The Property Management Industry Has a Memory Problem

Introduction

About eighteen months after a change of managing agent at a building in Croydon, a box arrived. Inside were dozens of keys, most of them unlabelled. Some months earlier, several ring binders of operation and maintenance information had surfaced from an archive. Both had belonged to the building all along. Neither had formed part of its handover. Every one of those keys opened something, and somebody once knew what. That knowledge did not arrive with the box.

Over nearly four years, I have taken on buildings from large national operators and small local firms alike, and the pattern has been the same each time. Handovers typically consist of a data dump and, at best, the offer of a Teams call to talk it through. The documents arrive. The understanding of how the building actually works, including why decisions were made, what was tried before and which problems keep returning, does not.

This paper argues that one of residential property management’s most persistent problems is not a shortage of information but a failure of memory. Buildings outlive the managers, directors, contractors and agents who look after them, yet the knowledge those people accumulate is rarely retained by anything that outlives them. Each change forces the building to be relearned, and leaseholders pay for the relearning.

The constituency affected is substantial: an estimated 4.9 million leasehold dwellings in England in 2024-25 (MHCLG, 2026). The symptoms are well documented. The Competition and Markets Authority identified poor service quality, insufficient transparency, unexpected or unnecessary costs and difficulty obtaining redress (CMA, 2014). More recently, repairs, service charges and management have remained the leading sources of leaseholder complaints and enquiries (MHCLG, 2025; TPO, 2025; LEASE, 2026). I would argue that many of these are, at root, symptoms of lost knowledge: costs nobody can explain, repairs repeated because their history is unknown, and decisions revisited because their reasoning has disappeared. They are not only that. Property-law and housing research explains many of the same failures through fragmented ownership, collective-action problems and the balance of power between owners and their agents (Carr, 2011; Yau, 2011; Bright and Weatherall, 2017). Lost knowledge does not replace those explanations; it compounds them.

Directors of RTM and resident management companies, often volunteers, are poorly placed to notice. They see the physical building and its immediate needs. How much information the building holds, and how much of it is leaking away, is far less visible.

Knowledge management has been applied to property management before: as a route to the digital transformation of property services in China (Shen et al., 2022), in studies of knowledge-management practice in property-management companies (Najib Razali and Juanil, 2011), and in information tools for building maintenance and facility management, including a building registry (Talamo and Bonanomi, 2015). Research on multi-owned housing has examined the position of leaseholders (Cole and Robinson, 2000), the role of strata managers and volunteer committees (Easthope, Randolph and Judd, 2012) and the expertise that commercial managers hold over lay boards (Lippert and Steckle, 2016). The idea that knowledge should move from the individual to the building has even begun to appear in property-technology marketing (Proprli, 2025). In the literature reviewed for this paper, I have not identified a study specifically examining knowledge loss when a residential building changes managing agent. That is the situation this paper examines.

From Building Information to Building Intelligence

Building knowledge is more than a collection of documents. A Fire Risk Assessment records defects, an asset register identifies equipment and an email may record a decision. Knowledge emerges when those records are connected and understood. Organisational memory allows that understanding to survive the people who acquired it.

Walsh and Ungson (1991) describe organisational memory through the acquisition, retention and retrieval of information for present decisions. Stored records become useful memory only when their context can be recovered (Ackerman and Halverson, 2000). Some building knowledge is explicit, in reports, contracts and correspondence. Some is tacit, held by people who have learned how the building behaves. As Polanyi put it, “we can know more than we can tell” (1966, p. 4). The challenge is to preserve what can be articulated without assuming that everything people know can become a record (Nonaka, 1994; Nonaka and Takeuchi, 1995).

That memory should extend as far back as useful records permit, rather than beginning with the current agent’s appointment or the previous accounting year. Original design choices, unsuccessful repairs and deferred works can all explain the building’s present condition.

I would therefore propose a Building Knowledge Hierarchy: Data → Information → Knowledge → Organisational Memory → Building Intelligence. It draws on the data-information-knowledge tradition (Ackoff, 1989), while recognising that knowledge is not mechanically distilled from data (Tuomi, 1999; Frické, 2009). This is a ladder of organisational capability, not a new theory of knowledge. Its upper stages describe whether knowledge survives and whether it informs decisions.

