Every Building Is a Balance Sheet: Spatial Memory for Real-World Assets — Jordi Spranger, Spranic
Read the talk
Every Building Is a Balance Sheet: Spatial Memory for Real-World Assets
Jordi Spranger demonstrates how repeated building scans become a queryable record of objects and changes, then explores what that record could support in maintenance, robotics and insurance.
From a talk by Jordi Spranger
At a glance
Ideas worth remembering
Repeated sensor walks create a history of a space: recognized assets enter a semantic graph, and retained visits align along a time axis.
The kitchen demonstration connects natural-language questions to objects and identifies three displaced chairs with a reference to the supporting capture.
Insurance renewal is a proposed use of the observation history: compare initial and renewal states against policy-relevant facts. Premium savings and automated certification are not demonstrated outcomes.
The books close while the building record stands still
A mortgage or insurance record can be updated every quarter while the building underneath it goes years without a fresh inspection. Jordi Spranger, founder of Spranic, opens this collaboration with Neo4j around that mismatch. His illustrative timeline starts with an assessment when a financial instrument is created, followed by another around year five. Between those visits, financial reporting continues, but knowledge of the property may remain scattered across photographs and PDFs.
The scale makes the mismatch consequential. Spranger puts global real estate at roughly $400 trillion and focuses on commercial property, including around $4 trillion in US commercial mortgages. These are market-context figures in the presentation, rather than measurements produced by Spranic. Each financial record ultimately rests on a physical asset whose condition can change between assessments.
An underinsurance example sharpens the problem: Spranger cites research covering 1,500 properties in which roughly 70% of buildings were underinsured. He explicitly leaves the cause unresolved. Old or insufficiently detailed property records are his proposed explanation, rather than a conclusion established by that study. The product question is therefore narrower and actionable: how can a building's observed state become easier to update and inspect?
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Turn a walk into a semantic record
“What would it take to make a building as queryable as a ledger?” A ledger can return an answer in milliseconds; a building question may require a site visit followed by a search through notes. Spranic's proposed input is an off-the-shelf sensor carried through the property. In the conference-hall example, Spranger describes walking through the room and having the scan a minute later. The useful relationship to inspect is between the captured space and its database representation: as the map forms, recognized assets are stored in a graph with semantic meaning.
Semantic meaning makes the record more useful than a collection of pixels. It represents what is in the space and allows information to be connected to those assets. The reason given for using a graph is flexibility: different properties can collect different information and connect it in ways suited to their assets. The talk explains that data-model choice without specifying the recognition models, graph schema or object-matching algorithm.
A repeat walk adds history. Both visits are retained and aligned along a time axis, whether they occur a quarter or a year apart. The record can then describe a space, its contents and its observed evolution. Its freshness still depends on someone walking and measuring the property again; the mechanism described is a sequence of observations rather than continuous monitoring.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Ask what changed, when it was observed and who needs to act
The next requirement is a record that can support decisions. Spranger calls these “ledger questions” and separates three useful demands:
- What changed? Compare two walks of the same space object by object and against their surroundings.
- When was it observed? Preserve each observation's origin so a reported change can be traced to its capture. With separated visits, this establishes an observation history; it does not necessarily identify the instant the physical change occurred.
- Who needs to know? Make an observed difference available to the people who can respond to it.
Structural damage and expired fire extinguishers illustrate the proposed action layer. A finding with evidence could prompt corrective work or inform an insurer's assessment of property upkeep. These examples introduce intended uses of the record; the kitchen demonstration that follows tests the simpler, concrete case of recognizing objects and locating a change between visits.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
A kitchen map answers questions about use
The kitchen example begins with two captures, presented side by side. The system recognizes them as the same space and aligns time period A with time period B. That shared spatial reference is what makes the visits comparable: each capture becomes an observation of the same environment along a timeline.
What does querying a semantic map actually return? The demonstration uses two ordinary requests:
- Find a way to warm up. “Where can I heat myself?” points to the radiator in the map.
- Find something to drink from. A question about a glass or mug points to glasses and mugs in the kitchen.
The first request is especially instructive because it names a need rather than an object. The returned location connects that need to a recognized asset. In both cases, the map supplies a place to look, making the answer spatially useful.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Three displaced chairs make the comparison concrete
The next query asks for the delta between the two kitchen passes: what changed in this environment? The reported result is specific—three chairs in the middle have been displaced. The answer also provides a reference back to the actual capture. The useful comparison is therefore both spatial and evidential: which objects moved, and which observation supports that answer?
