Smart Building AI Compared: FrostLogic vs. BrainBox, Clockworks, Nuuka

Every vendor comparison page has the same plot. The company writing it wins every row of the table, and the reader learns nothing.

We’d rather you pick the right tool, even when it isn’t ours. So this comparison starts from an honest premise: FrostLogic Explore, BrainBox AI, Clockworks Analytics, and Nuuka are all serious products, and they answer different questions. If you know which question your building is asking, the choice mostly makes itself.

The four questions, roughly:

• “Can the AI just run my HVAC?” That’s BrainBox AI.

• “What’s broken, and what is it costing me?” Clockworks Analytics.

• “Can one platform hold all my building data and trim energy use?” Nuuka’s pitch.

• “Out of everything my sensors are saying, what should I act on today?” That one is ours.

What follows is how each works, with numbers from their own published materials.

BrainBox AI: autonomous control, now part of Trane

BrainBox AI connects to your existing BMS, learns the building’s thermal behaviour, and then writes new setpoints back to HVAC equipment every five minutes. No human in the loop. The cloud decides, the building obeys.

The scale is real. Over 14,000 commercial buildings and more than 100 million square feet run on it, per company statements from 2025. Trane Technologies, the Irish-domiciled HVAC group, announced its acquisition of BrainBox in December 2024 and closed it in January 2025. BrainBox is now a product line inside a US-listed conglomerate rather than an independent vendor, sold as a subscription through Trane’s global channel.

Pricing is unusually transparent for this market: the AWS Marketplace listing shows US$0.25 per square foot for a 12-month contract, with steep discounts for 24 and 36 months. The headline claim is up to 25% HVAC energy cost reduction. Worth noting: every published result is a vendor case study with “up to” framing. We found no independent measurement and verification study in public sources.

A Nordic commercial buyer should read the fine print, though. The control loop lives in the cloud, the terms of service make the customer responsible for connectivity, and internet failures sit explicitly under force majeure.

The data terms deserve a slow read too. Customers keep title to their data, but the trained building model belongs to BrainBox, data export happens on termination within a 30-day window, and the contract defaults to Delaware law. There is no published EU data residency commitment.

We also could find no documented Nordic deployment, and no mention of BREEAM or Nordic Swan anywhere in their materials. Certification reporting isn’t what this product is for.

Clockworks Analytics: the diagnostician

Clockworks is the platform we respect most in this lineup, because it shares our core conviction: facility teams need a prioritised list of issues, not another dashboard. They’ve held that line since 2008, when the company spun out of MIT’s Building Science Department.

The product does fault detection and diagnostics on HVAC systems. A gateway pulls BMS data to the cloud every five minutes, a physics-based engine builds a model of each piece of equipment, and every detected fault gets scored 0 to 10 by its impact on energy, comfort, and equipment reliability, with the avoidable cost in currency attached. Their landing page cites more than 500 million square feet under analysis, and a September 2025 Microsoft customer story puts cumulative documented client savings at $69 million. Customers include Kaiser Permanente, Johns Hopkins, and MIT itself. The company raised a $16.1M Series E in 2023 and remains independent, with Schneider Electric as both investor and channel partner.

The boundary is scope. Clockworks is deep on HVAC mechanics and deliberately stays there. Their materials describe no forecasting, no what-if simulation, and no certification compliance tracking. It tells you what’s broken with impressive precision. It doesn’t predict next week’s energy demand or track your BREEAM thresholds between audits.

For Nordic buyers there’s a practical wrinkle: Clockworks reaches Scandinavia through service partners, chiefly GK and ISS, rather than directly. That works well if you already buy technical FM from those firms. It adds a layer if you don’t. Hosting is Microsoft Azure, US company, no published EU residency commitment, and pricing is quote-only.

Nuuka: the Nordic data platform that picked a lane

Nuuka is the closest to home, in both geography and history. Founded in Helsinki in 2012, it won the 2018 tender to monitor energy and water across roughly 1,000 City of Helsinki buildings, and its reference list reads like a Finnish institutional who’s who: YIT, NCC, Varma, the cities of Oulu and Lappeenranta.

It’s also a company mid-pivot. Over 2024 and 2025 Nuuka rebranded from a broad building data platform to “Nuuka AI” with three products: an Optimizer that writes HVAC commands back to the building every one to five minutes (up to 44,000 adjustments a month, by their count), a Diagnostics module for fault alerts, and an Analytics module with 200+ KPIs. Claimed savings run 10 to 30% on energy, vendor-published. Credit where due: like BrainBox, Nuuka does close the loop on HVAC control, which is more than most analytics platforms can say.

The interesting part is what the pivot left behind. Nuuka’s older positioning leaned on sustainability reporting, compiling data for GRESB and GRI submissions. The new site mentions ESG support mostly in passing, and we found no continuous certification tracking against BREEAM, LEED, or Nordic Swan criteria in their published materials. The compliance story is data compilation for annual submissions, not monitoring between them.

Nuuka is also small. Around 30 employees, modest funding history, and a fresh round from YIT Ventures in November 2025. Small can mean focused and responsive. It can also mean a roadmap that follows its largest shareholders. Hosting is Microsoft Azure, with no published data ownership or export commitment we could find.

The comparison table

Claims in competitor columns come from each vendor’s published materials as of June 2026. “Not in published materials” means exactly that: we couldn’t find it, which is not proof it doesn’t exist. Check with the vendor.

When each platform is the right fit

Choose BrainBox AI if you run a large portfolio of similar buildings, HVAC energy is your dominant cost, and you’re comfortable handing control to a cloud service backed by Trane’s channel. Retail chains and logistics portfolios fit this profile. It’s the strongest pure energy-autopilot in the market.

Choose Clockworks if you operate complex mechanical plant where equipment failure is expensive, think hospitals, labs, and campuses, and you have (or contract) a technical team to work through a diagnostic queue. Seventeen years of HVAC diagnostic depth is hard to argue with.

Choose Nuuka if you’re a Finnish or Nordic portfolio owner who wants HVAC optimization with local references and a vendor that speaks the market’s language, and your compliance needs end at annual GRESB submissions.

Choose FrostLogic Explore if your problem is broader than HVAC: you need anomaly detection across BMS, energy meters, and IoT sensors, forecasting you can plan against, certification compliance tracked continuously rather than audit by audit, and you want to keep decisions with your team. Or if your legal department has opinions about where building data lives.

What makes Explore different

The honest summary of the table: each competitor automates one layer. BrainBox and Nuuka automate control. Clockworks automates diagnosis. Explore automates the layer in between, the one where someone has to look at everything the building is saying and decide what matters today.

That’s a deliberate choice. Autonomous control optimises within the rules it knows. A decision queue surfaces what the rules don’t cover, like the anomaly that looks like weather but isn’t, or the compliance threshold that will breach in three weeks. Explore runs six detection methods with causal filtering, forecasts each metric up to seven days out with confidence bounds, and tracks BREEAM, LEED, and Nordic Swan criteria against live sensor data continuously. When you want to test a setpoint change, you simulate it before anyone touches the building.

Regulation is moving the same direction. The recast Energy Performance of Buildings Directive already requires building automation with continuous monitoring and fault detection for large non-residential buildings, and extends to systems above 70 kW by the end of 2029. Annual snapshots won’t satisfy a directive written around continuous monitoring.

And then there’s the question none of the marketing pages answer: whose building is it? With Explore, your data and your trained models are exportable from day one, hosting is in the EU, and the Enterprise edition runs on your own infrastructure if you prefer. We think that should be table stakes. The table above shows it isn’t.

Not sure which approach fits your building? Start a conversation.