Data “Smiths Elevators.” This is a fact, but it tells us almost nothing in isolation.

Information “Smiths Elevators maintains Lift 1.” The data has acquired context and a relationship.

Knowledge “Smiths Elevators has attended the same recurring fault on Lift 1 three times. The affected components are obsolete and replacement should now be considered.”

Organisational memory “The directors considered replacing Lift 1 in 2022 but deferred the works because a major works programme was already underway and the service-charge fund could not accommodate both projects.”

Building intelligence “Given the condition of the lift, its failure history, the obsolescence of its components, the previous decision to defer replacement and the present financial position, planned replacement next year is preferable to another substantial repair.”

A future manager reading only the latest engineer’s report may approve another repair. With the history, they can see that replacement was deferred rather than rejected, and that the reasons for deferral may no longer apply. Building intelligence means applying that memory through human judgement.

The record must also distinguish what is known from what is assumed. A contractor’s observation, an engineer’s report and a statutory inspection do not carry the same evidential weight. Each assertion should retain its source, date and supporting evidence, including any contradiction. Remembering an unsupported claim more efficiently does not make the organisation more knowledgeable.

Why Conventional Property Management Loses Knowledge

The de facto operating model of residential block management places the property manager at the centre of a building’s universe. At its best, this works exceptionally well. Given time, an experienced property manager develops an extraordinarily rich understanding of how the building actually works. They know the directors and contractors, which lift repeatedly breaks down, which section of roof leaks in particular weather and why a previous quotation was rejected. In knowledge-management terms, they become a substantial reservoir of tacit knowledge, and may hold much of the organisational memory the building needs.

Property management service models commonly emphasise the dedicated property manager, while The Property Institute’s qualifications and training focus on practitioners’ skills, competence and role-specific responsibilities (TPI, n.d.). The rationale is clear: clients want somebody who knows their building. The difficulty arises when the individual knows the building better than the organisation does. Annual leave, illness and competing demands make that individual periodically unavailable, and whoever stands in may find themselves reconstructing the building from shared drives, spreadsheets and telephone calls. The organisation has not forgotten the building, but its ability to recall it has deteriorated. If the property manager leaves, that temporary vulnerability becomes permanent. Research on knowledge loss finds that departures damage organisational performance well beyond the individual role, and that the damage can be measured (Daghfous, Belkhodja and Angell, 2013; Massingham, 2018).

Work silos compound the issue. Finance, compliance, repairs and maintenance often sit in different teams, in different offices and on different software systems, with communication still centred on Outlook. Each function may operate competently within its own boundaries; the knowledge-management problem exists between them. Finance may know that £18,000 has been spent repairing a lift over three years. Repairs may know the same component has failed repeatedly. The property manager may remember that the directors considered replacement. The budget may nevertheless contain another repair allowance, because those facts have not been connected. The failure is not always that information is absent, but that information which should be together remains separated.

Storage alone does not create organisational memory. A survey may be perfectly retrievable while the discussion that followed remains in email and the resulting decision appears only in meeting minutes. Five years later, each record remains, but the narrative connecting them has disappeared. Email is the clearest case. Outlook may be one of the richest repositories of knowledge in a conventional managing agent: it is where decisions are debated, contractors explain, directors challenge and residents describe problems. Yet in the handovers I have received, email history has never been transferred in any meaningful way. The documents, accounts and compliance records travel. Most of the context through which the building was managed is left behind.

The dedicated property manager should not disappear. An experienced property manager is one of the richest sources of knowledge a managing agent has. The weakness lies in an information architecture and organisational model that fails to extract, connect, preserve and redistribute what that person learns. The organisation should become more knowledgeable every day an experienced manager is in post. The conventional model concentrates on ensuring that someone knows the building; the greater challenge is ensuring the organisation remembers it.

The Handover Problem

Research into the transition from construction to facilities management consistently identifies a gap between transferring information and transferring the knowledge required to operate a building. Studies have found weak information fidelity, poor interoperability and fragmented exchanges between construction and management systems (Hosseini et al., 2018; Matarneh et al., 2019; Zhu, Shan and Xu, 2021), reinforcing what Tan, Zaman and Sutrisna (2018) describe as an industry “silo mindset”. Although this research is not specific to UK residential block management, it supports the view that handover is not just the transfer of information but the transfer of the ability to understand and operate a building. The same gap appears when a developer hands a new building to its owners, where problems created at handover can persist for years (Easthope and Randolph, 2016), and recent work has proposed frameworks specifically for managing information loss at handover (Abdelkarim, Ahmad and Naji, 2024).

