Heritage Informed August 2026

This edition leads with the first part of a two-part special feature on AI in the built environment: where the sector actually stands, why the reluctance to talk about it openly has the risk backwards, and what the professional bodies have and have not said. Part two follows next month. The rest of the digest picks up the same thread from the practical end.

We look at the two government tools now in circulation, at the council that submitted a heritage impact assessment describing a building 195 miles away, and at what the Planning Inspectorate expects you to declare. We also cover local lists and non-designated heritage assets, and note that the revised National Planning Policy Framework has still not appeared: the December 2025 consultation closed in March, so chapter 16 of the December 2024 Framework remains the policy in force.

Plus our project spotlight: a double consent at a Grade II listed hall in the Stour valley.

AI in the Built Environment: the Conversation We Are Still Too Embarrassed to Have

Those of us old enough to remember the dot-com boom, and the digital revolution it set in motion, will be no strangers to another technology revolution now working its way through the built environment. And much like an earlier moment in tech history, honest use of the new tool once carried a certain reluctance to admit it. There was a brief window, in the years before Tinder, when telling your friends how you’d actually met your partner felt like something you’d rather not get into, evidence perhaps that you couldn’t manage things the old-fashioned way. That stigma faded remarkably fast once it went mainstream. By the mid-2010s, for a generation who’d grown up online, it wasn’t just accepted, it was simply how people met. One British research participant, reflecting on getting the app a few years after it launched, put it plainly: she’d once thought dating apps were “for saddo people,” and then, “everyone seems to be on it.”

AI in professional practice carries a similar whiff of embarrassment for some, an unspoken sense that reaching for the tool is an admission that judgement or expertise needed a crutch. That instinct misunderstands something the built environment should recognise better than most industries: it has never stood still. Pen gave way to CAD. Handwritten site notes gave way to the typewriter, then the word processor, then the PC. Drawings once traced and copied by hand are now issued as digital files, and paper copies, once the only record a practice could rely on, are steadily giving way to digital storage, better for retrieval, better for the environment. Client files that once filled cabinets now live on a server. None of these transitions were failures of professional integrity. They were the profession doing what it has always done: absorbing a new tool, arguing about it for a while, and then getting on with the work.

Where the sector actually stands

The pace of adoption varies sharply by discipline. RIBA’s AI Report, the third in the series and published in late July 2026, found 74% of UK architecture practices now use AI on at least some projects, up from 59% a year earlier and 41% in 2024. Use is concentrated in early design visualisation, regulatory compliance checking and practice management. Three quarters of users report a productivity improvement, though fewer than one in six think their designs are actually better for it. On the planning side, government has moved just as fast: MHCLG’s Extract tool is now available to every English council and is expected to save the average authority around 255 hours of manual document work, while an AI prototype trialled with three councils aims to halve householder decision times. RICS’s own research, by contrast, finds 45% of construction and property organisations report no AI use at all, with fewer than 1% having fully embedded it. Heritage consultancy sits closer to that cautious end of the spectrum than to architecture’s, and there are genuine reasons for that caution, alongside genuine reasons to engage properly rather than quietly.

The stigma has it backwards

Here the dating-app comparison breaks down, usefully. Undisclosed AI use in a professional submission now carries a real liability exposure that undisclosed online dating never did. Two rulings, Harber v HMRC in 2023 and Ayinde v LB Haringey in 2025, found fabricated case citations presented as genuine legal authority, apparently AI-generated and never checked. The Planning Inspectorate expects AI use to be declared in casework evidence, and a costs decision of 2 January 2026 on an appeal at Corsham (APP/Y3940/W/25/3370482), involving a conservation area and listed buildings, recorded an Inspector’s “serious concerns” that a statement of case had been produced using AI. He gave the party the benefit of substantial doubt and made no finding on the point, but noted that undeclared use would amount to unreasonable behaviour. What was once a credibility risk to disclose has become, in the space of about two years, a credibility risk not to.

