infrared.city
An AI-powered urban microclimate simulation platform that returns wind, thermal comfort, solar and daylight analyses in seconds instead of days, reachable from a web app, native CAD, BIM and GIS plugins, a Python SDK or a REST API.
What is infrared.city?
infrared.city is an urban microclimate simulation platform published by Infrared City GmbH in Vienna. It replaces conventional computational fluid dynamics with AI surrogate models trained on large libraries of pre-computed simulations, so analyses that traditionally take hours or days return in seconds. The vendor claims results in one second against three days for traditional CFD, at 0.1% of the cost, from models trained on more than 500,000 simulations across over 100 cities, with accuracy above 90%.
The catalogue covers ten documented analyses: wind speed, annual pedestrian wind comfort across up to sixteen directions, thermal comfort on the UTCI scale, heat stress, cold stress, microclimate, daylight availability, sky view factor, direct sun hours and solar radiation. Recognised comfort criteria are exposed explicitly, including Lawson 1970, Lawson 2001, Lawson LDDC, Davenport, VDI 3787 and NEN 8100 for both comfort and safety. Solar figures come from Radiance-based sky matrix calculations fed by direct normal and diffuse horizontal irradiance taken from climate files.
Two AI layers sit above the raw results. The AI Assistant answers project questions conversationally and draws on region-specific building regulations and comfort criteria. AI Reports turns simulation output into KPIs, design-option comparisons, problem-area flags, intervention ROI and stakeholder-ready documents, with customisable templates and shareable interactive links.
The platform is reachable four ways: the web application, native plugins for Rhino/Grasshopper, Revit, QGIS and ArcGIS, a published Python SDK, and a REST API at version 2 with API-key authentication, asynchronous jobs and signed webhooks. A single token currency spans all three channels, at ten tokens for a standard analysis and fifty for pedestrian wind comfort, while AI workflows are priced by area bracket. SketchUp and Speckle are announced but not yet shipped.
The product began as InFraReD, a research project of the City Intelligence Lab at the Austrian Institute of Technology, several years before the company itself was entered in the Austrian commercial register in September 2023. Named testimonials come from SvN Architects + Planners and Skidmore, Owings & Merrill.
What it does
- Run wind, thermal comfort, solar and daylight simulations over an urban site in seconds
- Compare massing options and measure the resulting difference in pedestrian comfort
- Test interventions such as tree planting, ground materials and shading structures
- Generate automated reports and share them as interactive web links
- Call the same analyses from Rhino/Grasshopper, Revit, QGIS or ArcGIS without exporting geometry
- Drive batch studies through the Python SDK or REST API with asynchronous jobs and webhooks
- Ask the AI Assistant for guidance grounded in local regulations and comfort criteria
When to use infrared.city / When not to
A quick filter to help you decide if infrared.city is the right fit.
When to use infrared.city
- Architects and urban designers who need environmental feedback while massing is still moving
- Computational design teams pairing the engine with Grasshopper optimisers such as Galapagos or Wallacei
- Sustainability consultants producing comfort and heat-stress evidence for project deliverables
- City authorities and public works teams weighing interventions such as tree planting, cool surfaces or shading
- GIS analysts and developers embedding climate layers into district-scale studies or a digital twin through the API
When not to use infrared.city
- Teams needing a reference-grade CFD sign-off, since these are AI surrogate models with a claimed 90%+ accuracy
- Buyers who require published contractual terms before signing up, as no public terms and conditions page exists
- Organisations with a data residency requirement, because no hosting location for customer data is published
- Procurement processes that demand a security certification such as ISO 27001 or SOC 2, none of which is claimed
- Anyone wanting mobile access or a non-English interface, neither of which the vendor offers
How to use infrared.city
A typical end-to-end flow, from setup to results.