The five cases below are practitioner observations from buildings whose management or reconstruction I have been involved in. They draw on handover records, correspondence, financial records and subsequent work to establish what had happened. They are not a representative sample and cannot establish how common these failures are. They illustrate how information can survive while the understanding needed to use it does not.

Taking over residential buildings brings that sharply into focus. My first meaningful exposure to the problem came at the building where I joined the newly constituted board of an RTM company amid concern about rapidly rising service charges. The explanations offered for those increases were individually plausible, but the board did not have a sufficiently complete picture to judge whether they explained what was happening.

What eventually arrived from the managing agent was effectively a data dump. Reconstructing the building’s history required invoices, contracts, bank transactions and other records to be compared with what directors, residents, contractors and suppliers knew about the building. That process identified duplicate expenditure, costs relating to other properties, demised repairs charged through communal funds, longstanding supplier relationships which had not been reviewed, unresolved utility accounts and significant credits held by suppliers.

The necessary information had existed, but not as a coherent organisational memory. The building had reportedly passed through approximately half a dozen property managers in the previous decade. The binders and keys described at the start of this paper arrived long after the administrative handover had been completed. The handover had taken place, but the building’s knowledge had not transferred with it.

Later we were appointed to manage a higher-risk building, taking over from a well-regarded national operator. The handover consisted principally of several gigabytes of files. The volume was impressive; the usability was not. Documents had inconsistent or obfuscated titles, version history and context were frequently absent, contractor information was incomplete or outdated, and no meaningful email archive was provided. The files included nearly 1,000 photographs, many without useful descriptions, and more than five months later an adequate financial handover had still not been completed (MYFM, 2026a). A large volume of information did not produce a corresponding quality of knowledge. Multiple versions of a report can be less useful than one authoritative version if nobody can establish which informed the subsequent decision.

The consequences were potentially significant. Under the previous managing agent, the building had entered a publicly supported building-safety remediation process and received pre-tender support funding. Within five months of our appointment, before we had progressed the funding or instructed any works, a desktop audit by the London Fire Brigade raised questions which brought the most recent Fire Risk Appraisal of External Walls (FRAEW) into focus. That appraisal, completed two years earlier, had concluded that, under PAS 9980, the external wall construction did not require remediation (MYFM, 2026a). Whatever the eventual technical or funding conclusions, a potentially decisive assessment had existed for two years without being connected to the process it most affected. The document had travelled; the knowledge had not. We reported it to the funding body, which is now reviewing whether the project should proceed.

Smaller buildings revealed the same problem in different forms. At one, acquired through a distressed transaction with no incumbent managing agent, the building’s memory had to be reconstructed from leaseholders, residents, contractors and the physical building itself (MYFM, 2026c). It is the limiting case of a handover: the custodian had disappeared entirely, and nothing was transferred at all. At a ten-flat block, a nearly flawless financial handover arrived with little context on earlier repairs, litigation and enforcement costs, matters which in a building that size could materially affect the annual budget (MYFM, 2026d). At a third, operational responsibility transferred a month early to allow time for the remaining handover. Even so, insurance and contractor information, the closing financial position and the transfer of approximately £25,000 of leaseholder reserve funds were still outstanding when the appointment ended (MYFM, 2026e). The building did not enter a transitional state simply because its managing agents did.

The statutory and professional frameworks can help. Sections 93 and 94 of the Commonhold and Leasehold Reform Act 2002 give an RTM company rights to certain information and accrued uncommitted service-charge funds. An ordinary change of managing agent is a different transition. The current RICS residential management code recommends agreeing comprehensive handover arrangements in the management agreement and recognises that most documents produced or received during an agent’s appointment belong to the client (RICS, 2026, sections 7.4–7.6). These provisions matter, but neither rights to records nor an orderly transfer can recreate reasoning that was never recorded. A compliant document handover may still be a poor knowledge handover. We receive much of the know-what and some of the know-who (Lundvall and Johnson, 1994). What is usually weakest is the know-how, the know-why and the know-what-happened-before.

