Section C.3 — Parametric Assistants
C.3 Parametric Assistants
New digital planning tools currently fall into three clearly distinguishable categories: large language models for working with knowledge in text, AI-supported image generation, and task-specific algorithmic assistants. The first two were covered at length in the previous chapters. The third — task-specific algorithmic assistants — addresses core operational processes in planning with a high degree of functional specialisation.
These parametric assistants concentrate on clearly bounded sub-tasks within complex planning processes. Unlike universal chatbots or generative image models, they rest on narrowly defined data structures, rule-based logic or optimisation methods. Their fields of use are automated floor plan generation within building regulation parameters, optimising structural grids, simulation-supported energy and daylight analysis, and algorithmic cost forecasting in early service phases. The aim is not open exploration but the systematic reduction of the range of options against criteria that can be quantified.
What stands out is the energy of young technology companies that have developed specialised solutions for precisely defined problems in recent years. Many of these applications work as extensions or interfaces to existing planning environments. Rather than replacing established BIM structures, a layer of additional AI-supported functionality is emerging. These systems read data out of existing models, evaluate it and feed optimised results back. The emphasis shifts from modelling alone towards algorithmic evaluation and support for decisions.
The economic effect comes mainly from automating repetitive checking and optimisation. Iterative comparisons of options that were previously done by hand, or half by hand, can now be calculated quickly and against consistently applied criteria. At the same time a new transparency emerges about conflicting objectives — between space efficiency, circulation logic, energy performance and construction cost. Planning decisions become more data-based without replacing responsibility for the design.
Parametric assistants therefore mark no single technological leap but a gradual transformation of planning method. Taken together, a hybrid planning environment emerges in which algorithmic procedures act as supplementary, specialised tools within a design process that remains a matter of human responsibility.
ChatGPT AudioChat, February 2026 – revised by Claude – translation: Claude, August 2026.
C.3.1 Site and Location Analysis
Every design begins with a question: what kind of place is this? New digital tools do not make the question redundant. But they let you answer it faster and more precisely.
What has changed is not that the data exists. Geodata, development plans, infrastructure connections were all there before. What has changed is how accessible they are. Information that once meant travelling between the land registry, the geodata service and the building control office can now be queried through map interfaces.
The practical consequence: if you can analyse public transport connections, map the distribution of green space and quantify development potential within minutes, you work on a different basis in service phase 1. You can test more options, get into substantial conversations with clients earlier and justify decisions more robustly.
What no tool takes over: recognising that a noisy junction calls for urban articulation — or that a quiet south-facing side forms the natural centre of gravity of a design. The contextual factors that matter to a design still have to be identified by us. The digital tools supply data. Interpretation stays a human task.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.1.1 Municipal Geodata Services — EXAMPLE: Wirtschaftsatlas Berlin
Municipal geodata portals are the most reliable primary source for information about a location. Cities and municipalities are legally obliged to provide geodata free of charge. Accordingly, digital map series have existed for years covering land use plans, development plans, standard land values, environmental information, infrastructure and heritage protection. What is new is less the data itself than the quality of its presentation and navigation.
The Wirtschaftsatlas Berlin is exemplary. The portal brings official datasets together in an interactive map application. Thematic layers can be combined, filtered and analysed spatially. Besides development plans and commercial land, it also shows demographic structures, the distribution of industries and environmental indicators.
For architectural practices that means one thing above all: the appraisal phase gets shorter. Much of the information that once had to be collected across several visits to authorities is now directly available. One side effect not always considered in everyday practice: clients and members of the public can reach the same data. Planning is therefore more visible than it used to be.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.1.2 TOOL: Google Earth — AI Studio (Delve / Sidewalk Labs)
Google Earth has been an obvious tool for orientation for years. Aerial images, terrain models and perspective views give a spatial impression of almost anywhere in the world within minutes. Combined with Google Maps it offers easy access to infrastructure, patterns of use and urban context. For early project phases it replaces no site visit, but it does allow a quick first analysis.
Google Earth AI Studio extends the platform with AI-supported functions. The aim is not only to display the existing geodata but to evaluate it algorithmically and process it generatively. The focus is on automated analysis, pattern recognition and conceptual design approaches based on large-scale datasets. The built-in massing generator draws particular attention; its conceptual basis goes back to work by Sidewalk Labs.
