Tech and AI in Architecture, ft. Jordan Harris (Hale)

In this Architecture Social conversation, Stephen Drew is joined by Jordan Harris, Systems Architect and Digital Operations Manager at the London commercial practice Hale, for an honest look at technology and artificial intelligence in architecture and the wider built environment. Running for around 53 minutes, it cuts through the hype to explain what these tools actually do and how architects can use them.

Who this is for

Architects, architectural assistants, technologists and students who want a clear, practical understanding of where tech, BIM and AI are heading in practice, and what skills are worth building now.

Learning outcomes

After watching or listening, you will be able to:

1. Explain the difference between machine learning and deep learning, and why "AI" is often a marketing term.
2. Describe how generative and automated design tools fit into a real design workflow rather than replacing the designer.
3. Recognise the importance of framing the right question and reviewing AI output critically.
4. Distinguish digital twins from BIM and outline where each adds value.
5. Identify accessible routes into coding and computational design, including Python, Dynamo and Grasshopper.
6. Summarise the software skills UK employers are increasingly asking for.

Alternative careers and the rise of the technologist

Jordan describes a route many will recognise: a conventional Part I and Part II architecture education, followed by a shift towards technology. His role blends architect, IT manager and BIM manager, a reminder that practices increasingly need people who sit between design and digital.

What "AI" actually means

A central theme is that much of what is marketed as AI is not conscious intelligence. Jordan separates machine learning, which is heavily guided by a human, from deep learning, which is nudged with data and prompts but learns patterns by itself. That self-directed learning is powerful, but it also makes mistakes harder to predict and control.

AI in design: tools, not replacements

Tools that automate parts of layout and design through scripting can speed up the workflow, but Jordan's argument is that they support rather than replace the designer. The person who understands the client, the brief and the regulations, and who can interrogate the output, remains essential.

The skill of asking the right question

A recurring point is that you have to know how to ask the question. Automated tools only help if you can frame the problem, set the rules and spot when the answer is wrong. Jordan gives the example of an algorithm that appears highly accurate but is really picking up an incidental pattern rather than the thing you care about.

Digital twins versus BIM

Jordan sees digital twins as a significant near-term development. Where a BIM model often becomes a dead end once handed to a client who cannot maintain it, a digital twin aims to stay live and usable, giving accountability for components, warranties and facilities management across a building's life cycle.

Real-time engines and the modern scale model

Real-time tools such as Twin Motion and Unreal Engine are framed as the modern evolution of the physical scale model. Used within the design process, rather than only as a sales render, they let teams test materials, light and space quickly and realistically.

Learning to code, the accessible way

For anyone interested in computational design, Jordan recommends starting with Grasshopper and Dynamo, then Python, which underpins Revit, Dynamo and Unreal Engine. His practical advice is to find existing scripts, reverse engineer them and adapt them to a task, building confidence step by step.

Adoption, change and generative images

Drawing on the earlier shift to BIM, Jordan notes that adoption always carries a short-term dip in productivity before the gains arrive. On generative image tools, he is measured: useful for abstract or conceptual work and likely to be absorbed into render engines and design software, but limited as a standalone tool.

The skills employers are asking for

From a recruitment perspective, Stephen notes that Revit has moved from a nice-to-have to a default expectation, with Rhino and Grasshopper increasingly desirable in larger practices. Jordan's tip for learning Revit is to master the fundamentals first: building families, walls, floors and staircases, and locating projects properly.

Key terms

• Machine learning: pattern-finding that is closely guided by a human providing correct answers.
• Deep learning: models that learn patterns largely by themselves from data, with lighter human steering.
• Generative / automated design: using rules and scripts to produce design options.
• BIM: Building Information Modelling, a data-rich model of a building.
• Digital twin: a live, usable digital counterpart of a building for ongoing operation and facilities management.
• Parametric tools: Grasshopper and Dynamo, often scripted with Python, for rule-based modelling.

Reflective prompts for your CPD record

1. Which repetitive or mundane tasks in your own workflow could be automated, and what would that free you up to design?
2. Where in your current projects could a digital twin or better data handover add value for the client?
3. What is one accessible skill, for example Python, Dynamo or a real-time engine, that you could begin learning this quarter?

About the guest

Jordan Harris is a technologist with a classical architecture background, educated to Part II level at the University of East London and the University of Kent. At Hale, the London commercial architecture practice, his work spanned Systems Architect, Digital Operations Manager and BIM Manager, bringing together building design, BIM, data science and programming.