A change of managing agent is also a particular kind of loss. Most research on knowledge loss concerns individuals leaving an organisation that continues. Here the whole organisation that held the building’s memory is replaced, while the building and the leaseholders who fund it remain. Most of Walsh and Ungson’s (1991) retention bins, including the outgoing firm’s people, culture, procedures and structures, leave with it; what remains is whatever was recorded and handed over. In de Holan and Phillips’s (2004) terms, this is accidental forgetting of established knowledge. It is not always accidental. Lippert and Steckle (2016) show how commercial managers’ expertise can displace lay owner-directors in condominium governance, and Orlikowski (1992) showed how professionals’ incentives can work against sharing knowledge. An outgoing agent has little commercial reason to ease its successor’s task. Industry guidance acknowledges that a handover between strata managers is unlikely to include every email (SCA WA, 2025), but the specific loss of organisational memory at that transition remains a gap in the literature reviewed here.

The Golden Thread: Government Has Already Recognised Part of the Problem

Following the Grenfell Tower tragedy, Dame Judith Hackitt’s Independent Review of Building Regulations and Fire Safety identified fundamental weaknesses in the creation, maintenance and transfer of key building information (Hackitt, 2018). One important recommendation became the Golden Thread: key information capable of passing to future building owners and supporting safety management throughout the building’s life (Hackitt, 2018; BRAC, 2021).

The Building Regulations Advisory Committee conceived the Golden Thread as both the information needed to understand and keep a building safe and the processes needed to keep that information accurate, understandable, accessible and up to date. Current Building Safety Regulator guidance adds that it should be digital, secure, usable, version-controlled and available to those who need it as a “single source of truth” (BSR, 2024).

The Golden Thread’s proper purpose is building safety, and it should not be criticised for failing to do something it was never designed to do. But its principles of lifecycle continuity, recorded decisions, usable information and reliable transfer are also principles of good knowledge management, and their value does not stop at safety. A lift does not become less dependent on its maintenance history because the building is only six storeys high. Insurance, service charges, contractor performance, warranties, procurement and asset condition all inform future decisions, yet largely sit outside the statutory Golden Thread. Where the statutory Golden Thread applies, its requirements must be met. Every residential building should ideally also maintain a persistent organisational memory of the wider knowledge required to operate, maintain and improve the whole asset.

UK leasehold reform has largely treated information as something leaseholders should be able to demand: through regulation and qualifications for managing agents (Regulation of Property Agents Working Group, 2019) and through a simpler, more flexible Right to Manage (Law Commission, 2020). Those reforms matter, but they assume the information exists to be demanded. The Golden Thread is the exception. It accepts that a building must retain its information across changes of owner, but only for safety information in higher-risk buildings. This paper argues for extending that principle to the management knowledge of every block.

A Persistent Building Knowledge Model

If building knowledge is fragmented across people, documents and systems, another repository is not enough. I propose a Persistent Building Knowledge Model in which knowledge attaches primarily to the building and its constituent entities, rather than to the employee, inbox or application through which it arrives. A property manager may discover it, a contractor generate it and a resident report it, but none should become its sole repository. Knowledge should belong to the building.

The value lies in the relationships. A lift has a maintenance history, engineering recommendations, resident reports, expenditure and board decisions. Conventional systems may hold each separately as a ticket, invoice, report or email. The model connects them so that the organisation can understand and apply what it knows (Alavi and Leidner, 2001).

Documents remain essential evidence. An engineer’s report supports knowledge about an asset’s condition; meeting minutes record what directors decided. The source should be retained while the knowledge it supports becomes discoverable through the building or asset concerned. The same applies to email and WhatsApp: useful context should not remain accessible only through somebody’s inbox or phone.

The record must preserve how the present came about. Who maintained the asset before? Why was the contractor changed? Why were works deferred? Recording decisions with their alternatives, reasoning and implications is established practice elsewhere (Tyree and Akerman, 2005; Jansen and Bosch, 2005). In block management, capture should follow consequential decisions about works, funding and obligations, rather than demand a separate narrative for every routine action.