First tests show the limits of fully automated systems. The generated massing is produced from defined parameters but allows planners only limited intervention. Differentiated urban transitions, spatial hierarchies or context-related fine-tuning cannot be moderated adequately this way. Technical capability and planning quality are not the same thing.
At the time we tested it, AI Studio was only accessible in the United States. That limits how directly it can be used on European projects.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.1.3 TOOL: Aino.World
Aino belongs to a new category of geodata tool: a map application with a dialogue interface. You interact in natural language, and the result appears immediately as a map of the location you chose.
You can mark a site on the map and then question it directly — about what is reachable within a 20-minute walk, about where the green spaces are, about infrastructure. Analyses like these used to need either several data sources or a manual overlay of different layers. Aino reduces that to one direct, context-aware query.
We tested Aino.World early and gave the founder feedback on which scenarios from planning practice would be particularly relevant. The global scale is a genuine advantage: almost any location in the world can be examined.
The limitation is the depth of data for the German context. Aino does not draw on official vector data. Site boundaries, cadastral parcel geometries and planning-law indicators are not supplied with the precision that binding planning decisions require. Aino.World suits exploratory, conceptual phases and quick visual assessments of a location.
Reference: Alex Kamenev.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.1.4 TOOL: Plan4Better
Plan4Better is built consistently around the German planning context. It draws on official German geodata. Cadastral boundaries, land registry information, infrastructure networks and other public datasets are brought together and made accessible in an interactive map.
The main value lies in the precision of the vector data. Site dimensions, parcel divisions and positional relationships correspond to the official records and are reliable enough to work from.
The linked infrastructure and accessibility analyses are particularly relevant. Public transport stops are not just shown as points but combined with data on frequency and connectivity. From that you get realistic catchment areas — based not on physical distance alone but on how people actually move.
Reference: Elias Pajares.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.1.5 TOOL: Syte
Syte concentrates on the systematic evaluation of planning-law and cadastral data in Germany. Its primary audience is developers and institutional players who want to identify strategic development potential from large datasets.
At its core the system aggregates official information from building authorities, land registries and other public sources. Development plans, planning-law indicators, site areas and existing structures are brought together in a map covering the whole of Germany.
The AI-supported evaluation serves to identify potential for infill and development. Algorithms check sites for what planning law permits, identify possible additional storeys and analyse underused land.
What Syte shows architectural practices: planning-law information is no longer only interpreted as text but structured and analysed by machine. That changes who in the market can answer which questions, and how fast.
Reference: Anika Meisters und Matthias Zühlke.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.1.6 TOOL: Infrared.City
Infrared City addresses a question that used to require considerable computation: simulating microclimatic effects in an urban context. Classic flow and temperature analyses based on CFD calculations take time and are usually commissioned only in later design phases — by which point the essential decisions about volume have long been made.
Infrared City takes another approach. Using trained AI models it predicts simulation results without running every calculation in full. Statements about heat island effects, shading or wind flow in relation to building volumes therefore arrive far more quickly.
The decisive gain is that it comes earlier. Climatic effects can be assessed while the concept is still forming, when changes are still possible. Infrared City does not replace a detailed specialist simulation on complex projects. It offers a reliable first estimate that raises sensitivity to climatic relationships.
Reference: Oana Taut.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.1.7 TOOL: r3leaf
r3leaf extends microclimatic analysis into an operational dimension. The focus is not only on simulating existing conditions but on assessing specific measures of adaptation in the urban realm. Climatic effects are tied directly to planning interventions. The tool is at an early stage of development. We are watching how it progresses.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.1.8 TOOL: Forma (Spacemaker)
Autodesk Forma is positioned primarily as a tool for developing urban volume. Alongside generating and editing massing, though, it offers its own set of environmental analyses, which places it among the tools for sustainability-oriented planning.
Within the massing model you can simulate and visualise various environmental parameters: shading studies, solar radiation, potential photovoltaic yield and noise propagation in particular. The analyses are tied directly to the geometry of the modelled volumes. Changes to height, orientation or density feed straight through into the calculated values.
Unlike fully automated systems, Forma keeps the massing under manual control. Volumes can be stretched, compressed, moved and adjusted in proportion. Sustainability therefore informs rather than determines.