This is the demonstrated meaning of spatial memory. The system keeps two observations, relates them to the same space, identifies an object-level difference and makes the supporting capture retrievable through a query. The example establishes that concrete capability; it does not establish detection accuracy across buildings or performance on safety-critical defects.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Use the same record for upkeep and insurance renewal
The closing section returns to the conference-hall scan and asks what other work could consume the same record. Proposed uses include repairs, maintenance, compliance checks, insurance and interior design. Robotics adds another possibility: a robot could use the representation as a digital twin of the space. These are downstream applications of the observed asset record, rather than separate demonstrations of completed integrations.
Insurance gives the proposal a concrete timeline. A walk at the start of a policy records the property's initial state. A second walk at renewal reveals differences and evidence of upkeep. An insurer could compare those observations with facts relevant to its policy and use them to price coverage more specifically. The same history might support a higher premium for poor upkeep or a more favorable assessment of a well-maintained property; the argument is for pricing based on observed conditions.
The business case is that insurance depends on facts someone can verify. Spranger closes with three external indicators:
- Construction classification. He cites one in three commercial buildings being recorded under the wrong construction class.
- Certification incentives. He describes hurricane-premium deductions in the 35–60% range for certified buildings, although the spoken account is unclear about the percentage's denominator and eligibility scope.
- Claims experience. He cites 73% of insurers reporting fewer hurricane-damage claims for checked and certified buildings.
These figures motivate better property evidence. They are not measured premium reductions or claims outcomes from Spranic.
Certification introduces a further requirement: observations must establish the particular compliance facts a program needs. Spranger proposes using AI to check such facts from the digital asset record and extending that approach to other qualifications inside a building. The recording demonstrates neither an insurer deployment nor a completed certification workflow, so the economic outcome remains a proposed application of the scanning and comparison mechanism.
The final ambition is a more transparent, property-specific market built on knowing the building's actual state. The moved chairs provide the small working example behind that ambition: preserve visits, compare the same space over time and return the observation behind a reported difference. Maintenance teams, insurers and other systems would then have a shared record from which to ask their own questions.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Read the complete timestamped transcript
- 0:01
Hello. I'm very excited to give this talk today at AI Engineer NYC twenty twenty-six. Um, this is a collaboration between Spranic and Neo4j. Spranic, I am the founder of. It was founded in June twenty twenty-six. Neo4j is a graph database in-- based in Sweden, Malmo. And, um, a special thanks here at Zaid Zam, who is the collaborator on Neo4j's side. This is about every building is a balance sheet and a spatial memory
- 0:31
for real-world assets. The largest asset class on Earth is real estate, and it's about four hundred trillion globally in size, and, um, this is more than global equities, debt, and gold combined. If we look on the right, you see a lot of squares. Each square is about one trillion, uh, in worth, and you can see the residential part is the biggest. It's around three hundred trillion. Then you can see the commercial, it's around sixty trillion, and then agriculture, around fifty
- 1:01
trillion. And, uh, today, we will focus more on the commercial side, but all of this, um, real estate is impacted by essentially a gap. And the gap is that our financial books, uh, where we have, uh, l-lending and insurances and, um, mortgages, all of this is book kept, um, and this is updated every quarter. But the actual underlying assets, which is the real estate that they are
- 1:31
based on, um, they're not updated every quarter. In fact, they're only updated maybe every few years, maybe just before the financial instrument for the, for the real estate is being created, and also maybe again once it needs a modification or a new financial instrument on top of it, maybe in, in year five is a re-re-realistic timeframe. But between year zero and year five,
- 2:01
you really don't have an update on the building state or the real estate state. So that means that while the book closes so many times, there's really a black box or maybe something that is based on some pictures and some PDFs about the building, and this is a real gap. Um, and, uh, you can see it. So according to, uh, Crawl, um, they did a research with fifteen hundred properties. Essentially, seventy percent of the buildings are
- 2:31
under-insured. And, uh, why exactly that is, um, I don't exactly know. It didn't conclude, but my read is that the m-- the record is quite old, and it's maybe also not as detailed, so the insurance is not very highly specified and really fulfilling the needs of the real estate. And, uh, this impacts around four trillion US commercial mortgages and, um, yeah, every one of them rests on the building. Uh, so the question:
- 3:01
What would it take to make a building as queryable as a ledger? And a ledger here can be-- give you an answer in milliseconds, but a building today is giving you only an answer after a site visit, and even then, maybe in like some notes, PDFs, very dispersed and intransparent, not unified. Um, so this is where the idea comes in, where every walk becomes a memory. And, uh, how this works is you just take a off-the-shelf sensor, you walk the space,
- 3:32
and you map it. And you can see it up there actually being created. That was me, uh, going through a conference hall and the whole walk or the whole scan here took literally just walking through it, and a minute later, you had the scan. Um, so you walk through the property, whoever that is, property management, et cetera, can be equipped with it. They map it, and as they map it, you get all the assets in the building being recognized and stored in a database, which is building below. This is the graph
- 4:01
database. And, um, so this is then has a semantic meaning, which means we can really understand the space. And as we walk it again, um, both walks are kept, so walk on Monday and a Monday next year or in the next quarter, and these are now aligned on a time axis. So we have a space, we know what's in the space, and we know the evolution of the space according to every, every time we walk it and we measure it. And this is all kept as a graph.