- Create an account on the platform, with no credit card required
- Collect the one-off grant of 900 tokens, valid for 42 days after registration
- Draw or import a site boundary in the web application
- Pick an analysis type, such as wind speed, pedestrian wind comfort, UTCI thermal comfort or solar radiation
- Read the results as map layers and adjust massing, vegetation or ground materials to compare options
- Alternatively install the plugin for Grasshopper, Revit, QGIS or ArcGIS and follow the dedicated tutorial
- For scripted work, install the Python SDK, which requires Python 3.9 or later
- Set the INFRARED_API_KEY environment variable, and INFRARED_BASE_URL if a custom endpoint is needed
- Submit large batches as asynchronous jobs and collect them through signed webhook notifications
- Generate an AI report and share it as an interactive web link with the project team
Pros & Cons
Pros
- Response times measured in seconds, fast enough to sit inside a design iteration loop
- Recognised comfort standards exposed explicitly, including Lawson, Davenport, VDI 3787 and NEN 8100
- Runs inside the tools teams already use, with no geometry export or format conversion
- Public Python SDK and documented REST API, with one token currency across web, plugins and API
- Free entry with 900 tokens and no credit card required
- Fully verifiable vendor identity: Austrian commercial register entry, valid VAT number, matching business purpose
- Team and academic affiliations published by name, with named client testimonials
Cons
- No public terms and conditions page, so the contractual framework cannot be reviewed before signing up
- No published hosting location for customer data
- No security certification claimed, neither ISO 27001 nor SOC 2
- Neither a DPA nor a platform subprocessor list is published
- Nothing published about whether customer data trains the models, and no opt-out is offered
- The free tier expires 42 days after registration and does not refill; billing is annual only
- SketchUp and Speckle are advertised as coming soon but are not available
Pricing & Plans
A free entry point exists but is not a permanent free plan: every new account receives a one-off grant of 900 tokens at signup, without a credit card, and that grant expires 42 days after registration and does not refill. The lowest recurring paid price is EUR 240 per year for the Starter plan, billed annually, which is the only billing period offered. Advanced costs EUR 1,200 per year and Pro EUR 6,000 per year. All three paid plans unlock every simulation model and differ only in monthly token allowance and sustained request rate. Consumption is metered in tokens at a list rate of EUR 0.10 each, with a standard analysis costing 10 tokens and pedestrian wind comfort 50; the extra-token rate falls from EUR 0.08 to EUR 0.02 as the plan grows. Enterprise work is quoted individually.
- EUR 0 - Students and small studies - 900 tokens granted once at signup
- expiring after 42 days
- EUR 240 per year - Small sites - 500 tokens per month
- 10 requests per second
- extra tokens at EUR 0.08
- EUR 1
- 200 per year - Quarter scale - 4
- 000 tokens per month
- 30 requests per second
- extra tokens at EUR 0.05
- adds comparison and AI reports
- EUR 6
- 000 per year - District to city - 50
- 000 tokens per month
- 60 requests per second
- extra tokens at EUR 0.02
- price on request - digital twin
- live climate layer for a city
- or co-development on the API
Data, GDPR & hosting
A consolidated view of how infrared.city handles your data.
GDPR overview
GDPR implementation is detailed but never advertised as compliance. Infrared City GmbH identifies itself as controller and cites its legal bases explicitly: consent under Article 6(1)(a), contract under 6(1)(b), legal obligation under 6(1)(c) and legitimate interest under 6(1)(f). Data subject rights are enumerated, covering access, rectification, portability, objection, restriction and erasure. The competent supervisory authority is named, the Austrian Datenschutzbehorde at Barichgasse 40-42, 1030 Vienna. Transfers outside the EEA are said to rely on an adequacy decision, contractual guarantees or explicit consent under Article 49(1)(a). A seven-year accounting retention obligation is cited under section 132 of the Austrian Federal Fiscal Code. The phrase GDPR compliant appears nowhere on the site, so compliance is applied in substance but never claimed in words.
Who owns the data?
Ownership of project data cannot be established from the published documents. The vendor publishes no terms and conditions page, and five candidate paths all return the site's 404 page, so no contractual clause on the ownership of uploaded geometry or simulation output is available for review. The only identified controller is Infrared City GmbH, named in a privacy notice dated 29 November 2023 with Angelos Chronis as director. That notice governs the marketing website rather than the simulation platform: it describes visitor analytics, newsletter and waitlist data, and payment handling, but is silent on who owns the models, sites and results a customer creates.
Reuse rights
The published documents do not state whether a customer may freely reuse the data and results produced on the platform, because no terms and conditions page is available. The privacy notice covers only personal data on the marketing website: browsing data such as IP address, device, operating system, mobile carrier, browser version and access time, gathered for site operation and statistics; newsletter details passed to Mailchimp on consent; waitlist details including company name; and payment details handled by PayPal and Stripe. It records no automated decision-making under Article 22 GDPR, and says nothing about customer data feeding model training. Reuse rights for simulation output therefore have to be settled directly with the vendor.