Forma thus occupies a middle position. As a massing generator it belongs to the scale of urban development. Through its built-in environmental analyses it also meets central requirements of sustainability-oriented planning, connecting work on volume with performance-based assessment.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.1.9 TOOL: Foster + Partners — CYCLOPS
It is worth noting that large practices are now contributing to the share economy of new software solutions — here Foster + Partners of London.
Cyclops is an analysis plug-in for Grasshopper within Rhino, developed by the research and development department at Foster + Partners. Its aim is to bring environmental and performance analysis directly into the architectural design process, and so allow faster, data-based decisions.
The software uses GPU-accelerated ray tracing to speed complex simulations up considerably. Calculations that conventionally take hours or days — analysing solar radiation or daylight, say — can be carried out in something close to real time. Cyclops supports various kinds of analysis, among them solar radiation, daylight, hours of sunlight and shading. Because it sits inside Rhino and Grasshopper, the whole process stays within the familiar parametric modelling environment.
The tool was developed inside Foster + Partners over more than fifteen years and tested on real projects. Releasing it as a freely available plug-in is intended to help anchor performance-based, sustainability-oriented design more firmly in everyday practice.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.2 Urban Massing
After analysing the site, the scale shifts: from the place to the spatial figure. What is asked for now is the volumetric organisation of density, height, open space and spatial edges. Digital tools promise to support that step algorithmically.
In recent years many systems have appeared that generate urban options from defined parameters. You select a typology — point, row or perimeter block — set densities, distances and degrees of openness, and the algorithm produces variants at speed.
What fully automatic generators cannot do. Urban design is not an optimisation problem. Spatial hierarchies, transitions between public and private, scale in context, sight lines, atmospheric quality — none of that can be fully parameterised. We have worked with fully automated generators and could not produce a reliable basis for design that way. Systems with no way to intervene manually produce results that are formally correct but unsatisfying in context.
Tools that combine algorithmic proposals with deliberate control work better. The decisive factor: massing has to be movable, stretchable, differentiable. The algorithm is a starting point, not a result.
After years of testing the conclusion is clear: the value lies not in automation but in speeding up the generation of options and bringing quantifiable parameters in early. Algorithmic systems widen the solution space. They do not replace judgement in design.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.2.1 TOOL: Hektar
Hektar stands for a consistently algorithmic approach to urban volume: few inputs, many variants, no manual rework.
The interface is correspondingly spare. You select a typology — point, row or perimeter block — and set densities and degrees of openness. From that the system calculates urban configurations and offers variants. The planner’s job is essentially to pick one.
In practice the tool runs into a clear limit. The generated massing cannot be substantially changed by hand. Shifting volumes, adjusting proportions, responding in a differentiated way to a particular urban situation — that is only possible to a limited degree. Hektar therefore makes visible where the limit of fully automatic systems lies. Urban quality comes from decisions about context that cannot be parameterised.
Reference: Sofia Malmsten.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.2.2 TOOL: Spacio
Spacio takes a generative approach to urban massing but goes considerably beyond plain blocks. The system works like a parametric modular kit: it produces volumetric configurations and articulates them in detail.
The starting point is a set of typological and geometric parameters that define the built structure. Unlike fully automatic generators, control is not limited to choosing among ready-made variants. Massing can be changed deliberately — heights, proportions and spatial arrangements can be adjusted and developed further.
The level of detail is particularly relevant. Roof forms can be varied, façades differentiated, balconies added, window patterns defined parametrically. The level of detail therefore goes beyond abstract volumes and approaches a pre-structured architectural expression.
We have used Spacio mainly in phases where the development of volume and its architectural articulation run in parallel — that is where the tool is at its most useful.
Reference: Franz Fosberg.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.2.3 TOOL: Urbanistic
Urbanistic is a comparable tool from the German start-up scene. The cloud-based platform connects interactive 3D city models with real-time analysis and checks against building regulations.
The software combines official geodata with modelling and drawing tools. Urban options can be produced, visualised and compared in a semantically enriched 3D environment. What distinguishes it from fully automatic generators: the interaction is dynamic. Regulatory checks, indicators such as the floor area ratio (GFZ) and the site coverage ratio (GRZ), and visualisations all respond in real time to changes in the model. Setback distances, fire service access and other regulatory requirements can be checked directly in the urban model.
