Get new posts
Talks and discussions on engineering, and engineers, in an AI age.
See what's onJoin the community for new pieces, the best of the discussion, and what is coming up.
The ground is moving and some of the cards are wobbling. But this isn't a collapse story: professions that adapt for their clients could come out stronger. Here's who is saying what.
The ground is moving under architecture, design and engineering, and some of the cards are already wobbling: how juniors learn their craft, how expertise reaches clients, what professionals are trusted to know. The professional bodies' own evidence says AI is already part of daily work, and almost nobody has the training for it yet. But this isn't a story about everything tumbling down. Professions that roll with the punches and use these tools for their clients' benefit could come out stronger, and that's the future we want to explore with them.
Some quick background, so you know why I care about this. I trained as an architect at the Bartlett, then went into practice in London, including at Edward Cullinan Architects (now Cullinan Studio), designing university buildings. That put me on the professional's side of the table: serving clients, understanding the contracts, managing the risks, and working within the constraints of real-world design. It matters, because as a professional you rarely get to do simply what you think is best. Much of your work is shaped by what your clients ask of you.
I then went back to the University of East London to study Computing and Design, exploring what we'd now call generative AI for architecture, long before it was really possible. That led to a post as senior lecturer in professional practice, and as professional studies adviser to students on their year out.
Professional studies often gets less love than the design studio, but it's where students meet that reality for the first time. You're still designing, just with the real world switched on. And it raises the bigger questions. What do clients know? What do we know that they don't? What do we claim to know? And what exactly are we professing?
Back then, the internet was already shaking those answers loose. Suddenly anyone could look up what used to sit behind the professional's door. AI is round two, and it goes further. The internet gave everyone access to knowing. AI is giving them access to doing. Richard and Daniel Susskind saw this coming back in 2015: their book The Future of the Professions predicted that increasingly capable systems, AI among them, would change how professional expertise reaches people.
There's a tension here I keep coming back to. Professions run on codes: codes of conduct, codes of practice, the way things are done. Innovators have always pushed against those codes, because doing something new usually means doing it differently. That tension isn't going away in the AI age. If anything, it's about to get louder.
This is the first post in an Agent Build Club series on the future of work in the professions that design and make the physical world. Rather than add another opinion, it starts with who is saying what. Here's the route. First, the evidence that these professions are already changing. Then why so few people are ready for it. Then what's really at stake: where the next generation gets its judgement from. After that, how I think professionals can get ahead of it, and where the series goes next, one profession at a time.
Across every design and engineering profession, the people who represent practitioners are now reporting AI as part of daily work, not a pilot.
Architecture. The RIBA's AI Report 2026, its third annual member survey, found the share of practices using AI on most projects rose from 59% in 2025 to 74% in 2026. Use has moved well beyond images: among AI users, 40% now apply it to regulatory compliance checking, 32% to practice management and 24% to specification or technical design. The RIBA is now writing an AI Overlay to the Plan of Work and a policy on responsible use. (Member survey, self-reported.)
Engineering. The IET's 2025 Skills Survey found 58% of UK engineering employers already use AI, though only 18% use it regularly. The Institution of Civil Engineers rewrote its Code of Professional Conduct in April 2026 and now says the profession can no longer defer the question of how AI reshapes the civil engineer's role.
Design and make. Autodesk's 2026 State of Design & Make: AI Pulse, drawing on 2,500 leaders in architecture, engineering, construction and manufacturing, reports 98% using at least one AI tool and 59% using or planning agentic AI within a year. Its separate jobs report finds AI roles in these industries up 147% in two years. (Vendor research: read alongside independent sources.)
Product and digital design. The AI in Design 2026 report from Designer Fund and Foundation Capital (906 designers, 60+ countries) found weekly AI use jumped from 54% to 91% in a year, with half of respondents having shipped AI-generated code. Figma's State of the Designer 2026 reports most designers working faster, but alongside new pressure on craft.
Electronics and chips. At DAC 2026, Synopsys, Cadence and Siemens all launched autonomous design agents that plan, run and verify parts of the chip design flow. Nvidia's Tim Costa summed up the industry's direction: the future of engineering is agentic.
The pattern is consistent. Adoption is no longer the question. How the work, and the worker, changes is.
The same sources that report rapid adoption also report that the people doing the work have not been prepared for it.
Put simply: the tools have arrived in practice before the training has arrived in education or CPD.
The deepest change is not to senior practitioners but to how the next generation learns. In every one of these professions, judgement has traditionally been built through routine early-career work: the drawings, calculations, checks and first drafts that AI now handles first.
This is the part I recognise most from professional studies. The year out, the logbook, the jobs nobody else wanted: that was where judgement was supposed to be built, and where people like me checked that it had been.
These are beliefs and early signals, not settled outcomes. But they point the same way: if the old apprenticeship of judgement is being automated, something has to replace it, and very little has been put in its place yet.
Most of the conversation so far sits in one of two places. There is the debate: reports, panels and position statements about what AI means for a profession. And there is tool adoption: firms buying software and hoping skills follow. Both matter. Neither, on its own, prepares a person to work differently.
Our view at Agent Build Club is that the professionals best placed to shape this future are the ones doing both at once: taking part in the debate and getting their hands on the work.
This is the principle behind our workshops. We start on paper, not on a screen: mapping the people, decisions and evidence in a real professional task before any tool is opened. The machine is good at generating many viewpoints and options; it is poor at choosing between them. So the machine diverges, and the professional converges. Architects will recognise this immediately: it is how option studies and the crit already work. Engineers will recognise it from design reviews and options appraisal. That is a skill that can be taught, and it is the start of a new route to judgement.
The professions that design and engineer the physical world are changing, ready or not. The people who get stuck in early, arguing about it and trying it out, are the ones who'll shape what comes next.
So this series is a promise. While plenty of people are frowning at AI adoption, I want to go and talk to the ones already doing things differently: the innovators, the experimenters, the rule breakers. I don't mean people ignoring the codes. I mean people who've noticed the codes were written for a different way of working, and who are finding new ways of doing, thinking and being professional.
This post is the trunk. Each branch takes one profession and asks three questions: what's changing, who's ready, and what it opens up.
I'm starting with the two I'm closest to. Architecture, because that's where I started, and in 2027 I'll be back teaching architecture students in the UK and Europe, this time teaching them context engineering: how to set up the brief, the sources and the viewpoints that an AI works from, before it produces anything. Electrical and electronic engineering, because I'm working with the University of Surrey's AI Innovation Accelerator alongside its EEE courses, and the question of what tomorrow's engineers need is very live there.
Curious about the future of engineering, and of engineers, in an AI age? We're running talks alongside the blog. Have a look and sign up on our Luma calendar. Come along and argue with us. And if you're one of the people doing things differently, I'd really like to hear from you.
