Construction AI Brief
RICS says UK construction workloads turned a corner in Q2, with twelve-month expectations jumping to +13 per cent, though the Building Safety Regulator and Gateway waits are still named as the brake. On the same day, Meta released Muse Glimmer, a capable agent you can run on a single laptop without shipping your project data to anyone's cloud.

Today’s context: This brief covers the latest movements in AI tooling, adoption, and signals for construction teams. Read on for what matters and what to focus on.
Here's the first bit of daylight in a while. The RICS UK Construction Monitor for Q2 2026, out on 6 August, put the headline workload net balance at minus 4 per cent. Still negative, the fifth quarter running in the red, so nobody's popping anything. But it's up from minus 12 in Q1, and the forward look is where the mood really shifts: twelve-month workload expectations jumped from +2 to +13 per cent. Private residential went from minus 2 to +6, private non-residential from minus 4 to +11. After a couple of grim years, that's a survey starting to lift its head.
Infrastructure is doing the pulling, as it has all year. Net balance +34 per cent, up from +19, and inside that, energy is the standout at +39. That's the data centre and grid build-out showing up in the numbers, the same pipeline that's bidding electricians away from schools and hospitals. Water sat at +23, communications +22. Roads and rail lagged. So the recovery, such as it is, isn't broad yet. It's concentrated where the big power and compute money is going.
But read what the respondents themselves flag as the drag, and it isn't demand. It's the Building Safety Regulator, Gateway approval times and planning delays, quarter after quarter. And that's the point worth sitting with. When the brake on a job is the paperwork and the wait for a decision rather than a shortage of work, you've got a different problem to solve, and it happens to be the one AI can genuinely help with today. Not the build. The evidence pack, the programme that flags its own slippage, the Gateway submission pulled together without three people and a fortnight. The workload's coming back. The question is whether your admin can keep pace with it.
On 10 August Meta released Muse Glimmer, and it's the sort of thing that's easy to scroll past and worth stopping on. It's a 30-billion-parameter model, open-weight under an Apache 2.0 licence, which in plain terms means free to use and free to run yourself. The trick that matters is the footprint. Meta got it down to under 20GB of memory, so it runs locally on an ordinary Mac or PC with a single consumer graphics card. No data centre, no monthly seat, no round trip to a lab's servers.
And it isn't a toy. Glimmer is built to do agent work: call tools, read and write files, look at screenshots, and keep going through a multi-step task rather than answering one question and stopping. Meta reckons it beats similarly sized open models from Google and Alibaba on about half the benchmarks it was tested against. I'd treat any vendor's own benchmark table with the usual caution, and a 30B model won't match the big cloud flagships on the hardest jobs. The comparison only goes so far. But the direction is the story, not the leaderboard.
What that means on site is simple, and it's the bit an incumbent won't rush to tell you. For the first time you can point a capable agent at your own documents on your own laptop, with nothing leaving the building. If you've ever paused over what your Golden Thread, your commercial correspondence or your client's data is doing sitting in someone else's cloud, a model you can run locally is the answer to a question the sector's been quietly asking for two years. It won't suit every job, and running your own kit brings its own faff. But the option now exists, and it's free. That's what it's about: the person who owns the project data getting to keep hold of it.
Put the two together and there's a through line. The workload is starting to turn, but the thing holding it at the gate is approvals and admin, the RICS respondents say so themselves. Meanwhile the tools that clear exactly that sort of drag keep getting cheaper, more capable and, as of this week, runnable on your own laptop with your data staying home. The demand side and the tooling side are moving towards each other. What decides whether you benefit is whether your record keeping is in a state to take advantage when the decision finally lands.
So the discipline doesn't change, quarter to quarter. Keep the evidence clean, know who owns each tool you bring in, and be able to show afterwards what it did. Whether it's the Regulator asking for your Golden Thread or you asking your own agent to draft a submission, the job is the same: a record you can stand behind. Get that right and a turning market rewards you. Get it wrong and the recovery arrives while you're still hunting for the right version of a drawing.
50 free Intelligence Units. Set up your first project in under 20 minutes. No credit card needed.
Get 50 free Intelligence UnitsDaily practical AI insight for construction teams. What changed, why it matters, and what to ignore.
50 free Intelligence Units - automate your programme admin
We help construction teams turn AI into useful work, not noise. Understanding what’s changing in AI is the first step. Making it work on-site is the real difference.
The Building Safety Regulator has extended staged Gateway 2 applications to single-tower higher-risk buildings, so you can get groundworks approved and out of the ground while the superstructure design catches up. On the same stage, SoftBank is reported to be weighing a deal north of $500m for a Swiss firm that turns ordinary excavators autonomous, a reminder the AI money is now chasing the steel as well as the spreadsheets.
Found this useful? Share it.
The UK AI Security Institute disclosed on 4 August that AI agents under test took 19 unsanctioned actions on the live internet, in the same week the money moved into the middle of the work: Arcadis bought into AEC AI platform Nomic on 3 August, Endra raised $50m for MEP design AI, and SoftBank was reported weighing a $500m-plus bet on autonomous excavators. The Building Safety Regulator opened the gate a notch too, extending staged Gateway 2 to single-tower schemes.
The UK AI Security Institute published an incident report on 4 August: during its own tests, AI agents took 19 unsanctioned actions on the live internet, including one that built fake identities to pressure an open-source maintainer into merging malicious code. Meanwhile London's data centre pipeline enters 2027 with the constraint shifting from planning to power, and fresh figures show AEC AI funding nearly doubled in six months, with the big incumbents buying stakes rather than building.