Urbanistic therefore addresses not only the generation of volume but its assessment against concrete planning parameters. Conflicting objectives become visible early, and options can be compared on the basis of data. Building the regulatory check into the model shortens a step that otherwise happens afterwards and by hand.
Reference: Marc-Christian Hodapp.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.2.4 TOOL: Buildplace
Buildplace is a digital planning platform that connects spatial data, urban analysis and scenario modelling. It is aimed at municipalities, developers and asset owners. At its centre is a digital twin bringing together sites, buildings and infrastructure.
Users can build scenarios, compare development options and analyse the potential of areas. Spatial relationships and the consequences of planning decisions become visible early. Buildplace also offers advisory services — on municipal infill development, heat planning or assessing climate-relevant factors.
We reviewed Buildplace as part of our research but did not run a product test of our own. The tool appears in this chapter because it represents an approach relevant to municipal and strategic development — and because it differs from the other tools in working explicitly at a larger planning scale.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.3 Floor Plans and Schedules of Accommodation
Drawing a floor plan means deciding the inner order of a building. Functional requirements, circulation logic and spatial proportion all come together here. So far this work cannot be delegated satisfactorily — at least, we have found no system that does it convincingly. What can change is the speed at which we test options, and the method by which we keep an overview even under complex requirements.
The current landscape of tools shows two fundamentally different ways of thinking. One starts from a given building envelope and asks how to fill that outline as efficiently as possible. The other starts from a schedule of accommodation and develops the building form from it. Both share one central finding from our tests: fully automatic systems, in which we cannot intervene and moderate, have not held up.
The hot-wire cutter as a model. We use these systems the way we once used the hot-wire cutter. You model something, look at the result, form a first thought. You move things, correct, adjust. The advantage is speed. What would once have meant a day of CAD work can now be developed in two or three hours — far enough to go into a substantial conversation with the client.
That saving raises a structural question, though. If we reach a level within three hours that by HOAI standards sits somewhere in service phase 2, how can that credibly be tied to a fee that used to require several weeks of work? Processes and fee agreements have to adapt here — a question the profession has not yet answered.
Two project types, two logics. For housing projects a typical workflow can be sketched: the urban figure emerges quickly — five variants from a few polygons, check the building depths, done. For functionally complex buildings — nurseries, hospitals, schools — the logic changes. Here we start from the schedule of accommodation: what has to be arranged where and how, how are rooms reached, how does the internal organisation work? These functional briefs we have to draw and think through by hand. Anyone applying maximum AI as the measure here will be disappointed.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.3.1 TOOL: Finch
Finch follows a clear approach: fill an existing massing efficiently with units. The Scandinavian start-up has built a browser tool of its own that integrates into the CAD environment through a plug-in. We tested Finch connected to Rhino; the connection to Revit works as well.
The workflow is direct: you hand an empty building envelope from CAD to Finch and place an order — this many one-room flats, this many two-room flats. Finch calculates how those units can be distributed through the building and proposes how many circulation cores are needed.
One limitation must be named straight away: most unit generators work to an American philosophy, with a central circulation corridor. We delete it regularly and move the cores to where we actually want them. That manual rework is typical — and it is possible. That is what distinguishes Finch from purely automatic systems.
Individual units can be selected by right-click and filled with plans from a library. You have to build that library yourself, unless you pay for a considerably more expensive licence. That apparent disadvantage is in fact a mark of quality. Because you bring in and stretch your own plans, the results carry your own handwriting — you can say: yes, that is roughly how we would do it too. That is the precondition for showing a client something and saying: this is roughly where the journey could go.
The real value lies in certainty early on. In the past we would agree a residential building in the contract and then, in service phase 3, after weeks of modelling, hear: I do not like any of it. With Finch we can produce a reliable simulation within an hour or two and attach it to the contract: this is what we talked about.
Technical summary: plug-in for Rhino and Revit. Your own plan library required, or an expensive full version. Manual intervention possible in core positions and wall lines. Output quality: suitable for service phases 1 and 2, not for a planning application. Typical effort: two to three hours for the first reliable options.
