Sources and verification

One standard applies throughout: a source counts as verified only where its own text has been read. Search engines were used to locate candidate sources; a search result summary was never treated as the source itself.

A reference list tells you what an author says they consulted. It does not tell you how far to trust each line of it. Every entry below therefore carries a tier stating what was read — not how reputable the publisher is.

Each question in the assessment names the source it is drawn from. This is where that source is stated in full, with the clause. Sixteen of the twenty are anchored to a published standard or industry-body document whose own text was read. Four are not, and say so.

Until this pass was completed the criteria carried candidate anchors that had never been checked against the sources they name. That is recorded here because the difference matters to a reader deciding how much weight the instrument carries: a criterion described as anchored, but never checked, is an assertion about the literature rather than a use of it.

Read from the publishers’ official preview documents. Clause numbers and quoted wording below are taken from those texts.

Building information modelling, verbatim

ISO 19650-1:2018, 3.3.14 — “use of a shared digital representation of a built asset to facilitate design, construction and operation processes to form a reliable basis for decisions.” The standard attributes this to ISO 29481-1:2016, 3.2, modified: the word “object” was replaced with “asset”. Cite 19650-1 as the operative source and mention 29481-1 where the lineage is material.

Three requirement types the glossary once collapsed

ISO 19650-1 separates them by what each relates to: EIR (3.3.6) — “information requirements in relation to an appointment”; AIR (3.3.4) — “in relation to the operation of an asset”; PIR (3.3.5) — “in relation to the delivery of an asset”. An early draft of the toolkit’s glossary merged EIR and AIR into one entry. The plain-language gloss remains for the practitioner; the distinction is carried in the footnote.

The instruments behind claims that BIM is required. Tiers are mixed and are given per row. Where a primary text was obtained it is cited directly; where it was not, the row says so.

On the UAE specifically

BIM is mandatory in Dubai only, under Dubai Municipality circulars. There is no federal UAE BIM mandate, and the other emirates have not mandated BIM, although major projects have adopted it voluntarily. Any statement that “the UAE mandates BIM” is inaccurate; the accurate form names Dubai Municipality.

On the EU specifically

Article 22(4) is routinely miscited as an EU-wide BIM mandate. It is not one. It authorises member states to require BIM electronic tools, and where they do, contracting authorities “shall offer alternative means of access until such time as those tools become generally available.”

What the buildingSMART playbook does and does not cover

Global openBIM Mandates, 2025 Edition [12] carries chapters on Australasia, China, the Czech Republic, Denmark, Dubai, Japan, Poland, Singapore, Spain, Spain (Catalunya), the USA and South Korea. It has no chapter on Malaysia, the United Kingdom or the European Union, so those three rows remain on their per-jurisdiction secondary accounts. This is recorded because an earlier version of this register expected that one document would re-cite the whole table.

The register above records what this project cites. This section records what it contributes: measurements made by the authors on named models, using stated instruments, drawn from no external source. These are the entries for which this project is itself the citable authority.

Reproducibility of the scoring

A separate and stronger claim the instrument makes by construction: identical responses, core version and locale pack produce an identical score, because scoring is a deterministic function rather than a model output. This is what makes test–retest reliability claimable, and it is checkable by any third party from published versions.

Recorded so that they are visible rather than forgotten. Each is either softened in the published copy or awaiting verification, and none may be given a footnote until it moves out of this section.

This register is only worth publishing if it is allowed to overturn the toolkit’s own copy. Four live claims were found to be wrong during compilation and were corrected in the published pages.

C-1 — Malaysia mandate overstated as covering private projects

Was published as BIM being mandatory for Malaysian projects of RM 10 million and above, “public and private”, with LOD 500, PeDATA and SKATA handover requirements attached.

Finding. PK 1.15 is a Treasury circular issued by the Ministry of Finance. Its scope is Government development projects of RM 10 million and above. A Treasury circular governs government procurement and cannot bind private-sector projects; private adoption is encouraged, not mandated. The “public and private” framing traces to vendor marketing material, not to the circular. The LOD 500, PeDATA and SKATA requirements derive from separate CIDB guidance, and attaching them to the mandate conflated two instruments.

