Monthly Register — June 2025

The carpet at the end of the second afternoon tells you more than the programme did. Pipe-and-drape half struck, a blank foam-board leaning where a graphic had been, somebody's lanyard on the floor next to an empty counter and a water bottle. Digital Construction Week ran at ExCeL London on 4 and 5 June. NXT BLD and NXT DEV followed at the Queen Elizabeth II Centre on 11 and 12 June. Two London weeks in one month, and a fortnight after the second hall closed the sensible thing is to ask what is left of them.

Two things, I think. OpenBIM, and artificial intelligence applied to design and delivery. Neither was invented in June, and both got harder to dodge. The measure worth applying to either is whether it changes a published container or only a slide.

What moved

Take exhibitions as weather. Two of them in a London June produce the predictable systems: stages, demonstrations, a claim that the next generation of BIM has arrived, and a long queue for coffee. Weather is not the story. What the weather was about might be.

Start with openBIM in the narrow sense that matters on a job rather than in a logo. Can a party who was never in the authoring licence open the published set? And is that set the same object the job will plot from? A slide that says "open" while the issued set remains a native file plus a PDF pack has not moved the container an inch. An IFC — or whatever open schema the appointing party actually named — carrying revision, suitability, and enough geometry and property to be reused downstream is a different proposition altogether. Most live jobs still fail that test. The halls did not create the failure; they put it back in front of the people who could do something about it.

Artificial intelligence was the fortnight's other unavoidable sentence, and the category is wide enough to be nearly useless: design automation, drawings produced from a model, a chat window sitting on last week's export. The useful question is narrower, and it is the one the first-quarter print already failed. Does the tool read a register, or does it wrap a folder? Ask it to name the current issued set — these sheets, this model state, this revision, released to these companies, for this purpose — and watch whether it can, or whether it summarises whatever happened to be in the download. A new cover on an unofficial PDF set is not progress. It is a funded sentence.

Neither signal belongs to a single software house, which is worth saying plainly to anyone who came home with a shortlist. Both apply to any model platform, any common data environment, and any assistant bolted onto either. A London June counts as the month's weather because that is where the UK conversation occupied two buildings for a week each, not because one stage happened to be newer than another.

The pair of claims is what I would write down for June, then. Openness promised on one set of stages, intelligence promised on the other, and the issued set still a clerical job on the great majority of live work unless the published container itself changed.

What changed in the standards

No new ISO 19650 part in June. The guidance layer formerly issued as the UK BIM Framework — restated in May as the IMI Framework — put out no replacement text to mark the halls, and Part 6, on health and safety information, remains the January text.

What can be dated sits on the IFC side of the openBIM conversation. On 12 June, buildingSMART International published a note on four tracks of work with ISO on the IFC standard, formally the ISO 16739 series. The framing is careful, and worth reading in the original rather than in somebody's summary of it. Track 1 — an update of the 2021 collaboration document between ISO/TC 59/SC 13 and buildingSMART — is an editorial and procedural exercise already in progress, including the possibility of maintenance-organisation status for IFC. Tracks 2 to 4 cover modularisation of IFC starting with IFC 5, a proposed update to IFC 4.3 that does not retract the current Part 1, and an IFC 5 core begun as a separate ISO part so that the 4.x series is not withdrawn. Those are under discussion, in development, or being prepared as proposals. None of them is a new published schema, and none is a commencement.

That is the accurate scale of it, and the accurate scale is more useful on a job than the excitable version. "IFC 5 is here" is not a clause. A New Work Item Proposal that has not yet been raised is not a container a document controller can accept a delivery against. The sentence the note actually supports is this: the next generation of the open schema is being split so that current 4.x deliveries are not orphaned when 5 begins, and industry input is invited while the modules are still soft.

One thing to watch

Two weeks on from Westminster, what matters is not whether anyone remembers a stage, but whether the same two arguments are still running in the office: whether the published set is an open container a party outside the authoring licence can use, and whether the AI tool is reading that set or decorating last week's export.

Watch the word "open" especially closely. If it turns out to mean a dashboard over the same natives, the job has bought a view rather than a state. If the published IFC cannot be opened, or opens as stripped geometry with the properties left behind, the demo did not survive the plotter. If three applications each claim to be the open environment and the unofficial PDF set has not retired, June changed nothing where it counts. Apply the same reading to the assistants: one that cannot name revision and suitability is a cover sheet, and one that writes into a register it cannot later reconstruct has simply created another unofficial set.

Which suggests one thing to ask for. If you are shortlisting anything you saw this month, put the demonstration in your own data and ask whoever is selling it to produce the published set for a package as it stood on a date you choose — in a format someone outside your licence can open, with the revision and suitability attached. That is a procurement condition rather than a technical one, so it belongs in the appointing party's requirements, where it has teeth. It will slow a purchase down by a week. On the evidence of most stack reviews, that is a week well spent.