Post Mundane

01 Why most of this fails Evidence

95^%

of pilots deliver no impact.

That is why we don’t open by selling you a tool. A tool projected onto a process nobody mapped is how most of that 95% happened — and if you have already been through one of those, you are not sceptical, you are correct.

Find out why

02 What changes Before / after

The same decision, taken earlier — while you can still act on it.

01 Sites you can look at

One at a time → a city, overnight. You look at what a broker has already worked, a parcel at a time, because that is what the week allows. Screened against your own box instead, the shortlist is drawn from everything that fits — and the ones that do not pencil are gone before they have cost you a Tuesday.

02 The number that breaks it

Found in month three → found on day one. Every deal turns on one figure: the most you can pay for the land, and the first assumption that fails when it moves. Found early, it is a negotiating position. Found late, it is a write-off you are already committed to.

03 Judgement that stays

In people’s heads → queryable, and added to. Twenty years of knowing which sites are worth a second look currently leaves the building when its owner retires. Held where it can be asked questions, it compounds instead — and your apprentices learn the patterns in months rather than a decade.

See the approach

The tool’s map view: an aerial photograph of a district with a masterplan drawn over it, seen inside the dashboard as a zoomable map, with a shortlist of sites listed beside it.

The same dashboard over a full visualisation of the scheme: towers, a public square and street trees rendered in daylight, with the visualisation controls and the scheme’s cost and value beside it.

The same dashboard with the pitch deck it generated filling the canvas: a cover slide for 127 O’Carroll Street, a rendered view of the scheme at golden hour beside the title, and the form that produced it at the foot of the screen.

03 Our philosophy Five layers, in order

Mapping our process to your business goals.

Exploded — process layers

Select a layer

04 How we work Five phases

Five phases. One line we walk together.

Every phase ends with a decision made with you, not for you: carry on, or stop here with something useful in hand. Either way, you know why.

01Tied to your own KPI

One scope, zero risk

We test whether AI can move your margins against the objectives you already have — not against a demo. If it can’t, we say so and recommend nothing. What we do build is measured on a number your board already watches.

02Your people in the build

Upskill your team

AI doesn’t remove a team. It changes how the team works. So we embed with the people who will run the system while it is still being built — and the capability stays in the building after we leave.

03In your files, your accounts

Your data, real ownership

We build from your own documentation. That means tools and workflows shaped around the way of working that made you successful — not a product that asks your firm to work its way instead.

04Found by your own team

Expansion

Once the first scope has produced a real outcome, we look at what to scale. The people we upskilled find the next leverage points with us — they know the process better than we ever will.

05Last, on purpose

The brain

Every file your firm holds in one place, and anyone can find anything in seconds. It is the end goal for anyone serious about AI in development — and it is where most firms begin, which is why most of them never see the money. We prove it on real projects first, then make the whole firm run on it.

05 The readiness audit Deliverable · five sheets

Nobody should build anything before this.

01 The process, on one page

We sit with the people doing the work and draw how a deal actually moves through your firm — including the steps that only exist in somebody’s inbox. Most firms have never seen the whole of it at once.

02 Leverage points, ranked

Every candidate scored the same way: the hours it gives back, the risk it takes off the table, what it costs to build. Ranked, so the conversation is about the order — not about whether.

03 Tied to numbers you already own

Each one mapped to a KPI your board already reports, with today’s baseline measured while we are in the building. Without the baseline, nothing built afterwards can be shown to have worked.

04 One recommendation, argued

Build it, buy it off the shelf, or leave it alone this year. Sometimes the honest answer is a third-party tool and no further fee to us — and when it is, you get that in writing.

05 A scope you can say no to

If there is something to build: what it does, the KPI it answers to, the date it lands on, and the deadline we carry with you. Scoped and priced before you have committed to any of it.

Book the audit

AUD · 01–05

AUD-01 Process map

AUD-02 Leverage, ranked

AUD-03 KPI mapping

AUD-04 Build, buy or wait

AUD-05 Scope & deadline

06 The AI Studio Build status, stated

S1 Development Tooling

04 / 04

Test-fit Vitrux — envelope solver Solves the zoning envelope — setbacks, angular plane, streetwall build-to, upper stepback — then lays a floor plate inside it with core, back-of-house and egress placed. Live Feasibility Pro forma Costs, revenue, the financing stack and returns from a site and a scheme — with the sensitivities that decide whether you walk. Live Feasibility Package builder Pro forma, floor plate and massing produced as one package, so the numbers and the drawing can never drift apart. Live Design Design iterator Holds a design philosophy as parameters and refits it to each lot — sun, light, site — so a scheme adapts without being redrawn. Live

These are tools our clients use day to day. The studio is where they get built: leading tools developed for your firm and delivered into it, not licensed to you off a shelf.

07 The team Three of us

Three of us, in three cities.

We do not split into a seller and a builder. Both founders sit in the discovery and both write the code that comes out of it, so the people who scope your work are the people who deliver it. Between London, Ottawa and New York we cover the working day twice over, which is less a slogan than a scheduling fact.

Will Dunne

Founder · Advisory & engineering · London

Scopes the work and builds it. Came here from years running his own AI consultancy, and still sits with clients directly rather than behind a delivery team — the systems you end up using are ones he has written.

Qmin Lee

Founder · R&D & engineering · Ottawa

A background in architecture before running his own business; he and Will met when their two companies worked together. Leads R&D here, on sovereign tooling — the systems a firm owns outright rather than rents.

Scott Sivan

Partner · New York

Forty years in real estate development, working across the US, the Middle East and Europe. Sits with us on the tools themselves: what gets built, and whether the market it is built for actually wants it.

08 Three things we actually believe Position

Intelligence, not artificial

The thirty years of judgement inside your firm is the asset. Our job is to store it, not to replace the people holding it — which is also the honest answer to whether this costs anyone their job.

Leverage points, not software

The hard part was never building software. It is knowing where to put it. If you already have a data team, they don’t need us to write code — they need someone to find the workflow they haven’t gotten to.

We are early, and we say so

We are not going to show you a wall of logos we haven’t earned. You don’t need our track record — you need one result on your own data. That is exactly why the first step is small and the second one is priced.

09 Start Thirty minutes

Tell us about one live file.

The most useful thirty minutes we can spend together is on a project you are actually working on right now — its phase, what is slow, and what you would need to see to believe any of this. Bring that and we can be specific.

Decisions your firm takes in a quarter, taken in a week.