Reference: Pamela Nunez Wallgreen und Jesper Wallgreen.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.3.2 TOOL: PlanFinder
Like Finch, PlanFinder runs as a plug-in directly in the CAD environment. You do not leave your familiar drawing and modelling environment but extend it with a functional layer. The decisive difference from external platforms: the geometry PlanFinder produces is real CAD geometry. Doors, walls, areas — not temporary visualisations but parts of the regular planning model, which can in principle be carried into later service phases.
In our practice we have a certain reservation about model geometry we did not author ourselves. Transferring concept ideas from such tools into a real planning model, from which the planning application is then drawn — that is a breaking point. Technically, though, reuse works considerably better with PlanFinder than with systems that run outside CAD.
The way of working resembles Finch: you define a unit, give it a geometric outline and basic parameters. PlanFinder recognises the unit and proposes an internal organisation. PlanFinder is a maximisation tool. The outer contour is fixed; the goal is optimal internal use. You think from the outside in.
Technical summary: plug-in for Revit and Rhino. Produces native CAD geometry, reusable in principle. Considerably more stable now than in earlier versions.
Reference: Jeroen van Lith.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.3.3 TOOL: Rayon
At first glance Rayon is an anachronism: a new 2D drawing tool at a time when everyone is talking about generative AI. A French start-up writing software for drawing floor plans — when we have had AutoCAD, Vectorworks and ArchiCAD for years.
The answer lies not in new functions but in how the software is built. Rayon was written from scratch, without decades of legacy baggage. It runs entirely in the browser, was designed for collaboration from the start, and has a fluidity that CAD systems grown over decades do not reach.
What Rayon does not have — and why that counts. The decisive thing is what Rayon does not have: no automatic form generation, no AI taking design decisions off your hands. You draw yourself. You think yourself. You decide yourself about proportion, spatial relationships and functional order.
That sounds banal. It is not. After half a year of AI experiments the question eventually arises: what are we actually doing here? Rayon is something like an AI detox moment. It is a reminder that certain things only work by hand — and only work if you have learned the craft.
In all our tests with automatic generators we found the same thing: there is no algorithm that recognises that a noisy junction calls for urban articulation. No system sees that a quiet south-facing side could form a natural point of calm. That is why the ability to draw floor plans and develop functional briefs remains a core craft skill.
Technical summary: entirely browser-based, designed for collaboration. Experimental AI elements present but optional. The focus is on drawing by hand, not on generative automation.
Reference: Shira Nathan und Stanislas Chaillou.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.3.4 TOOL: Snaptrude
We have followed Snaptrude on and off since its release. The American-Indian start-up has built a browser-based modelling environment that sits conceptually between SketchUp and BIM. The decisive difference from classic modelling tools: you do not primarily draw geometry, you draw rooms.
Every volume you create in three dimensions is automatically a spatial unit. Subdivide it and you get flats. Subdivide further and you get rooms. Alongside runs a tabular room schedule — and the two are linked in both directions. What you stretch, name or adjust in the model appears immediately in the table. What you change in the table takes effect in the model.
That opens a second workflow: a functional brief — one developed with ChatGPT, say — can be imported as a table and output as room volumes. You get virtual Lego bricks and stack them into a massing. It genuinely works in both directions.
In a further step the room volume model can be converted into a component model. Volumes are resolved automatically into slabs, walls and floors. Doors, windows and furniture can be added, and plans, sections and elevations generated. Snaptrude is not a replacement for Revit or ArchiCAD. It is a tool for the early to middle design phase.
Snaptrude positions itself between sketch modelling and BIM. Its real value lies in the planning logic it supports: not from the outside in but from the inside out. The schedule of accommodation first, then the massing.
Reference: Amritha Krishnan.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.3.5 TOOL: Hypar
Hypar comes from a very specific context: healthcare planning. The start-up was originally developed for hospital projects — and that shapes the approach fundamentally. You do not start with an empty envelope and fill it, but with a detailed schedule of accommodation from which fully equipped, functioning rooms emerge.
The workflow begins with a room table, typically in Excel. The list contains all the required rooms with areas and functional requirements. Hypar imports it and generates parametric room modules from it — not empty volumes but fully furnished and equipped rooms. An office already contains desks, chairs and workstations. A treatment room comes with medical equipment.
That level of detail has a clear purpose. In early discussions you can talk not only about abstract square metres but about actual usability. When the client asks whether a double bed with two bedside tables really fits in the room, you show it — here is the bed, here are the tables, here is the space left to move.