Provenance tells us where knowledge came from; accountability determines who answers for its use. An AI-assisted budget requires review and approval by a named person. A repair needs someone responsible for its lifecycle. Conflicting evidence should remain visible, and new information should update the record without erasing the history behind it.

Not everything can be captured. Relationships, judgement and a feel for how a board or contractor behaves remain with people (Pathirage et al., 2008). Nor do stored records become memory unless they carry enough context for someone who was not there to understand them (Ackerman and Halverson, 2000). The aim is to preserve what can be recorded and lower the cost to the next person of rebuilding the rest. The organisation should not need to ask, “Who remembers what happened?” It should ask, “What does the building’s memory tell us happened?”

Building Information Standards, Digital Twins and the Missing Layer

The Persistent Building Knowledge Model builds on established approaches. ISO 19650 addresses information management across a built asset’s life, including operation and information exchange (ISO, 2018; ISO, 2020; ISO, 2022). CIBSE’s building logbook records how a building is intended to work and be maintained, while the European Digital Building Logbook envisages a broader repository supporting informed decisions and information sharing (CIBSE, 2006; European Commission, n.d.). The principle that information should remain with the building is not new.

Digital twins and facilities-management knowledge graphs also connect information about buildings and their systems. Their applications extend beyond sensors to integration, retrieval and compliance, and twins can incorporate historical and social context (Boje et al., 2020; Deng, Menassa and Kamat, 2021; Navarro Bringas, 2026). They are not inherently incapable of holding the knowledge proposed here.

The difficulty is making these principles work in residential leasehold management. Responsibility is divided among freeholders, volunteer boards, managing agents and, where applicable, accountable persons. The organisation maintaining the record may be replaced every few years. Although research addresses distributed information and the reconstruction of existing building records (Werbrouck et al., 2024; Volk, Stengel and Schultmann, 2014), the handovers described here expose a particular problem: recovering operational understanding after the custodian has changed.

That understanding is contractual and financial as well as physical. A lift’s model number belongs in an asset record. The reason its replacement was deferred, who was responsible for paying and what that meant for the reserve fund should be just as readily available.

Provenance has established foundations too. W3C PROV and building ontologies provide ways to trace individual assertions to their origins (Lebo, Sahoo and McGuinness, 2013; Rasmussen et al., 2019). The model should adopt these foundations, allowing a manager to distinguish evidence from inference and see whether an assertion has been contradicted.

Its contribution is therefore to bring established principles together around the interrupted life of a leasehold building. Where existing systems preserve that understanding, they should be used. Where they cannot, the missing capability must be supplied.

From Building Knowledge to Practical Value

Connecting fragmented building knowledge can reveal opportunities which would otherwise remain hidden. The following outcomes illustrate that possibility; they do not isolate the financial effect of knowledge management from the interventions it informed.

At one building, a fuller understanding of risk, claims, remediation and previous insurance arrangements contributed to a reduction exceeding £100,000 in the annual insurance premium, comparing 2023, the year before our appointment as managing agent, with 2024, the year after. Analysis of bills, consumption and supplier accounts supported historic water reconciliations and, with metering and monitoring, a reduction in building-wide consumption of more than 20% (MYFM, 2022–2026).

This example illustrates a progression seen across the cases: information → understanding → intervention → outcome. Value arose when information was connected, interpreted and applied, not merely possessed. Each successful intervention should then add to the building’s memory, so that the next manager does not have to reconstruct an insurance challenge or a water reconciliation from scratch. Good knowledge management compounds; its absence creates recurring costs. Building knowledge is also an asset created largely through leaseholders’ money. They paid for the surveys, engineers, managers, contractors and advice from which it arose, and they should not continually pay for the building to learn the same thing again.

If knowledge belongs to the building, it cannot belong exclusively to the managing agent. The management agreement should identify the client’s rights to its records, the agent’s responsibilities for maintaining them and the arrangements for access, correction and transfer. Those arrangements should survive a change of agent or the failure of a software supplier. At handover, the record should include current evidence, unresolved issues and consequential decisions with their reasons, in a documented form the successor can use. Otherwise, the single point of knowledge failure has merely moved from an individual to a firm. Any serious managing agent, BlockGenie included, should be judged partly by how completely it can hand that memory over.

