How we work · the capability
Structural intelligence.
Information tells you what a regulation says. Structural intelligence tells you what that regulation means for the architecture you are trying to build.
Both answers are true. Only the second one changes what gets built. This page puts five of those questions twice, then follows one of them down to the layer where it is actually decided.
01 · The distinction
How we workA rule is a fact. What it does to your structure is a different answer.
The text of an instrument is knowable, current and, in these centres, largely public. Knowing it is the beginning of the work. What it does to the chain you are proposing is the end of it, and the two are separated by everything this page is about.
The distinction resolves into four statements. The fourth is why it matters commercially rather than intellectually.
The text of an instrument, the version currently in force, and the date it last moved. It is finite, it is checkable, and a great deal of it can be read by anyone who knows where to look. A reading that stops here has produced evidence and no design.
The consequence of that text for the entity chain, the instrument, the sequence and the exit you have in mind. Which layer of the chain the rule attaches to. What it forecloses two jurisdictions later. Which of the choices you have already made it has just made irreversible.
A rule has one text and any number of structural consequences, and which consequence applies depends entirely on the structure standing in front of it. The same article is decisive in one architecture and inert in another. That is why the second answer cannot be looked up, only worked out, and worked out against a particular chain.
Managers rarely lose a transaction because a rule could not be found. They lose it because a rule that everyone had read turned out to attach to a layer nobody had drawn yet. The reading is not the product. The consequence is.
02 · The same question, twice
The questions we are askedFive questions. Two answers each, and only one of them is a structure.
Each row is a question a manager brings to us in roughly these words. The middle column answers it from the text. The right column answers it from the architecture the manager is trying to build. Every cell in the middle column is correct. Not one of them tells you what to build.
| By question · by kind of answer | The information answer | The structural answer |
|---|---|---|
| May the fund hold this asset through a vehicle in another jurisdiction? | The vehicle exists in that jurisdiction and is available to a holder resident elsewhere. The formation route is published and the requirements are written down. | Whether your fund may hold it is not a question the vehicle answers. The questions are whether your own constitution permits an intermediate holding layer at all, whether that layer changes which court would enforce a pledge over its shares, whether it moves the point at which the asset stops being visible to the investors who negotiated for visibility, and whether it can be unwound at exit without a consent you do not currently hold. |
| Does this activity need a permission in the centre? | The activity is named in the rulebook, the permission that covers it is named alongside it, and the application route is set out. | The prior question is which entity in your chain is carrying on the activity, because the answer moves with the layer. Lift the function one level and the permission moves with it. Move it outside the centre and a different perimeter engages instead of this one. The permission is a consequence of where the function sits, and where the function sits is a design decision you have not taken yet. |
| Can the fund borrow against its commitments? | The constitution states a borrowing limit. The facility papers state a covenant. Both are readable, and both were drafted by people who knew what they were doing. | They are read against each other and against the close calendar, because that is the only place they meet. A limit expressed against commitments and a covenant expressed against drawn value move in opposite directions when a close slips a quarter. Neither document is wrong. The one that breaks first is the one nobody read against the other. |
| Which court hears a dispute about this security? | The instrument names a forum, and that forum has jurisdiction over what the instrument covers. | A forum clause settles where the argument happens. It does not settle where the asset is. What decides recovery is the distance between the court that hears the matter and the registry that has to give effect to its judgment. Where those sit in different systems, enforcement has a last mile the clause is silent about, and that last mile is a property of the ownership chain rather than of the clause. |
| Can the fund leave this holding without a consent it does not hold? | The transfer provisions sit in the constitution and in the shareholders agreement, and both are readable. Each states who must consent, on what notice, and in what form. | The consents are not in one document and they do not run on one clock. A change-of-control provision in a finance document, a pre-emption right in the shareholders agreement and a registration condition at the asset each attach to a different layer of the chain, and the slowest of the three settles the date the exit is actually available. Which one binds is a property of the chain drawn at entry rather than of the offer that arrives at year five. |
03 · One question, followed through
Holding chainsOne question. Four layers, and the answer sits in the third.
Take the first question in the table above and follow it to the end. The information answer closes in two sentences. The structural answer takes four layers, and it changes which vehicle gets formed.
The information answer is this: the vehicle exists, it may be held from outside the jurisdiction, and here is the route to forming it. It is accurate. A manager who acts on it forms the vehicle, and discovers what it costs at the third layer rather than at the first.