Data retention & training
Hosting summary
No hosting location is published for the platform or for customer data. The only countries named anywhere on the site concern the marketing website rather than the simulation service: Microsoft Clarity and Google Analytics are described as transmitting and storing website usage data on servers in the United States, covered by a European Commission adequacy decision. Treating that as the platform's hosting arrangement would be a misreading. The privacy notice states that transfers outside the EEA occur only under an adequacy decision, appropriate contractual safeguards, or explicit consent under Article 49(1)(a) GDPR, but names no data centre, region or cloud provider for the simulation platform itself. The vendor is established in Austria and its payment providers are European, PayPal (Europe) in Luxembourg and Stripe Payments Europe in Ireland. Any residency requirement therefore has to be confirmed directly with the vendor, since the published documentation does not answer it.
Things to keep in mind
Risks and trade-offs to weigh before adopting infrared.city.
- AI surrogate models approximate rather than reproduce reference CFD; the claimed 90%+ accuracy is the vendor's own figure, so treat results as design guidance and not as regulatory sign-off
- Speed invites over-trust: results returning in seconds can be accepted without the sanity checks a slower workflow would have forced, and comfort criteria still need an engineer who understands them
- No public terms and conditions page means data ownership and reuse rights for uploaded geometry and simulation output are undetermined before you commit
- No hosting location, DPA or platform subprocessor list is published, which matters for public bodies and anyone with data residency obligations
- Nothing is published about whether customer data trains the models, and no opt-out is offered, so site geometry may be uploaded without a clear commitment on its downstream use
- The headline performance and adoption figures are unverified vendor claims and should not be repeated to stakeholders as established fact
- The privacy notice and imprint carry different company addresses and both date from late 2023, so published legal information lags behind the commercial register
Setup & Integrations
Technical difficulty
Low to moderate, depending on the route. The no-code path is straightforward: create a web account, collect 900 free tokens without a credit card, then draw or import a site and run an analysis, with the vendor stating that local climate data and technical configuration are handled automatically and no specialists are needed. The plugin route adds an install for Grasshopper, Revit, QGIS or ArcGIS, each with a dedicated tutorial. The developer route requires Python 3.9 or later, an API key and two environment variables. Domain literacy matters more than setup: wind comfort criteria, UTCI and sky view factor need interpreting.
Deployment
Integrations
Behind infrared.city
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What does infrared.city actually do?
How fast is it compared with traditional simulation?
How were the models trained, and how accurate are they?
Can I start without paying?
How much does one analysis cost?
Which design tools does it integrate with?
Is there an API?
Which comfort criteria are supported?
Who is behind the tool?
Where is customer data hosted?
Should you pick infrared.city?
infrared.city addresses a real bottleneck in sustainable design: environmental analysis has traditionally arrived too late and too slowly to influence the decisions that matter most. By substituting AI surrogate models for full CFD runs, it moves wind, thermal comfort and solar studies into the phase where massing is still negotiable, and it does so inside Rhino, Revit, QGIS and ArcGIS rather than asking teams to change how they work. The published grounding is unusually explicit for this category, with named comfort standards such as Lawson, Davenport, VDI 3787 and NEN 8100, a documented Python SDK and a REST API with asynchronous jobs and signed webhooks.
The vendor identity is fully verifiable, which is not always the case in this market: Infrared City GmbH is entered in the Austrian commercial register, its VAT number validates, its registered business purpose matches the product, and its managing director is the CEO named on the team page. The company grew out of a research project at the Austrian Institute of Technology, and a EUR 1 million pre-seed round is corroborated by the shareholder list at the register.
The reservations are documentary rather than technical. There is no public terms and conditions page, so the contractual framework, data ownership and reuse rights cannot be assessed before signing up. No hosting location for customer data is published, no security certification is claimed, and nothing is said about whether customer data trains the models. The performance figures are the vendor's own and carry no third-party verification, and the models remain approximations rather than reference CFD. For fast, iterative design exploration the tool is compelling; for regulated sign-off or a procurement process with residency and certification requirements, expect to ask questions the website does not answer.
- Choosing a selection results in a full page refresh.
- Opens in a new window.