C-2 — EU procurement directive miscited by omission

Finding. Directive 2014/24/EU Article 22(4) is permissive, not mandatory. The page now states that the directive allows member states to require BIM tools for public works but does not itself require them.

C-3 — the Dubai position was two circulars out of date, and its two figures were read as eras

Was published as the position under Circulars 196 (2013) and 207 (2015), sourced to a law-firm guide, with 207 described as having “widened” 196.

Finding. The primary texts were obtained from Dubai Municipality’s own document server, and two separate defects appeared. First, two circulars had been missed: 9-1-1 (2021) restates the standing position and 9-1-2 (2023) adds a mandatory Open BIM model submission with new building permit applications from 1 January 2024. Second, the two threshold figures are not two eras: 9-1-1 sets them out as two work-scopes in force together — over 20 floors and 200,000 sq ft for architectural and mechanical work, over 40 floors and 300,000 sq ft for structural work. Reading 207 as having widened 196’s figures collapses a scope distinction into a chronology.

The central assertion — mandatory in Dubai only, no federal UAE mandate — was correct and is unchanged. A secondary account can be simultaneously accurate about the instrument it describes and years out of date about the jurisdiction.

C-4 — four criteria anchors attributed wording to standards that do not carry it

Found by the verification pass in section 02. All four are of one kind: the anchor named a real standard but described it in words the standard does not use, or attached content to it that belongs elsewhere. None was a fabricated source; each was an imprecise attribution, and imprecision in a citation is what the tier system exists to expose.

The change was to four label strings. No question wording, no response option and no score was touched.

What this instrument is positioned against. Six of the seven sources below were obtained and read in full; the one that could not be is recorded at T2 with its content unclaimed. An earlier version of this register reported this material from search summaries, which §01’s own standard forbids — it was redone properly.

An LLM conducting the interview is an active research area

The instrument’s conversational step is not a novel idea and is not presented as one.

BIM maturity assessment is a crowded field

A 2020 evaluation by the Centre for Digital Built Britain with the UK BIM Alliance [24] inventoried fifteen tools and six methods. Its gap analysis names, in its own words, “rigid tools — one-size-fits-all; inaccurate and low granularity assessment; binary (yes/no) assessment focused on readiness and capabilities for compliance purposes; overlooking collaborative behaviour; inappropriate baselines and timing.” It reports 45% of surveyed organisations that assess maturity had built their own internal tool, and rates interoperability and IFC support at 96% industry importance. Succar’s BIM Maturity Matrix [25] adds a five-level index for “low-detail organisational self-assessment.”

What this comparison does not establish

The field survey [24] is dated February 2020 and the field has moved. The most recent comparator located, BIM-CAP [26] — by the same author as BIM-CAREM and very probably that model’s software implementation — could not be obtained on three separate routes and is recorded at T2 with no claim made about its contents.

No OpenBIM-specific instrument was found in the portions of [24] read, but that report runs to several hundred pages and was read selectively. That is a gap not found, not a gap demonstrated, and it is recorded that way rather than presented as originality.

On constraining an LLM to a schema

Requiring JSON output against a fixed schema is thoroughly established practice, and the relevant caution is the design rationale for how this toolkit treats what comes back: schema validation solves a format problem, not a truth problem — a schema-valid row can still be invented, stale, or drawn from the wrong source. Detecting fabricated citations is its own research area, working by grounding each citation against an external index. This project does something deliberately weaker: it does not verify, it quarantines at T4, where nothing can touch a score. This paragraph is written from secondary summaries rather than primary texts, and nothing is cited from it by number.

Stated as a set, in one place, because a reader should not have to assemble them from seven pages. Each is detailed in the section named against it.