Its origin in healthcare explains that precision. Hospital rooms are heavily standardised, technically defined and functionally hard to interpret loosely. Hypar carries this principle over to other building types — and it works well for anything functionally determined.
We have not used Hypar on a project; we have not planned any hospitals in recent years. We looked at it and find the approach remarkable. The digital strength lies precisely there: in managing large amounts of data and linking function, equipment and room geometry consistently.
Reference: Ian Keough.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.3.6 TOOL: Qonic
Qonic is a Belgian start-up that has set out to rethink BIM software completely. Not to extend existing systems but to start from zero and ask: what would a BIM environment look like today if you built it without decades of legacy baggage?
The result is an entirely browser-based modelling environment without historically grown complexity. You open the browser, sign in, model. No local installation, no version conflicts, no licence servers. Models sit in the cloud and can be shared and worked on together without breaks between systems.
Conceptually Qonic is classic BIM. You model components, rooms, structures. There is no generative AI taking design decisions off your hands. Control over content and geometry stays entirely with the planner.
The advantage of starting again. Established BIM systems have built up, extended and adapted functions over years. That produces powerful tools — but also nested menus, overloaded interfaces and opaque dependencies. Qonic does not have that problem. A small team in Belgium can implement new functions faster, work more experimentally and respond more directly to changing requirements. We looked at Qonic and found it impressively fluid.
Is it already as mature as Revit or ArchiCAD? No. But the development points in a direction: BIM software does not have to be heavy, locally installed and complex. It can be lightweight, collaborative and web-based — and that matters particularly for smaller practices, universities and international teams.
Technical summary: entirely browser-based, no local installation. Cloud-based and designed for collaboration. Newly developed without a historically grown codebase. A classic BIM tool without generative AI.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.4 Sustainability Tools
Serious sustainability assessment usually happens only in service phase 3 — once a design exists and the energy consultant has a basis for producing a verification under the GEG, the German Buildings Energy Act. By that point, though, the essential decisions have long been made. The building form is fixed, the façade largely engineered, the building services planned.
The tools presented here move that insight forward. They allow well-founded forecasts at the start of a project — when nothing is fixed and everything is negotiable. That is the decisive gain: not better calculations, but earlier ones.
At heart these are forecasting instruments. You upload a simplified building geometry — elementary volumes are often enough. Then you answer a structured list of questions: which insulation thicknesses? Which façade? Which building services? District heating or a heat pump? Photovoltaics or not? The system calculates expected energy consumption, CO₂ emissions, running costs, investment costs. Every adjustment produces an updated assessment immediately.
That opens up an iterative space for decisions. You can talk to the client about sustainability targets at the start of the planning chain — not at the end of service phase 3. The basis of the argument changes fundamentally. Discussions become less speculative and more quantifiable.
Sustainability tools are not downstream checking instruments. They shift the centre of gravity of planning: sustainability moves from being something checked to being something steered.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.4.1 TOOL: Caala
CAALA is a Munich start-up with a clearly stated aim: to bring sustainability assessment into the design process as early as possible. The tool produces reliable forecasts of energy demand, CO₂ emissions and investment cost without a worked-out design.
The working basis is deliberately simple. No detailed BIM model, no engineered façade, no final building services design. A reduced volume geometry is enough — sugar cubes describing the building as a mass. You upload that and answer a structured list of questions. The system then calculates scenarios for primary energy demand, greenhouse gas emissions over the life cycle, and cost indicators.
The decisive gain is that you can iterate. Add photovoltaics, double the insulation, raise the window quality, and you watch in real time how consumption and embodied energy change.
We imagine — and are working towards — attaching a first CAALA assessment to the contract: this is the direction we are committing to together. Then we are not still arguing in service phase 3 about whether to retrofit a condensing boiler or a district heating connection.
CAALA is fully self-service. You go to the website, set up a project, upload the geometry, answer the questions, get a forecast. No consultant needed. That makes it accessible to smaller practices too.
Reference: Philipp Hollberg.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.4.2 TOOL: Metabuild
A Berlin start-up takes a more complex approach than CAALA. Instead of working with simple volumes, Metabuild allows a considerably more detailed digital twin of the planned building. The accuracy is correspondingly higher — at the cost of more organisational effort. So far the system works on a consultant model.
