TL;DR: AI use in UK architecture practices jumped from 41% to 74% between 2024 and 2026, but RIBA’s latest survey shows the gains haven’t reached fees, design quality, or confidence in the profession’s future. A separate survey of 230+ UK practices, run with RIBAJ, shows why: the worst-rated problems in day-to-day practice are financial, recovering costs, forecasting fees, and most AI use still isn’t pointed at them.
RIBA’s 2026 AI Report is out, and it looks like the use of AI in architectural practice has gone from a side experiment to something that nearly three-quarters of UK practices now use on at least some jobs.
The report delved into the findings: productivity is up, fees aren’t. Only 17% of architects think their designs are better because of AI, and barely a fifth feel more optimistic about where the profession is heading.
We think the survey only paints half a picture. RIBAJ and Hyphen Digital asked over 230 UK practices where their actual pain points live day-to-day, and the answer wasn’t design. It was money: recovering costs, forecasting fees, knowing whether a project actually made a profit. Those are the problems AI adoption has barely touched so far.
Here’s what both surveys say about AI in architecture practice and where the real opportunity sits.
How many UK architecture practices are actually using AI in 2026?
74% of UK architecture practices now use AI on at least some projects, up from 59% in 2025 and 41% in 2024, according to RIBA’s 2026 AI Report. But adoption is shallow: only 31% use it on every or most projects, and just 15% call it embedded in their day-to-day work.
That’s a substantial jump. Two years ago, most practices had never touched AI on a live project. Now most have. Larger practices are well ahead, with over 90% of those with 50+ staff using AI compared with 63% of practices with under 10 people. Adoption also doesn’t track experience the way you’d expect: architects in their first five years and their 25th to 30th year use AI at almost the same rate, around 80 to 85%. It’s only architects past 30 years in the profession who lag, at 59%.
The bigger tell is the maturity figure. 62% of practices describe themselves as still exploring AI rather than using it properly. There is a big difference between trying something once and building it into how you work on a day-to-day basis. Changes to processes and workflows take time and effort.
Is AI actually making architecture practices more profitable?
Not yet, for most. 73% of AI users report a productivity improvement and 57% report a positive return on investment (ROI), but only 7% say AI has allowed them to increase project fees, and just 17% believe their designs are better because of it. Optimism about the future hasn’t moved either.
Productivity and profitability aren’t the same thing, and this is where the two come apart. Architects using AI are genuinely working faster: three in four say so. But speed only turns into margin if a practice can capture it, either as additional fee income or as extra capacity for other paid work. Right now, 58% of AI users disagree that it’s led them to increase fees.
It’s not just a UK pattern either. A global survey of nearly 800 architects and designers by Chaos found that 61% feel pressure to adopt AI despite their own reservations, and only a minority are more than “somewhat satisfied” with what it produces.
Confidence in the work itself is thin too. 77% of RIBA respondents agree AI can never replace human creativity, but only 30% think it enhances creativity, and just 17% think their own designs are better for using it. Optimism about the future of the profession sits at 17-21%, while roughly half actively disagree.
Why the confidence gap? What RIBA’s data doesn’t tell you
RIBA’s own breakdown of AI use gives a clue. The most common application by far is early-design visualisation, at 56% of AI users, followed by standards and compliance checking at 40% and general practice management at 32%. Whilst that’s useful work, AI doesn’t touch how a practice actually makes money: fee setting, cost recovery, project profitability.
RIBA President Chris Williamson made a related point in the report’s foreword, noting that “professional oversight of design remains essential” if tomorrow’s buildings are going to be safe and sustainable. That’s true of design. It’s just as true of the fee proposal, the contract, and the numbers behind them, and that’s the oversight nobody’s talking about yet.
So we asked. RIBAJ and Hyphen Digital surveyed over 230 UK architecture practices on the business side specifically: fees, scope, cash, the parts of running a practice that don’t show up in a design portfolio.