1. Four of the twenty criteria are not anchored to any external source. A3 data portability, C1 tool concentration, C3 adoption barriers and C4 switching cost rest on the instrument’s own design rationale. Construct validity for those four is not demonstrable from the literature and depends on the pilot study. Section 02.
2. The instrument has not been validated. Expert validation and pilot testing are a planned phase and have not been carried out. No criterion weighting, threshold band or proportionate target in this version has been calibrated against evidence; they are provisional. Version 0.1 throughout.
3. The industry benchmark is empty. Zero assessments have been submitted, so any comparative statement in a report is a statement about an empty set and is labelled as such. No figure is substituted for it. Finding F6, section 05.
4. Findings F1–F4 have n = 1. They are measurements of individual named models. They establish that a condition occurs and is detectable; they establish no frequency, prevalence or industry rate, and must not be quoted as industry statistics. Section 05.
5. Responses are self-reported and unaudited. The organisational half of every assessment is what a respondent says about their own firm. No third party witnesses, checks or attests to it. The model half is measured, but from a file the respondent chooses to supply. Acknowledgement gate, in the assessment.
6. Three standards were read only as far as their publishers’ previews allow. For ISO 19650-2, ISO 19650-3 and ISO 12006-2, clause titles are quoted and clause bodies were not read. Anchors resting on those documents are anchored to a clause whose title is verified and whose full text is not. Section 03.
7. One source remains at T2 and six at T3. ISO 16739-1:2024 is cited from its catalogue record only. The Malaysia, United Kingdom and European Union mandate rows rest on secondary accounts whose primary instrument texts were not obtained. Sections 03 and 04.
8. Regulatory positions are current as at the date each row was read, and mandates change. This was not hypothetical: the Dubai position published here was two circulars out of date until 23 August 2026, and the error came from a secondary account that was accurate about the instrument it described. Correction C-3, section 07.
9. The toolkit measures losslessness, never lawfulness. It reports whether information is present and well-formed. It does not certify compliance with any regulation, standard, code or contract — no software does, in any jurisdiction. Local authority compliance rests with the Appointing Party and the Lead Appointed Party. Disclaimer Notice.
10. Material returned by a respondent’s AI service is unverified by construction. The guided conversation step lets a respondent bring back current context from an AI service they chose and ran themselves. Rows naming no checkable source and no date are discarded on the way in, but the rest are T4 — not verified, cite nothing from them. They never affect a score, never alter a readiness profile and are never submitted. A row leaves T4 only when someone opens the named source and reads it. Sections 01 and 02.
11. No single headline score is issued, by design. Coverage and well-formedness are independent axes, so a one-word verdict would be arbitrary rather than merely coarse. A reader wanting a single number will not find one here, and that is the finding at F3 rather than an omission. Section 05.

Standards are given in the issuing body’s own citation form. Each entry states how it was verified and on what date.

