How we work.

A stated commercial objective is translated into structural requirements. A manager that wants exposure to logistics real estate has stated a commercial objective, not yet a structural one. It resolves into direct ownership or economic exposure; control or influence or neither; income or capital appreciation; financing exposure; the form of interest held; the duration wanted; the liquidity profile assumed; the risk and return characteristic being sought. Each of those is established in writing rather than assumed, because each answer moves the whole universe of workable structures. Where the objective is economic exposure without direct operational ownership, the set of available structures is a different set from the one available where control is required.

The investment thesis as it stands. The mandate and the constitution the fund runs under. The return the transaction has to produce and the date by which it has to produce it. The constraints your investment committee has already accepted, including the ones accepted informally.

The objective written in structural terms, with the internal conflicts inside it made explicit. Duration against liquidity is the pair that conflicts most often, and a stated objective that has not reconciled them is a structure waiting to be rebuilt.

Which jurisdiction, which vehicle, which instrument. A structure chosen before the objective is defined becomes the thing the objective is trimmed to fit, and nobody records that the trimming happened.

The existing environment is mapped in full: the fund, the manager, the investment vehicle, the asset, the jurisdictions, the ownership relationships, the instruments, the counterparties, the financing, the governance, the administration and the regulatory interfaces. This is a structural map of the problem rather than a description of the market. The regulatory reading here is contextual, never recitation: the useful question is not what the ADGM or DIFC framework says in general, but which parts of it become relevant given this particular architecture, and what that implies for the structures still open. A rule stated in the abstract tells a manager nothing it can act on.

The constitution, the limited partnership agreement or the articles, the offer document as it stands, the management agreement, the delegations, the facility papers, and the side letters already signed. The appointments already made, because an appointment is a constraint on the structure rather than a detail of it.

The map: every entity, every jurisdiction, every instrument and every interface between them, with the constraint each one carries and the document or instrument that creates it, dated as read.

Asking which route is best. The map is the ground the routes are drawn on. A route drawn before the map is a route that has ignored a constraint you have already signed, and that constraint does not become less binding for having been missed.

Pathways are constructed. Each is a particular combination of entities, jurisdictions, vehicles, ownership structures, investment instruments, financing, contractual relationships, co-investment arrangements and governance mechanisms. The question being answered is narrow and concrete: what are the possible ways to connect this fund to that economic exposure. Each route is designed around purpose, constraints, dependencies and implementation, in that order. That ordering is the reason the word is architecture: a shape that satisfies purpose and ignores implementation is a drawing, not a structure.

Decisions on the trade-offs stage one left standing, and a realistic view of what your investors will accept. A route that requires an investor base to agree to something it has never agreed to before is a route with an undeclared dependency in it.

Two, three or four routes drawn to the same depth: the entities, the jurisdictions, the flows, the security, the holders of record at each tier, the constraint that shapes each part, and the point at which each route fails.

Asking which one we prefer. A route offered before it has been compared is a preference presented as analysis, and once a preference has been stated the comparison that follows it is no longer a comparison.

Judgement enters here. The routes are set against the thirteen structural axes below and the differences are stated in the terms the decision is actually taken in, not in the terms that are easiest to write. What returns is a structural decision framework: the axes, the position of each route on each of them, and the routes rejected with the constraint that removed each one. We structure the decision environment, and the framework is built to be argued with, because a comparison that arrives with the answer already chosen has stopped being a comparison.

What your investors, your investment committee and your board weigh most heavily, and what they will not accept at any price. An axis nobody in the room cares about is an axis that lengthens the comparison without changing it.

The comparison across thirteen axes, and the rejected routes recorded with reasons. A disagreement about a route then lands on reasons rather than on positions, which is the only kind of disagreement that can be resolved in a meeting. This is the layer of the blueprint the investment committee reads first.

Asking whether the transaction is a good investment. That is a question about whether you should, and whether you should is yours. How you could, and what each route costs structurally, is ours.

Two disciplines run together. First the surviving route is attacked with the eight questions set out below, each one aimed at an assumption the structure is quietly resting on. Then the dependencies are mapped, because complex transactions are rarely difficult because of one large problem. They are difficult because entity A can only be formed once question B is answered, because financing C depends on ownership structure D, because the ownership route depends on jurisdiction E, because the regulatory position depends on what entity F is actually doing rather than on what it is called, and because the administrator has to be comfortable with the operating model before any of it can be built. Dependencies create sequence, and sequence is where transactions are lost.

The assumptions you are actually relying on, including the ones nobody has written down. The views your appointed advisers have already given, and the ones they have declined to give. An assumption that is not disclosed is an assumption that is not tested.

The failure points named, each with the condition that would trigger it. The dependency map. And the order in which the open questions have to be answered, which is the input the next stage builds the sequence from.

Asking for the structure to be certified sound. The objective is not to prove the structure perfect; it is to identify where it is fragile. A stress test that returns nothing has not been run properly, and a structure with no named failure points has not been read.

Two things, in order. Each unresolved question is routed to the professional who owns it: a legal question to counsel, a tax question to your tax adviser, a question about a regulatory permission to the appropriate regulated or legal specialist or to the regulator itself, an administration question to the fund administrator, a financing question to the lender and its counsel, and the investment decision to you. Then the implementation framework is set: the entities required, the advisers required, the validations required, the documentation dependencies, the operational dependencies, the regulatory dependencies, the sequence and the milestones. The role stays architectural throughout: the framework says which step is taken, in what order, by which role, and against which part of the drawing.

The parties you have appointed or intend to appoint, and the dates that actually bind: a first close, a long-stop, a committee that sits monthly, an expiry on exclusivity. The sequence is built backwards from whichever of those is real.

The implementation framework, drawn for the machinery that will run it: each step carrying the role that owns it, the steps that cannot be taken twice marked as such, and the sequence built backwards from the dates that bind.

Fixing a date on a step whose predecessor has no named owner. A sequence with an unowned step in it produces dates that move, and the dates that move are the ones the transaction was planned around.