We are an AI-native consultancy working only with UK property investment firms. We map how your firm actually runs — capital, diligence, transaction, operations — find where the money and the time are actually going, then build what that specific bottleneck requires.
No product, no pilot, nothing pre-built. One measurable bottleneck at a time, with the arithmetic shown.
From £2,000, fixed. Refunded in full unless the session identifies a bottleneck worth at least 25 basis points of the capital you plan to deploy over the next twelve months, and credited in full against the Intensive if you proceed.
A firm deploying capital is a sequence of decisions, each one gated on information arriving in time. Most of what looks like an AI problem is an information-arrival problem, and you cannot tell which is which until the machine is drawn. So that is the first thing we do, and it is what the Foundation Session produces.
A deep underwrite takes a day or more of somebody senior, so how much of the market gets examined is set by calendar space rather than by opportunity. The deals nobody had time for are not rejected — they are never seen.
The restriction that kills a deal was findable in week one and surfaced in week five, after the legal spend. Abortive costs are the price of information arriving in the wrong order.
Rent roll, arrears, void exposure and refurb status live across a managing agent's system, an accountant's ledger and two spreadsheets that disagree. By the time the picture is assembled it describes a position you have moved past.
Refurb overruns are visible on site long before they are visible in a report. The cost lands as days between practical completion and money back in the account, and nobody owns the chase.
Licensing renewals, EPC and MEES exposure, Article 4, gas and electrical certification — tracked in a diary at best, and in somebody's recollection at worst. It is a liability that only becomes visible late.
The vendor's real asking position, the rents the area achieves, what the refurb quotes came back at. Fifteen hours of work on a deal you passed, and all of it dies in a spreadsheet nobody opens again.
A tool gets bought, used for a fortnight and abandoned. A person leaves and takes their judgement with them. Infrastructure is built once, owned outright, and everything else in the firm runs on top of it. That distinction decides almost everything about how the work should be approached, so it is worth stating before anything else.
The model is available to every firm bidding against you, at the same price, on the same day. What is not available to them is your buying box, your red lines, the way your partners weigh a location, and the reasons you passed on the last four hundred deals. Any edge lives in the second thing, so that is what we build around.
A deal dies in week five because a restriction that was findable in week one surfaced late. A quarter closes on numbers assembled by hand. Neither of those is a model problem, and no amount of intelligence pointed at them will fix the sequence they arrive in. Drawing the machine tells you which problems are which.
There is no product here, no module list, and nothing pre-built waiting to be configured for you. We map how your firm actually runs, quantify which bottleneck is costing the most, and then build what that specific problem requires.
A consultancy that already knows what you need hasn't looked at your business yet.
Every engagement starts from a baseline countersigned by both of us before anything gets built, and ends with the same measures taken again. Recovered time converts to pounds at a rate fixed at signature and never revised upward. This is the whole reason a fixed fee and a refund threshold are possible at all.
The most likely first question is what happens when it misses a break clause or a licensing restriction on a live deal. The answer is that the system's job is to make sure no document goes unread, not to make the decision.
The four surfaces above are where we work. This is how. Point a general model at a deal and you get general market advice — the same advice as everyone else who typed the same prompt. The edge is the layer wrapped around it: commoditised intelligence, pointed at proprietary context.
Institutional Context Architecture is the method we use to make that context legible to a machine. It is published here in full, because a firm that hides its method usually doesn't have one.
Hiring is linear: to examine twice as much of the market you employ twice as many analysts. A system is not — it costs roughly the same pointed at fifty assets or five hundred. But cost is not the argument, because payroll is a rounding error against debt service.
Reading, extracting and checking against thresholds. Genuinely mechanical, and it scales accordingly.
Verification doesn't compress. We claim both rates separately, and never the second as the first.
Buying from the top fraction of a percent rather than the top one percent. A permanent annual gain that lifts refinance valuations, which recycles capital faster, which compounds.
One example, on one surface, using residential figures and market-standard assumptions. Illustrative, not client data. Every input is shown so you can argue with it.
Hours from a lead arriving to a defensible go or no-go with a number attached. This governs how much of the market you can examine.
The gap between what the model assumed and what the asset delivered, by assumption line. The only measure that improves the buying box rather than reporting on it.
Cost of idle capital, forward income at risk over 90 days, days from practical completion to drawdown.
Every rung is a fixed fee with a stated end date. Nothing here is a retainer you have to enter in order to find out what the work is worth.
We map the machine across all four surfaces, then name the single bottleneck costing you the most, what it costs per year, and precisely what it would take to fix.
The bottleneck named at the readout, fixed — built for your firm, live and running, with the before-and-after numbers to prove what it recovered.
Successive bottlenecks across multiple surfaces, sequenced through the year as each one is measured and closed out.
A guarantee with no conditions attached is either dishonest or unaffordable, and a numerate buyer can tell which. So both of ours name exactly what has to be true for them to apply.
We can afford the refund because we qualify hard before starting. If your firm isn't a fit, the honest thing is to say so before you have paid us anything.
Book the Foundation SessionIf the session does not identify at least one bottleneck worth 25 basis points of the capital you plan to deploy over the next twelve months — stated in pounds, with the calculation shown — we say so in writing and refund the fee in full within fourteen days.
Deciding not to proceed when a qualifying bottleneck has been found is a different thing and does not trigger a refund. The work was still done.
If the success metric agreed in writing before the build is not hit by week eight, we keep working towards it at no additional fee for up to eight further weeks.
The metric derives from the baseline countersigned at the readout, so neither of us gets to redefine success later.
The Foundation Session is a half day plus pre-work and a readout, inside two weeks. An Intensive is eight weeks from signature, with something working and demonstrated against your own data by day 21.
Around twelve hours in total from the person who runs the workflow, plus a director who can authorise and unblock. Read-only system access within five working days of signature. Access delays come out of build time rather than the calendar, so they matter more than they sound.
A data processing agreement is in place before anyone touches personal data. Your data is not used to train anything, every processor is mapped with its lawful basis and retention rule, and adding a supplier is a documented change rather than an unrecorded one.
Every extraction cites its source down to the clause, low-confidence reads are flagged for review rather than passed on silently, and there is an audit trail of what was checked and when. The system's job is that no document goes unread. The verdict stays with your partners.
No. We don't ask firms to migrate. Whatever we build reads from what you already run and writes back into it, which is also why swapping any one tool later doesn't require rebuilding the foundation.
Fixed fees, published above. From £2,000 for the Foundation Session, £8,000 for an Intensive, retainer plus a capped performance ratchet for a Partnership. No day rates, and no open-ended scope.
We go through which of the four surfaces is hurting most, whether a Foundation Session would find anything worth acting on, and what happens if it doesn't.
Refunded in full unless the session identifies a bottleneck worth at least 25 basis points of the capital you plan to deploy over the next twelve months.
No mailing list. No sales sequence. One reply, from the person who would run the work.