1. ISO 19650-1:2018, Organization and digitization of information about buildings and civil engineering works, including building information modelling (BIM) — Information management using building information modelling — Part 1: Concepts and principles. ISO, Geneva. Technical committee ISO/TC 59/SC 13. Terms cited: 3.3.3–3.3.6, 3.3.9, 3.3.11, 3.3.12, 3.3.14, 3.3.15, 3.3.18, 3.3.19; Clauses 5.2, 7, 8, 12. Publisher’s official preview text read, 22–23 Aug 2026 (T1).
2. ISO/IEC 33020:2019, Information technology — Process assessment — Process measurement framework for assessment of process capability. ISO/IEC, Geneva. Clauses cited: §5.2.1–§5.2.7. Publisher’s official preview text read, 22–23 Aug 2026 (T1).
3. ISO 16739-1:2024, Industry Foundation Classes (IFC) for data sharing in the construction and facility management industries — Part 1: Data schema. ISO, Geneva. Supersedes ISO 16739-1:2018; corresponds to IFC 4.3.2.0. Catalogue record read, 22 Aug 2026 (T2). No clause wording is quoted from it.
4. ISO 12006-2:2015, Building construction — Organization of information about construction works — Part 2: Framework for classification. ISO, Geneva. Clauses cited: 1, 5. Publisher’s official preview text read, 23 Aug 2026 (T1).
5. Malaysia, Ministry of Finance. Pekeliling Perbendaharaan PK 1.15 — Pelaksanaan Building Information Modelling (BIM). Effective 1 July 2025. Secondary account read, 22 Aug 2026 (T3). Primary circular text not obtained.
6. Dubai Municipality. Circular No. 196 (2013) and Circular No. 207 (2015), on the mandatory use of Building Information Modelling. Secondary account read, 22 Aug 2026 (T3). Primary texts of 196 and 207 not obtained; their substance is restated in [15], which was obtained. Superseded in operative effect by [15] and [16].
7. United Kingdom, Cabinet Office. Government Construction Strategy. BIM Level 2 required on centrally-funded public projects from April 2016. Secondary accounts read, 22 Aug 2026 (T3).
8. Singapore, Building and Construction Authority. BIM e-submission requirements. Secondary accounts and BCA published guidance located, 22 Aug 2026 (T3); the substance is restated at T1 in [12].
9. European Union. Directive 2014/24/EU of the European Parliament and of the Council of 26 February 2014 on public procurement and repealing Directive 2004/18/EC, Article 22(4). Official reproduction of the article text read, 22 Aug 2026 (T3). EUR-Lex full-text retrieval was unsuccessful; wording taken from an official legislative reproduction.
10. buildingSMART International. openBIM definition, and the IFC certification programme. Publisher’s published definition read, 22 Aug 2026 (T3).
11. buildingSMART International. Software certification register. Register consulted directly for Autodesk’s certification holdings, 31 July 2026 (T3).
12. buildingSMART International. Global openBIM Mandates, 2025 Edition — A playbook of IFC Mandates from around the world. Chapters cited: Dubai, Singapore. Publisher’s own PDF obtained and read, 23 Aug 2026 (T1). Retrieval had failed at HTTP 403 the previous day.
13. ISO 19650-2:2018, … Part 2: Delivery phase of the assets. ISO, Geneva. Clauses cited: 5.1.1, 5.1.2, 5.1.4, 5.2.1, 5.3.3, 5.3.4, 5.4.3, 5.4.4, 5.4.5, 5.8.1. Publisher’s official preview text read, 23 Aug 2026 (T1) — cover matter through the opening of Clause 3, including the full table of contents from which clause titles are quoted. Clause bodies not read.
14. ISO 19650-3:2020, … Part 3: Operational phase of the assets. ISO, Geneva. Clauses cited: 1, 3.1.1, 3.1.2, 5.1.4, 5.1.11, 5.8.1, 5.8.2. Publisher’s official preview text read, 23 Aug 2026 (T1), on the same terms as [13].
15. Dubai Municipality. Circular No. (9-1-1) of Year 2021, On Application of Buildings Information Modelling, 23 August 2021. Primary circular text obtained from Dubai Municipality’s own document server and read, 23 Aug 2026 (T1), in the Municipality’s own English technical translation, which states that the Arabic version prevails in case of ambiguity.
16. Dubai Municipality. Circular No. (9-1-2) of 2023, on the mandatory submission of the Building Information Model (BIM Model) with the building permit application for certain types of buildings, 13 October 2023. Primary circular text (Arabic) obtained from Dubai Municipality’s own document server and read, 23 Aug 2026 (T1). The Municipality’s document index records the file date as 18 October 2023; the circular’s own text is dated 13 October 2023, and that date is used here.
17. buildingSMART International. Validation Service, published page text. Publisher’s own published source read, 23 Aug 2026 (T1).
18. buildingSMART International. buildingSMART International Software Certification Program, published page text. Publisher’s own page read, 23 Aug 2026 (T1), from an archived snapshot dated 7 Aug 2024. The snapshot date is stated because the programme’s use-case list changes.