Until recently, the cost of interpreting the enormous volume of emails, invoices, reports, minutes, photographs and financial records was a practical objection to such a model. Researchers had identified fragmented, unstandardised building documentation as a barrier to automated analysis before large language models arrived (Bodenbender, Kurzrock and Müller, 2019). In 2022 and 2023, reconstructing one building’s financial history took hundreds of hours of manual review. Three years later, an AI-assisted review at another block classified and compared similar material in days, identifying more than £25,000 of expenditure requiring challenge or investigation (MYFM, 2026b). The buildings and reviews were different, so this is not a measured productivity comparison. It nevertheless suggested that AI could reduce the cost of asking the question.

Traditional systems retrieve known records. AI can also interpret unstructured information, extracting defects from reports, possible duplication from invoices, decisions from minutes and commitments from email. It can propose implicit relationships, for example that an invoice represents the third repair to a component whose replacement was previously deferred. The question shifts from “Where is the information?” to “What does it tell us?” Research is moving the same way, with language models used to propose structured knowledge from heterogeneous documents for people to verify (Sun et al., 2025).

The danger is equally clear. AI can misinterpret documents, attach information to the wrong asset, infer relationships which do not exist and express uncertainty with excessive confidence. Grounding answers in source documents reduces the problem but does not remove it: studies of particular legal research tools and generative search engines found substantial hallucination and citation-support failures (Magesh et al., 2025; Liu, Zhang and Liang, 2023). Human review has failure modes of its own. People tend to under-check the output of automated aids (Parasuraman and Manzey, 2010), and explanations can increase acceptance of AI recommendations whether or not they are correct (Bansal et al., 2021). A named person’s confirmation is only a safeguard if it is meaningful: the source passage should be in front of them, and confirmed assertions should be sampled and audited. A poorly designed organisation equipped with AI simply produces poor decisions more quickly. The Persistent Building Knowledge Model is the framework; AI may make it economically practical at scale.

Law and Consent in a Persistent Data Model

Persistence has limits. Data-protection law requires personal data to be processed for a lawful purpose, kept no longer than necessary and protected appropriately (UK GDPR, Articles 5–6). Retention should follow the purpose and obligations attached to the record, rather than simply whether someone still lives in the building. Removing a name may not anonymise a history that still identifies the person, particularly in a small block (ICO, n.d.c). Where personal correspondence is minimised or deleted, the remaining building record must make clear what evidence remains and what can no longer be checked. The aim is to remember the building, not everyone who has passed through it.

Consent should not be the foundation on which a building-knowledge system rests. Much of the information required to administer a building is processed because it is necessary to perform contracts, comply with legal obligations, protect building safety or pursue legitimate management interests (ICO, n.d.a). Asking for consent in those circumstances may be positively misleading, because it suggests that individuals can prevent processing the organisation is required, or properly entitled, to undertake. Using an AI-assisted system does not by itself change that position. The law is concerned with why and how information is processed, not whether it sits in a filing cabinet, SharePoint or an AI-assisted system. AI does, however, demand precision. Classification and retrieval in support of the original management purpose may fall within the existing lawful basis, while model training, profiling or solely automated decisions with significant effects raise distinct questions of lawful basis, transparency and safeguards (UK GDPR, Articles 22A–22D, as amended by the Data (Use and Access) Act 2025; ICO, n.d.b; ICO, 2026).

The appropriate approach is therefore layered participation rather than universal consent. Core records and statutory communications are processed under the lawful basis appropriate to their purpose. Residents are given genuine choice over optional services, and reasonable alternatives to digital communication remain available to those who need them. Withdrawing consent disables only the optional processing that depended on it. The model’s sustainability depends on maintaining a disciplined boundary between what is necessary, what is legally required, what is reasonably expected and what is genuinely optional.

BlockGenie: A Managing Agent Built Around the Model

BlockGenie is intended to be a residential managing agent, operated by MY Facilities Management Ltd, whose organisation, processes and technology are built around the Persistent Building Knowledge Model. BlockGenie is the managing agent. BlockGenie OS is the operating system on which it runs.

Managing agents typically grow by adding buildings to each manager’s portfolio. That concentrates more knowledge in fewer people at precisely the point where it is most at risk, so growth itself becomes a source of organisational amnesia. BlockGenie is designed to grow the other way, so that each new building adds to the organisation’s memory rather than to an individual’s workload. In the early years we are deliberately growing within a defined geographic area, starting in Croydon, so that the model can be tested where it can still be verified on the ground.