Where the professional bodies stand

The response from professional bodies has been uneven, and the pattern is telling. The bodies furthest from heritage practice have moved fastest and most formally:

• RICS has gone furthest: a mandatory global standard, in force from 9 March 2026, putting “the surveyor’s judgement, rather than the AI system’s output, at the heart of professional practice.”
• RTPI published a Practice Advice Note in March 2025, advisory rather than mandatory, its position being that AI must “support, not replace” professional judgement, and will require an AI self-declaration on its Assessment of Professional Competence from November 2026.
• CIOB published an AI Playbook in July 2024, framed around a practical “white box” principle: understand how an AI system reached its output, don’t simply trust it.
• RIBA has an annual AI Report and an Expert Advisory Group monitoring the field, but no binding standard yet.

The three bodies most directly relevant to heritage practitioners, IHBC, CIAT and CIfA, currently have no formal AI policy at all. For a discipline built on documentary rigour and defensible judgement, that is a gap worth naming rather than quietly working around.

So what is AI actually for?

Ask what AI is for and you’ll get answers shaped by who’s asking. Speed up process. Reduce headcount. Provide better documents. Enhance knowledge. All of these are true in places, and none is quite the whole answer, and it’s worth being honest that cost and speed tend to lead the decision to adopt, not knowledge or quality.

Here’s a view from the tool’s side of the conversation, for what it’s worth. AI’s most defensible purpose in this industry isn’t replacing judgement, it’s extending attention. A planner, architect or heritage consultant has a fixed amount of time and a finite capacity to hold detail in mind. AI can hold more of the routine material, the archive, the policy history, the site context, so attention gets spent on the decision rather than the retrieval. Used that way, it enhances knowledge and produces better documents almost as a side effect. Used purely as a headcount-reduction tool bolted onto the same old workflow, it risks quietly eroding the very judgement the profession depends on, judgement that was built, historically, by doing the retrieval work now being handed over.

Where does the built environment actually sit compared with other professional sectors doing this same reckoning, and what tools out there are genuinely AI rather than just automation wearing an AI label? That’s a longer conversation than this piece has room for.

Part Two next month

Extract and Augmented Planning Decisions: What the Two Tools Actually Do

The feature above refers to both government tools in passing. They are worth knowing by name. Extract, built by the Ministry’s Digital Planning programme with the government’s Incubator for AI, became available free to every local planning authority in England on 4 June 2026. It converts scanned planning documents and historic maps into standardised, machine-readable data, reducing a task that could occupy an officer for up to two hours to an average of around two minutes. The datasets it targets are precisely the ones heritage practitioners depend on: conservation area boundaries, Article 4 directions and Tree Preservation Orders, much of it currently held as scanned PDFs and paper files organised differently by every authority.

Augmented Planning Decisions is the second and the more consequential. It is a prototype funded through an £8.2 million Ministry contract with Google Cloud, Google DeepMind and delivery partner Faculty, with alpha trials running since May 2026 at Barnet, Camden and Dorset. The tool analyses a householder application, surfaces the policies that apply and supports the drafting of the officer’s assessment. Every assessment is reviewed and approved by a qualified planning officer before a decision is issued. Subject to results, the Ministry expects to extend trials to up to ten further councils later this year, with national rollout from 2027.

An AI-Written Heritage Impact Assessment, and a Building 195 Miles Away

Exeter City Council was criticised in June after it submitted a heritage impact assessment, in support of an application to itself as local planning authority, containing two paragraphs that turned out to describe a different corn exchange altogether: one in Lincolnshire, some 195 miles away, taken from a newspaper article published in 2013. The borrowed passage referred to a local organ club and to a 1969 meeting of an urban district council that had been abolished in 1974.

The local civic society spotted it, objected, and observed that the statement appeared to have been written by AI and needed revision, adding that it was surprised the application had been signed off and submitted by the council in the first place. A replacement statement was uploaded to the planning portal with the misplaced history removed. The application, to cover deteriorating upper-storey window frames with hoardings on safety grounds, was subsequently approved.

