Here's the rewrite in the RISE voice:
AI in Architecture: What It Actually Does and What It Doesn't
There's a version of the AI in architecture conversation that sounds like a press release: generative design unleashing creativity, predictive analytics transforming decision-making, construction site monitoring revolutionising safety. All of it technically true, none of it particularly useful if you're trying to understand what these tools actually mean for the buildings that get designed and built.
Here's a more grounded account of where AI is genuinely useful in architectural practice, where its limitations are real, and how we think about it at RISE.
Energy performance modelling on screen alongside hand drawings and a physical scale model in the studio. The tools change. The judgement required to use them well doesn't.
Generative Design
Generative design uses algorithms to explore a solution space defined by parameters the designer sets: spatial requirements, structural constraints, material preferences, energy performance targets, cost limits. The software generates multiple design options within those parameters, far more than a human designer could manually produce, and ranks or filters them against the defined criteria.
The useful thing about this isn't that it replaces design thinking. It's that it expands the range of options tested before a direction is committed to. On a complex massing problem, or a structural optimisation challenge, or a question about how to maximise daylight across a floor plate within planning constraints, generative tools can surface solutions that wouldn't have been reached through conventional iterative design.
The less useful version of this is treating the output as a finished design. Generative tools are good at optimising against quantifiable criteria. They're not good at the judgements that make architecture worth building: spatial quality, material character, the way a building responds to its context, the atmospheric qualities that determine how a space actually feels to inhabit. Those still require human judgement, and they always will.
We use generative approaches selectively, for specific problems where the parameter space is well-defined and the optimisation criteria are clear. We don't use them to generate design directions.
Energy Modelling and Performance Simulation
This is where computational tools, including increasingly AI-assisted ones, are most directly relevant to the way we work at RISE.
Energy modelling software allows us to test a building's thermal performance before it's built: to understand how heat moves through the fabric, where the losses are, how different glazing configurations affect daylight and solar gain, what the effect of shading devices will be across the year. We use the Passivhaus Planning Package for this on projects where we're targeting meaningful energy performance, and the results directly shape design decisions at concept stage rather than being a compliance check at the end.
AI is beginning to accelerate parts of this process. Machine learning tools trained on large datasets of building performance outcomes can provide faster initial performance estimates, which is useful in early-stage option testing where the full modelling workflow is too slow for the pace of design development. They're less reliable for detailed technical analysis, where the specificity of the building's geometry, materials, and construction matters too much for generalised models to handle accurately.
The bigger shift here isn't AI specifically. It's the broader integration of performance analysis into the design process from the outset rather than treating it as a separate technical exercise that happens after the design decisions have already been made. AI tools accelerate that integration. The underlying methodology is what matters.
Construction Site Monitoring
Sensors, cameras, drones, and increasingly computer vision are being used on construction sites to monitor progress, identify clashes between different trades, detect safety issues, and provide project managers with real-time data about what's actually happening on site versus what the programme says should be happening.
On large, complex construction projects this is genuinely valuable. The gap between programme and reality on major construction projects is a persistent and expensive problem, and tools that make that gap visible earlier, rather than when it's already caused delays and cost overruns, have obvious utility.
For the residential and community projects we work on at RISE, the scale doesn't yet justify the infrastructure investment these tools require. But the underlying principle, that better real-time information about construction progress improves decision-making and reduces the cost of problems, is one we apply through conventional means: regular site visits, systematic inspection reports, and a continuous record of instructions and decisions that maintains clarity about what was agreed and when.
Operation and Maintenance
Buildings fitted with smart sensors and connected systems can feed performance data back in ways that weren't possible with conventional construction. Occupancy patterns, energy consumption, air quality, temperature distribution, equipment performance: all of this can be monitored continuously and used to optimise how the building operates.
For the buildings we design, the most relevant application is post-occupancy monitoring of energy performance. There is a well-documented gap between the energy performance modelled at design stage and the performance achieved in use. Understanding that gap, through measured data rather than assumed outcomes, is essential for improving the reliability of design predictions and for demonstrating to clients that the performance claims made at design stage are borne out in practice.
This is also where the relationship between AI and sustainable design becomes most tangible. A building that generates and responds to data about its own performance can be operated more efficiently than one that runs on fixed settings regardless of actual conditions. The potential for ongoing energy savings, without any change to the building's fabric, is real.
Where AI Doesn't Help
It's worth being direct about what AI currently can't do in architecture.
It can't make good design judgements. The qualities that make architecture worth building, spatial intelligence, material honesty, contextual sensitivity, the capacity to produce a building that people feel genuinely well in, are not reducible to optimisable parameters. They require the kind of integrative human judgement that comes from years of designing, building, and inhabiting spaces.
It can't substitute for construction knowledge. A generative tool that produces a structurally novel form without understanding how it would be built, or at what cost, or whether the contractor exists who could execute it, isn't useful. Architecture requires a continuous feedback loop between design ambition and construction reality that AI tools currently don't maintain.
It can't replace the architect-client relationship. The most important conversations in any project are the ones about what a client actually needs, as distinct from what they initially ask for. Those conversations require listening, experience, and judgement. They're not a data processing problem.
How We Think About It at RISE
We use computational tools throughout our work: energy modelling, BIM, structural analysis, daylight simulation. These aren't AI in the current sense of the term, but they're the digital infrastructure on which more sophisticated tools are building.
We're interested in AI applications that make performance analysis faster and more accessible at early design stages, that improve the reliability of post-occupancy monitoring, and that help identify coordination problems in complex construction packages before they become site problems.
We're sceptical of AI applications that position the tool as the designer, or that treat the architectural challenge as primarily a data optimisation problem. The buildings we most admire weren't produced by optimisation algorithms. They were produced by people who understood their sites deeply, listened carefully to their clients, and made thousands of well-judged decisions across the full span of a project.
AI will change how some of those decisions are informed and supported. It won't change what they fundamentally require.
If you're planning a project and want to understand how we integrate technology and design thinking, we're glad to have that conversation.
→ Email us at architects@risedesignstudio.co.uk
→ Or call the studio on 020 3947 5886
RISE Design Studio, Architects, Interior Designers + Sustainability Experts
☉ Architecture for people and planet
☉ Trading since 2011
☉ Company reg no: 08129708
☉ VAT no: GB158316403