BlockGenie OS is a proprietary building-knowledge system. It is in beta and being used for testing and evaluation, but is not yet used operationally to manage buildings. It is being built around the principles set out in this paper:

The building is the central object, with knowledge captured as a by-product of everyday work. Assertions retain their sources and dates, and AI-derived claims remain unconfirmed until verified by a named person. Consequential decisions require human approval and retain their reasoning. On a change of management, the building’s memory must travel in a usable form, even to a competitor. The operating system is ours; the building’s memory is not.

The AI-assisted review described earlier was not carried out in BlockGenie OS. It did, however, show that the economics of interpreting unstructured building records had changed, and it informed the system’s design.

Our current arrangements are typical of the problem this paper describes: accounting software, shared drives, email and, for some buildings but not others, a block-management package, with much of the knowledge that connects them held by people. Because BlockGenie OS will go live in stages, each measure will be baselined under these existing arrangements before the stage that affects it. The measures are:

• the time and accuracy with which a standard set of building questions can be answered;
• the proportion of invoices correctly linked to an asset or area;
• the completeness of asset repair and expenditure histories;
• the proportion of decisions with recorded reasoning;
• the time taken for a stand-in manager to become effective; and
• the time taken to onboard a new building.

Each measure will be repeated twelve months after the stage that affects it goes live. Sample checks will test unsupported assertions, missed contradictions and the effort required for verification. Someone unfamiliar with the building will also test whether an exported record is usable. The test is simple: does the organisation know more about each building, and can it recall that knowledge accurately and faster, than it did before?

We do not yet know the answer. Some knowledge may resist capture, verification may cost more than expected, and parts of the model may need to change. The argument of this paper does not depend on BlockGenie succeeding, but BlockGenie is how we intend to find out whether it holds.

Conclusion: From Property Management to Building Knowledge Management

Residential property management has largely been organised around portfolios, individual managers, accounting periods, documents and functional departments. Better software, workflows and portals can help, but a highly digitised managing agent can still suffer from organisational amnesia, and leaseholders pay for organisations to rediscover what the building once knew. The task is to turn information into knowledge, knowledge into memory and memory into better decisions.

The Golden Thread has established that critical safety information should remain accurate, accessible and persistent throughout a building’s life. This paper argues that the principle should extend to the building’s wider history: its assets, repairs, contractors, finances, failures, successful interventions and the reasoning behind its decisions.

Artificial intelligence changes the economics of maintaining that memory, but AI is not the thesis. The proposition is organisational: knowledge belongs to the building. People should contribute to it, documents should evidence it, systems should preserve it and managing agents should apply it. Every year of effective management should leave the building with a richer memory.

The Persistent Building Knowledge Model expresses that proposition, and BlockGenie is our attempt to build a managing agent that lives by it. Some of our assumptions will prove wrong. This leaves one final question: can we design a managing agent so that the organisation knows the building, rather than merely employing someone who knows the building?

Declaration of Interest

The author is the founder and CEO of BlockGenie, which is building and operating the model proposed in this paper. No external funding was received in the preparation of this paper.

Declaration of AI-assisted Preparation

During the development of this paper, the author used OpenAI’s ChatGPT and Anthropic’s Claude to support literature discovery, drafting and editing. The overall thesis, underlying argument, practitioner observations, case evidence and conclusions are the author’s own. AI-generated suggestions and edits were reviewed, challenged, amended and verified by the author, who takes full responsibility for the final content.

References

Practitioner Case Evidence

About the Author

Saif Bonar is the founder and CEO of MY Facilities Management, trading as BlockGenie, a residential managing agent built around persistent building knowledge, and the architect of BlockGenie OS, its proprietary operating system.

He holds an MSc in Information and Knowledge Management from London Metropolitan University and has experience across technology, business operations, facilities management and residential block management. His work includes Right to Manage transitions, service-charge review, procurement, contractor management, building safety, remediation programmes and building handovers.

The intersection of this practical experience with his academic work led to the central proposition explored in this paper: that failures of organisational memory contribute to persistent problems in leasehold management.

ORCID: 0009-0005-1652-3273