Declaring AI Use in Casework Evidence

The declaration requirement noted in the feature above is worth setting out in detail, because what has to be declared is narrower than most people assume. The Inspectorate’s guidance sets what it calls a golden rule: you may use AI to help prepare what you send, but you are responsible for making sure the result is accurate and appropriate. Where AI has been used to draft or substantially rewrite text, to produce a summary or an analysis, or to generate or alter images or video, you should say so. A short note in the covering letter, the statement of case or the proof of evidence is sufficient. Help with formatting and presentation does not need to be declared.

This sits alongside a tightened appeal regime. For applications dated on or after 1 April 2026, most appeals follow the part 1 written representations procedure, under which the inspector works largely from the documents that were before the local planning authority. New evidence is admitted only in narrowly defined circumstances, and late material can attract an award of costs. The Inspectorate has confirmed it is watching the roll-out of AI tools in local authorities closely and intends human control and professional responsibility to remain the rule on both sides of the process.

Local Lists Back in Focus: Non-Designated Heritage Assets

Historic England has issued a revised version of its local heritage listing advice, supporting communities and local authorities in setting up a local list or reworking an existing one. Coverage across England remains patchy, and the quality of the criteria used varies considerably between authorities. The draft Framework published in December 2025 would go further still, requiring councils to prepare local lists of non-designated heritage assets and expecting heritage impact assessment where such assets are affected by development.

It is worth being precise about status. That draft is not adopted policy. The consultation closed on 10 March 2026 and the revised Framework has not been published, so the current position remains chapter 16 of the December 2024 Framework, read with sections 16, 66 and 72 of the Planning (Listed Buildings and Conservation Areas) Act 1990 where designated assets are involved.

New From Historic England This Summer

Three publications worth downloading. Assessing Damp in Historic Buildings (13 July 2026) is a practical guide to investigation and diagnosis, and useful where damp is being cited to justify interventions a building does not actually need. Control of Microorganisms in Historic Buildings (21 July 2026) sets out approaches to UVC irradiation treatment on inorganic materials. And Delivering Clean Power: Commercial Renewable Energy Development and the Historic Environment, Advice Note 15, revised on 30 June 2026, covers the effects of terrestrial and marine renewable energy proposals on the historic environment: the document to have to hand for any solar or wind scheme near designated assets. All free to download at historicengland.org.uk.

Project spotlight: Double consent at Braham Hall

On 14 July, Babergh District Council granted both planning permission (DC/26/01713) and listed building consent (DC/26/01714) for the erection of a rear extension and internal alterations at Braham Hall, Brantham. The building is Grade II listed and, on the evidence of its list entry, a house of several builds: timber framed and plastered rear ranges of the sixteenth century or earlier, refronted in gault brick in the eighteenth or nineteenth century. It was the seat of the Braham family, and Thomas Tusser, author of Five Hundred Points of Good Husbandry, is believed to have lived and written there.

The route to consent was not a short one. Our involvement began with a site inspection in October 2025 and a Heritage Impact Assessment that went well beyond a desk exercise, incorporating a fabric assessment and an analysis of the timber frame to establish what was original, what had been altered, and where the significance of the building actually lay. That evidence base earned its keep, because the scheme went through revision and resubmission before the conservation officer was content.

What secured the approvals in the end was not an argument won at the eleventh hour but a design that moved. Working with the project architect, the proposals were amended in response to the conservation officer’s concerns, and the Heritage Impact Assessment was then updated so that the assessment described the scheme as revised rather than the drawings it had superseded. That last step sounds administrative. It is not. Applications regularly stall because the heritage document describes a building operation that has since changed shape, leaving the officer to reconcile two accounts of the same proposal.

The conditions attached are the ones a building of this age should attract, and they are instructive. Materials, fenestration, eaves and abutment details, internal doors and external vents all have to be approved before the relevant works begin. Adjacent fabric is to be made good in matching materials, forms and finishes. Most tellingly, the council asks that lost timber framing be accurately reinstated by reusing the mortices left in the surviving midrail, with oak studs matching the adjacent framing. None of that obstructs the scheme; all of it is discharged on detail, and every one of those conditions is answerable because the fabric had already been recorded. Both the Heritage Impact Assessment and the separate fabric assessment are listed among the approved documents. The project architect’s own verdict on the decisions was that they had arrived without onerous conditions, and that our input had been crucial in getting there.