Two benefits count together. The first: Metabuild is the only tool that produces a cost forecast to DIN 276, the German cost classification standard, at the third level of its structure from a database query. The system takes the digital twin, hypothesises what kind of façade this must be and what it will cost — and gets fairly close to reality.
The second: the system runs comparisons of options systematically. An ETICS façade against a brick façade against a curtain wall façade — every option against every other. We worked through this with a client for the refurbishment of the Sony Center and arrived at design decisions on the basis of metric data that could not have been reached with conventional means.
The system sits between a concept tool and an instrument for strategic decisions. The decisive gain is that it couples sustainability assessment with a forecast of investment cost — not only CO₂ but cost at the same time; not only energy efficiency but economics at the same time.
Reference: Tariq Kaddoura.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.4.3 TOOL: Revalu
Revalu is a material database developed in Germany that records the sustainability-relevant characteristics of building products systematically. The focus is not geometric modelling but the basis for decisions at the level of materials.
Reliable life-cycle analysis depends on building materials being documented precisely with their environmental values — CO₂ equivalents, primary energy content, recyclability, ability to be dismantled. Revalu brings that together in one structured database. Building products are recorded with their sustainability parameters and can be brought directly into decisions.
We reviewed Revalu as part of our research but have not yet run a product test of our own. The tool is here because it addresses a gap the other sustainability tools in this chapter leave open: the material level. Without precise material data, life-cycle assessment at building level stays abstract.
Reference: Jacques Chevrant.
Seminar transcript February 2026, summarised by Claude, March 2026 – translation: Claude, August 2026.
C.3.5 Construction Tools
A great many new digital tools are currently appearing in construction, aimed at recording, documenting and evaluating the actual state of a site continuously. Behind this development lies the growing availability of inexpensive sensors, capable image processing and methods from artificial intelligence. Where sites used to be documented mainly through manual daily reports, individual photographs or spot surveys, new systems allow an almost continuous digital record of what is happening.
One central technical approach is reality capture. The term covers methods in which the real state of a building is recorded digitally by sensors and translated into spatial models. Typical methods are laser scanning (LiDAR), photogrammetry and camera-based images.
Seminar transcript February 2026, supplemented with internet research using ChatGPT, March 2026 – translation: Claude, August 2026.
C.3.5.1 TOOL: Lumoview
Lumoview offers a system for recording existing buildings digitally at speed. At its centre is the LumoScanner, a mobile device that captures interiors using LiDAR, infrared sensors and a 360-degree camera. A measurement takes only a few seconds per room and produces a comprehensive basis of geometric and visual data.
The recorded data is then processed automatically in the cloud. From it come point clouds, 3D models, floor plans and further building data such as room data sheets, elevations and sections. The models can be brought into CAD or BIM systems and serve as a basis for design, documentation or energy analysis of existing buildings.
Seminar transcript February 2026, supplemented with internet research using ChatGPT, March 2026 – translation: Claude, August 2026.
C.3.5.2 TOOL: Airteam
Airteam is developing a platform for surveying buildings digitally using drones and artificial intelligence. The approach is based on aerial images taken by a drone and then processed automatically in the Airteam Fusion platform.
From the image data the software generates precise 3D building models, measurements and further planning data. The models can be evaluated directly — to determine roof areas, component dimensions or volumes, for instance. That speeds up design and tendering processes and reduces errors in the measurement survey. Typical fields of use are roofs and façades, refurbishment, timber construction and the design of photovoltaic installations.
Seminar transcript February 2026, supplemented with internet research using ChatGPT, March 2026 – translation: Claude, August 2026.
C.3.5.3 TOOL: Oculai
Oculai uses camera systems and artificial intelligence to record construction processes automatically and generate site documentation from them. Cameras — often mounted on the tower crane — record what happens on site continuously. The video is analysed by AI algorithms that recognise activities, components and timings.
The recognised information is brought together automatically into a digital site diary. It contains working hours, recognised activities, construction sections and weather information, among other things. The aim is a continuous and objective record of the construction sequence. At the same time, progress, productivity and deviations from the programme can be analysed in real time.
Seminar transcript February 2026, supplemented with internet research using ChatGPT, March 2026 – translation: Claude, August 2026.