- 4:31
Graph have a bunch of technical benefits, but one of them is they're extremely flexible in what information you want to gather and how you want to connect it. And so this can be hyper pertin-- personalized to each asset, to each, to each property. Um, yeah. And, uh, so now what is it that a record has to answer? So these are not vision questions. They are ledger questions. You really want to know in a property or in the real
- 5:01
estate what has changed. So two walks of one space compare object by object and, and surrounding by surrounding. You also want to know when it has changed, so every observation carries and captures where it came from. And you also want to know who needs to know information or deltas that have happened that we have now observed. So when you find something with evidence, for example, structural damages or maybe for fire insurance that fire extinguishers are, um,
- 5:31
off, off, um, yeah, they have de-- expired, then you need to know that, and you need to be informed and counteract on it. Or maybe the insurance can know, and they can in-increase their premiums by the performance of the upkeep of the property. So hyper-individualized insurance. Um, and here's really how it looks. So this is just me walking through a kitchen, captured twice, once on the left and once on the right. And you
- 6:01
can see you just walk the space, you directly get the map and, um, and as you walk the same space, they're being recognized as the same space, and you can align them between, um, the same space in a time period A and a time period B, and you have a timeline. And once you have a map, you can now use it as an AI-ready asset, which means that you can chat with it, you can talk with it, you can ask questions against it. For
- 6:31
example, here I ask, "Where can I heat myself?" And in the map, the answer is being given to the radiator. Or, "Is there a glass or mug to drink from in a kitchen?" It, it points you to the glasses and mugs in the kitchen. Um, so this is very powerful because it makes a map from pixels into something we can work with. Um, and now also the change detection. So this were two passes we made of this kitchen or of this space, and we can really now connect
- 7:01
these two and ask what is the delta between them? What has changed in this environment? And in this case, um, these three chairs in the middle have been displaced, and you can see when you chat with the, with the asset, with your data, you can actually identify the delta and get the ground source reference to the actual capture so you have proof of it ha- changing. Um, so what does this all mean? It means that you have now a building that is an AI-ready data asset.
- 7:32
So there's one record of what it i- what is there, and you can connect it to any downstream task. So on the left, we have this conference scan that I made, and then you can now check over time any repair that needs to be happening or some maintenance or some compliance checks or insurance or maybe interior design changes or maybe even you have some robotics that you want to place there and it's like a digital twin for it to use. So let's check the insurance. Um,
- 8:02
when you walk the building, when the policy starts, you have like a first s- initial state of the property, and on-- based on this, you can do a hyper-personalized insurance policy. And as you walk it again at re-renewal, you now see the delta and the upkeep and so on, and the insurance poli- insurance company can actually measure the performance of the property for relevant fact against its policies and give a very hyper-personalized insurance policy, which
- 8:32
means a very efficient premium, um, to insure such property and insure it in the right way.
- 8:40
Uh, yeah, and these are the numbers behind that. So the business case here for the insurance, for example, would be, um, that the insurance pays for what it can verify. And, uh, so currently what insurers know is that one in three commercial buildings are on the books with the wrong construction class. Uh, 35 to 60% of the buildings, uh, in a specific case, for example, for the hurricane premium, can actually get a deduction of, of such percentage if it's a certified
- 9:10
building. Certified building means it has certain compliance standards and that you can check from such a AI digital, um, asset of the, of the asset. And, uh, you also see that 70 per- 73% of the insurers ag-agree that, uh, they claim fewer, uh, claims for hurricane damage if they have been checked and certified against the building. And just-- this is just one such certification you can do automatically with AI,
- 9:40
plus a lot of other certification in inside of the building that qualifies maybe for such programs. But yeah, all three of them essentially depend on somebody knowing the building and, and the state of the building. Um, and that's it. That's essentially the talk. I think that the real estate class is very big, and it really needs to be transparent, and there's a huge inefficiency happening today that I think with the new insights of what AI is
- 10:10
capable to do, uh, we can make much more transparent and create a really new hyper-customized market around, around assets. Thank you very much. Um, if you want to find out more, please visit my webs- website, spranic.com, and thank you for your attention.