The real pain points AI isn’t solving and what they’re costing you
If you asked architects to rate what’s hardest to keep on top of, and design doesn’t feature. In the RIBAJ/Hyphen Digital survey, architects rated recovering out-of-scope costs their single worst problem, 5.2 out of 7, ahead of forecasting fee income (4.9) and tracking billable hours against budgets (4.2). Producing client reports barely registered at 2.5. Practices aren’t overwhelmed by admin. They’ve lost sight of where the money’s going.
55% of practices say more than a quarter of their projects run over budget or over time, and 28% say it happens on half or more of their projects. Most only find out after the fact: 32% catch it during a manual review, 24% don’t spot it until they’re invoicing, and 8% don’t find out until the project’s finished. Just 22% get an automatic flag.
Working out whether a project was actually profitable is just as messy. Only 41% call it “very easy” once a project ends; the rest either piece it together manually or find it genuinely difficult. 36% aren’t confident their practice recovers all the fees it’s owed in the first place, a figure that echoes RIBA’s own finding that only 7% of AI users think it’s improved their fees.
Governance hasn’t caught up either. Only 19% of practices have a documented AI policy, and 61% agree AI will make it harder for early-career staff to gain the skills they need, since the tasks AI automates best are often the tasks juniors learn from. Adrian Malleson, RIBA’s Head of Research, points to unresolved questions around liability, intellectual property and professional indemnity insurance that regulators and insurers haven’t caught up with yet. UK insurers aren’t yet applying AI-specific exclusions, but they’re starting to ask about AI use before they’ll quote. A practice that can’t manage scope creep properly isn’t just leaving fees unclaimed. It’s carrying risk nobody’s priced.
Where should architecture practices actually be investing?
Not primarily in more AI tools. 68% of practices still run finance and cost control on spreadsheets, and only 20% use AI specifically for that job, even though it’s the practice’s worst-rated pain point. 78% plan to increase tech investment over the next two years, and the barriers are cost and cash flow, not doubt about the value.
45% cite the cost of technology and 36% cite cash flow as the main barriers to further investment, compared with just 28% who cite doubt about ROI. Cost is holding practices back, not scepticism.
Right now, that spending mostly goes to visualisation and drafting tools, because that’s where the AI hype is. It should be going toward fixing revenue leakage from unlogged time and uncaptured variations (scope changes that go unbilled), and giving every live project real-time WIP visibility (work in progress, the gap between fees earned and fees actually invoiced), so a practice catches lost margin while there’s still time to fix it, not three months after the project’s closed.
How do you close the AI confidence gap in your own practice?
Start with visibility, not more tooling. Give every live project real-time fee recovery and WIP tracking so overruns get caught mid-project rather than at invoicing, where 24% of practices currently catch them. Write a short AI use policy. Then treat AI and practice-management systems as two separate investments solving two different problems.
Rejecting AI isn’t the point. The productivity gains are real; three-quarters of practices report exactly that. But without financial visibility, that speed just moves a practice faster toward the same margin problem.
The practices that stay profitable do a few things consistently: plan capacity properly instead of assuming everyone bills 7 to 8 hours a day, price fee proposals off real delivery data, capture variations as they happen rather than writing them off, and review WIP monthly rather than at year-end. What makes profitable practices different comes down to operational discipline, not superior use of AI. It comes from a system that makes the numbers visible while there’s still a project left to save, not from more AI.
Fix the foundation first
AI adoption in architecture practice has nearly doubled in two years, and that’s not slowing down. But RIBA’s own data shows that speed hasn’t translated into fees, better designs, or confidence about what’s next, and the RIBAJ/Hyphen Digital survey shows why: the problems costing practices the most are financial, not creative, and most current AI use doesn’t address them.
Keep using AI. Just fix the practice-management foundation underneath it first, so the productivity gains actually show up in margin instead of leaking straight back out. If you’re not sure which of your systems are worth investing in next, book a project management discovery call and we’ll help you work out where the real return sits before you spend on more software.