The chain the question actually runs through
- 00 The fund Where the commitments sit and where the terms were fixed. The constitution was drafted before this asset existed, and it is the document the whole answer has to survive. What it permits by way of an intermediate holding layer, and on whose consent, is settled text that nobody is going to reopen for one asset. The constitution
- 01 The holding layer The company the fund would hold the asset through. Its jurisdiction decides which court enforces a pledge over its shares, which registry has to give effect to that judgment, and whether the lender at layer 00 can take security that reaches past it. It also decides how many consents an exit needs, because a layer added for one purpose has to be removed for another. Jurisdiction A
- 02 The local vehicle The entity that may actually hold the asset where the asset sits. Whether it may be owned from outside, by whom, in what proportion and on what continuing conditions, is the constraint that decides every layer above it. This is the stratum the information answer named and stopped at, and it is the only one whose terms are not negotiable by anybody in this chain. Jurisdiction B · binding
- 03 The asset The thing the manager wanted in the first place. It has not moved once through any of this, which is precisely the point. The intention was fixed at the start and the structure was drawn to reach it. When the two conflict, it is the structure that is redrawn. Unchanged
Read top down, the chain looks like a series of choices. Read from the binding layer outward, it is one constraint and three consequences. The order of reading is the whole of the difference.
Three things follow from having read the chain in that order, and only the first of them looks like an output.
Not a new rule. The same conditions at layer 02, read before layer 01 was chosen instead of after, which changes the jurisdiction of the holding company, the security package the lender can take, and the number of consents standing between the fund and an exit. Three structural outcomes from one change in the order of reading.
A conclusion on foreign ownership law belongs to counsel qualified where the asset sits. The question goes to them framed as a question rather than as a request for reassurance, with the structural consequence of each possible answer already drawn, so counsel answers one question instead of working out what it is being asked.
Chains are drawn from the top because that is where the manager sits. Constraints bind from the bottom because that is where the asset sits. A structure drawn in the wrong direction is usually workable for months and unworkable at the moment it matters, which is the moment the money has already been spent.
Nothing in that reading required an instrument nobody could find. It required knowing which of the four layers to read first.
04 · What it is made of
Where structures breakSeven things held at once. Not one of them is a rule.
Seven understandings are held together over one transaction. Each is a way of reading a structure rather than a fact about one, and the value sits in the holding together rather than in any one of them on its own.
- How structures work
What a vehicle actually does to ownership, control, cash and liability once it is standing in a chain rather than described on a page. A legal form is a set of consequences before it is a name, and the consequences are what get inherited by everything above and below it.
- How structures interact
What happens at the junction between two of them. Feeder and master. Holding company and operating company. Parallel vehicle and main fund. Components that are each sound can be jointly unsound, and the unsoundness exists only in the joint.
- Where structures fail
The assumption that looks harmless at the term sheet and is decisive at the exit. Structures rarely fail at the point that was argued over. They fail at the point nobody recognised as a decision, which is why the work is to find the decisions that were taken without being noticed.
- How jurisdictions interact
Not what each system says standing alone, but what each does to the other when both are inside one chain. Two coherent systems can produce an incoherent combination, and the incoherence belongs to the combination rather than to either system. Reading them separately never finds it.
- How regulation shapes architecture
Which entity carries a named activity, which permission that entity then needs, and what changes when the function moves one layer up or one border across. Regulation is a constraint on the design rather than a description of it, and it is read as a constraint or it is not being read.
- How an economic objective becomes a legal and operational structure
The manager wants an exposure. The exposure has to be produced by entities, instruments, ownership relationships and cash flows that counsel can document, an administrator can operate and a regulator can recognise. That translation, from what is wanted to what can be built, is the whole of the discipline.
- What has to be settled before execution
The questions whose answers change the structure, separated from the questions whose answers do not. Most transactions carry dozens of open points and three or four of them are load-bearing. Sorting them is the difference between a sequence that holds and a sequence that is rebuilt twice.
Each of the seven is ordinary on its own. Holding all seven over one chain, at the same time, is the work.
05 · The lens
How routes are comparedThirteen readings of the same structure. All thirteen are taken before a route is preferred.
The lens is the thirteen structural axes, set out one by one at how routes are compared. The same thirteen questions are put to every structure, in the same order, whatever the asset is and whatever the centre is.
A reading taken out of turn answers a question the previous reading had not yet settled. Regulatory perimeter read before the entity that carries the activity has been placed returns an answer about an entity that does not exist yet, and the answer then travels through the file as though it did.
Not every axis is live on every transaction, and an axis that is dead on this chain is recorded as dead rather than skipped. The axis nobody asked about is the one that turns out to bind, and there is no way to know which it is until it has been asked.