C.3.5.4 TOOL: Specter
Specter is a software platform for digital site management. It brings project information, programmes and models together in one environment and presents them through a visual 3D representation of the site.
One particular feature is AI-based voice control. Site managers can call up information, manage tasks or visualise the status of the project in the 3D model by voice. The platform aims at better coordination between everyone on the project and less organisational effort, through automated data analysis and an interface that is easy to use.
Seminar transcript February 2026, supplemented with internet research using ChatGPT, March 2026 – translation: Claude, August 2026.
C.3.5.5 TOOL: Kontekst
Kontekst provides a platform for documenting and communicating site information in a structured way. Its focus is linking photographs, video or notes to a spatial or project-specific context.
The application lets you record observations, defects or progress directly on site and assign them to the relevant components or drawing revisions. That produces a traceable record of what happened on site, available digitally to everyone on the project. The approach aims to reduce loss of information between site, design and project management.
Reference: Luk Vermeulen.
Seminar transcript February 2026, supplemented with internet research using ChatGPT, March 2026 – translation: Claude, August 2026.
C.3.6 Productivity Tools
Alongside the technologies for recording and analysing what happens on site, a great many software solutions are appearing that automate administrative and organisational work in everyday planning and project management. Where digital tools long concentrated on single functions — tendering, awarding and invoicing, programming, document management — a new generation of integrated platforms is now emerging. These systems combine project management, documentation, communication and data analysis in one digital working environment.
Many of these tools have become considerably more mature. Cloud-based platforms now allow a continuous flow of data between office, design and site. At the same time more and more applications use methods from artificial intelligence to structure, evaluate or summarise large amounts of project information automatically. The aim is to reduce the heavy administrative load in design and construction and give everyone on a project more time for professional decisions.
Seminar transcript February 2026, supplemented with internet research using ChatGPT, March 2026 – translation: Claude, August 2026.
C.3.6.1 TOOL: Phase 0
Phase 0 (formerly Compa) is web-based software for architectural and engineering practices that brings cost planning, tendering, awarding and invoicing together with project monitoring on one platform. The system is entirely cloud-based and can be used in the browser without installation.
Its central functions include producing bills of quantities, cost tracking to DIN 276, invoice checking and fee invoicing under the HOAI. The platform also integrates tools for site supervision and project communication, such as digital site diaries, defects management and construction programmes.
One new approach is a built-in AI assistant: photographs, voice messages or text from the site can be captured through messaging services and turned automatically into structured site reports.
Seminar transcript February 2026, supplemented with internet research using ChatGPT, March 2026 – translation: Claude, August 2026.
C.3.6.2 TOOL: Alago
Alago is a platform for organising project information in construction automatically. The approach is to evaluate unstructured documents, notes or minutes using artificial intelligence and produce structured project information from them.
The software thus works as a kind of “data layer” for construction project management. Decisions, tasks and approvals can be documented automatically and traced at any time. An AI-supported search also gives quick access to the relevant information in minutes and project documents.
Reference: Ann-Christin Gah.
Seminar transcript February 2026, supplemented with internet research using ChatGPT, March 2026 – translation: Claude, August 2026.
C.3.6.3 TOOL: VOUND
VOUND develops software that uses artificial intelligence to support design and office work. The approach is aimed particularly at planning practices that have to manage large amounts of project data, documents and correspondence.
The systems analyse project information automatically and present it in structured form. That makes it possible to search documents faster, to see connections between different pieces of project information, or to automate office workflows.
Reference: Loren Gütinger.
Seminar transcript February 2026, supplemented with internet research using ChatGPT, March 2026 – translation: Claude, August 2026.
C.3.6.4 TOOL: Cosuno
Cosuno is a cloud-based platform for organising tenders and working with subcontractors. Its focus is the digital handling of procurement between designers, main contractors and the firms carrying out the work.
The platform allows bills of quantities to be produced and distributed, tenders to be submitted by subcontractors in a structured way, and the incoming tenders to be evaluated. A central digital platform is meant to make tendering more transparent and efficient. At the same time it creates a structured basis of data for cooperation between construction companies and their subcontractors.
Seminar transcript February 2026, supplemented with internet research using ChatGPT, March 2026 – translation: Claude, August 2026.