to adoption for risk-averse operators.
DiLT moves the category from "here's an anomaly" to "here's the fault, here's why, here's the fix, and here's what it's worth."
The market: a large, tailwind-driven category
DiLT's ideal customer is the institutional real-estate operator — asset managers, property managers, and facility-management firms running portfolios of buildings where ESG pressure and operating cost meet. With millions of non-residential buildings across the EU and a clear willingness to pay for technical monitoring, the addressable opportunity runs into the tens of billions of euros.
Critically, this is a market with genuine tailwinds rather than one that needs to be created: regulation is mandating measurement, energy cost is forcing efficiency, and the labour shortage makes automation the only viable path. Independent, analytics-first software also extends the life of existing hardware — a sustainability advantage over incumbents whose incentives point toward selling more equipment.
The team: research depth meets operator experience
Deep tech lives or dies on the team, and DiLT pairs two things that rarely sit in the same company:
- Academic depth. DiLT is a spin-off of TU Graz (Intelligent Systems Group) and TU Wien (Integrated Building Technology Group), co-founded by two of Europe's leading researchers in model-based diagnosis and data-driven smart buildings. An active research pipeline feeds the product roadmap.
- Operator experience. The company is led by a commercially minded CEO with deep HVAC-data expertise and a COO with 15 years in renewable energy and a prior exit — a team that knows how to move from research project to enterprise product.
Across our reference calls with buyers in aviation, industry, and institutional real estate, the feedback was consistent: a credible, differentiated technical approach delivered by a team that technical customers genuinely want to work with. That combination of research moat and go-to-market credibility is exactly what we look for at pre-seed.
Our thesis in one line
We backed DiLT Analytics because it is turning a fragmented, manual, expert-dependent problem into automated, explainable software — in a market that regulation, energy cost, and labour scarcity are all pushing in its direction. If the category standard for building operations shifts from anomaly detection to automated root-cause diagnosis, we believe DiLT is positioned to define it.
We're proud to support Theresa, Chris, and the whole DiLT team on that journey.
Frequently asked questions
What does DiLT Analytics do?
DiLT Analytics builds FaciliMind, an AI platform for automated fault detection and diagnostics (FDD) in HVAC and building energy systems — effectively predictive maintenance for buildings. It generates a semantic digital twin of a building and produces explainable fault diagnoses with root cause, recommendation, and ROI.
How is FaciliMind different from other building-AI tools?
Most tools stop at anomaly detection. FaciliMind automates the semantic modelling of the building and delivers explainable, causal diagnoses rather than black-box alerts — while staying analytics-only (it does not control the building-management system).
Why did Noctua Science Ventures invest?
Because DiLT combines a genuine research moat (a spin-off of TU Graz and TU Wien) with operator-led commercial execution, in a large market driven by EU regulation (EPBD), energy cost pressure, and the facility-management labour shortage.