A structure that is attractive under one axis and fatal under another is the ordinary case rather than the exception. The only defence is refusing to stop at the reading that flattered the structure, which is a discipline about when to stop rather than a fact anybody holds.
The lens is not a checklist for its own sake. It is the only reliable way to stop one attractive feature from hiding a consequence three readings away.
06 · The patterns that recur
The one-way doorsEight patterns recur. Each of them is a question asked before anybody spends money on documentation.
None of this is proprietary. Every published instrument an engagement cites can be read by anyone who knows to look for it, and we would tell you where. What the reading adds is the shape: which architecture this one resembles, which assumption inside it has broken before, and which question therefore goes first.
The eight below are the patterns the questions come from. They arrive as questions in a first meeting rather than as citations in a memorandum, because at that point a question can still change the structure and a citation cannot.
Structures that worked, together with the conditions under which they worked. A precedent stripped of its conditions is an anecdote, and anecdotes are what get repeated into failures.
Structures that did not work, together with the assumption that broke them. The assumption is the useful part. The structure itself is rarely repeated in the same form.
Structures that work in one direction only. Building them is straightforward. Reversing them is a different transaction, and usually a worse one, undertaken by people with less time.
Regulatory interfaces that recur. A small number of junctions produce most of the perimeter questions in this market, and knowing which they are shortens the reading rather than replacing it.
Operational problems that recur. A structure counsel can document and an administrator cannot operate is not finished, and the discovery normally lands after the constitution is signed.
Sequencing mistakes that recur. Almost every one of them is a step taken before the step that had to precede it, and almost none of them was visible as a step at the time.
Jurisdictional incompatibilities. Two systems, each internally coherent, that cannot both be satisfied by one chain. The incompatibility is invisible in either system read alone.
Dependencies that were never written down. The ones that surface only when something upstream moves, and that nobody thinks to mention because nobody has ever had to say them out loud.
Not one of the eight is a rule that can be looked up. Every one of them is a question about a chain, and it is answered against that chain or not at all.
07 · Speed
When to involve usEveryone reaches the answer eventually. The question is how much has already been committed when it arrives.
Any structural problem yields to enough time. Private markets rarely supply enough time. Exclusivity expires, a committee sits monthly, a first close is dated, and the transaction is often decided by whoever narrows the field first.
Here are the routes that survive the constraint, here is the single question that decides between them, and here is who has to answer it. That sentence, said in the second week rather than the fourth month, is what the reading is for.
Four things follow from that, and the third is the one most often misheard.
Not in finding the rule. In working out which of the available architectures the rule bites on, and in what order the remaining questions have to be answered so that each one is asked when its answer can still change something.
The cost avoided rather than the work performed, and the cost of an assumption is the number of parties who have already relied on it. That ladder is set out rung by rung at what the state is worth.
Arriving at the right question sooner rather than asking fewer of them. The surviving routes go to your counsel and your regulated advisers on the terms they always would, and they arrive with the field already narrowed, which is a shorter brief rather than a shorter process. Nothing in the sequence is skipped. What moves is where in the sequence the decisive question sits.
The universe of structures that could theoretically produce an exposure is large. The set that survives the constitution, the perimeter, the financing and the exit is usually two or three. Getting from the first number to the second quickly, and being able to show why each route fell out, is the work.
The answer arrives either way. What the timing decides is how many parties have already built on the version that was wrong.
08 · Contextual, not encyclopaedic
Our firmWe hold what the rulebook does to the structure in front of us. That is the smaller body of knowledge, and the harder one.
Research tells you what exists. Architecture tells you what it does to the thing you are trying to build. The second has no existence except in relation to a particular chain, which is why it cannot be stockpiled in advance of one.
Nothing here is a standing answer. Every reading is taken against your chain, your constitution, your calendar and your investors, and it is dated, because the instruments move and a reading that does not say when it was taken is not evidence of anything.
What a reading returns is the routes that survive, how they differ across the thirteen axes, what each one depends on, where each is most likely to break, and which questions have to be answered formally, by whom, before one of them is preferred. That is structural clarity, and it is not the same thing as certainty.
We take a transaction through structural design and execution-readiness. Execution remains with the manager and its appointed counterparties, and the structural analysis provides the framework within which appointed counsel undertakes the relevant validation.
Information can be obtained. What it means for your structure has to be worked out against your structure.
Write to us.
Complex transactions fail at the interfaces between otherwise workable components. We resolve the structural complexity between investment intent and transaction execution.
Contact us