19. Succar, B., Sher, W. and Williams, A. (2013). ‘An integrated approach to BIM competency assessment, acquisition and application.’ Automation in Construction, 35, 174–189. doi:10.1016/j.autcon.2013.05.016. Author’s final version read in full, 23 Aug 2026 (T1), from the publisher-linked repository copy. Bibliographic identity confirmed independently against the Crossref record.
20. Wuttke, A., Aßenmacher, M., Klamm, C., Lang, M.M., Würschinger, Q. and Kreuter, F. (2025). ‘AI Conversational Interviewing: Transforming Surveys with LLMs as Adaptive Interviewers.’ arXiv:2410.01824v2 [cs.HC], 12 March 2025. LMU Munich, Munich Center for Machine Learning, University of Mannheim. Full text read, 23 Aug 2026 (T1).
21. Cuevas, A., Scurrell, J.V., Brown, E.M., Entenmann, J. and Daepp, M.I.G. (2024). ‘Collecting Qualitative Data at Scale with Large Language Models: A Case Study.’ arXiv:2309.10187v3 [cs.HC], 3 December 2024. Carnegie Mellon University, ETH Zurich, University of Washington, Microsoft Research. Full text read, 23 Aug 2026 (T1). An earlier version circulated under a different title; the v3 title is used here.
22. Barari, S., Angbazo, J., Wang, N., Christian, L.M., Dean, E., Slowinski, Z. and Sepulvado, B. (2025). ‘AI-Assisted Conversational Interviewing: Effects on Data Quality and Respondent Experience.’ arXiv:2504.13908. NORC at the University of Chicago, 1 December 2025. Full text read, 23 Aug 2026 (T1).
23. Lang, M.M. and Eskenazi, S. (2025). ‘Telephone Surveys Meet Conversational AI: Evaluating a LLM-Based Telephone Survey System at Scale.’ arXiv:2502.20140. 60 Decibels Inc. and University of Oxford. Full text read, 23 Aug 2026 (T1).
24. Kassem, M., Li, J., Kumar, B., Malleson, A., Gibbs, D.J., Kelly, G. and Watson, R. (2020). Building Information Modelling: Evaluating Tools for Maturity and Benefits Measurement. Centre for Digital Built Britain with the UK BIM Alliance, February 2020. Publisher’s own PDF read, 23 Aug 2026 (T1) — executive summary, tool inventory, gap analysis and interoperability findings. Read selectively, and that limit is stated wherever it is cited.
25. Succar, B. BIM Maturity Matrix (BIm³). BIMe Initiative. Built on Succar (2009) Automation in Construction 18(3); Succar (2010); Succar, Sher & Williams (2012). Publisher’s own PDF read, 23 Aug 2026 (T1).
26. Yılmaz, G. (2026). ‘BIM-CAP: A tool for assessing building information modelling capabilities of architecture, engineering, construction and facilities management processes.’ SoftwareX, 24 February 2026. doi:10.1016/j.softx.2026.102575. Gold open access, CC-BY. Publisher record read, 23 Aug 2026 (T2). Full text is behind a bot challenge on three separate routes and was not obtained; no claim is made about its contents.
27. Yılmaz, G. (2017). BIM-CAREM: A Reference Model for Building Information Modelling Capability Assessment. PhD thesis, Graduate School of Informatics, Middle East Technical University, Ankara. Full thesis obtained from the METU institutional repository and read, 23 Aug 2026 (T1).
28. Yılmaz, G., Akçamete, A. and Demirörs, O. (2018). ‘A reference model for BIM capability assessments.’ Automation in Construction. doi:10.1016/j.autcon.2018.10.022. Publisher records read, 23 Aug 2026 (T2). Full text not obtained; its substance is read at T1 in [27].
29. Yılmaz, G., Akçamete, A. and Demirörs, O. (2023). ‘BIM-CAREM: Assessing the BIM capabilities of design, construction and facilities management processes in the construction industry.’ Computers in Industry. doi:10.1016/j.compind.2023.103861. Publisher record read, 23 Aug 2026 (T2). Full text not obtained.

How to cite

This page forms part of the toolkit release and is not separately citable. Cite the toolkit:

Yusof, M.F.B. (2026). OpenBIM Readiness Assessment Toolkit for Digital Infrastructure Management (Version 0.1) [Research instrument].

No DOI has been minted. A Digital Object Identifier will be assigned at the first version-frozen release, and this section will be updated to carry it. Until then, cite the version number and the access date. Any DOI attributed to this work before that release is not one the authors issued.

To cite a specific claim, cite the underlying source above directly rather than citing this toolkit as the authority for it. The toolkit is the authority for its own instrument design — the criteria, the scoring function, the claim and evidence delta method — and not for the content of ISO standards or national regulations, which should always be cited to the issuing body.

Funding, competing interests, data availability, ethics, use of generative AI and licence are declared at the Disclaimer Notice and are not duplicated here. A reader-facing account of where this instrument sits among existing BIM maturity tools is at How this differs from other assessments.

Correspondence

Author of record and corresponding author: Mohd Faizal Bin Yusof.

Queries about a figure, a source, or a request for the underlying data: yfaizal@gmail.com

Ethics committee approval for the participant-facing study: not required. The researcher has determined that this study does not require formal ethics committee review.