AI Tooling.

Built for one real estate investment team, handed over with the code, and then I leave. There is nothing here to subscribe to.

§ 01 — What it is

What it is.

You describe the thing your team does by hand every quarter — the screen, the memo check, the portfolio roll-up that takes two analysts a week. I build it, on your data and inside your infrastructure. You get the code, the documentation, and a tool that works.

Six years inside institutional real estate investment processes is the part that usually decides whether tooling like this survives contact with a committee. The engineering is rarely the constraint. Knowing which judgement can be handed to a model, and which cannot be handed over at all, is.

You keep the code. I do not keep a copy running.

§ 02 — How it works

How it works.

  1. 1

    We scope it against a real decision.

    Not a capability audit and not a roadmap. One process your team already runs, with a defined input and an output somebody has to put their name to.

  2. 2

    I build it.

    Python or R, in Databricks if that is where your data already sits, with a Shiny interface where one is needed. Built on your infrastructure, against your data, not lifted from somewhere else.

  3. 3

    I hand it over and leave.

    Code, documentation, and a working handover to whoever maintains it after me. The engagement ends there, by design rather than by neglect.

A tool nobody owns after handover is a tool nobody trusts by the second quarter.

§ 03 — Who it is for

Who it is for.

Real estate investment teams with one process worth building for, and somebody in-house who will own it afterwards.

Investment teams.

Where two analysts spend a week each quarter on a roll-up that should take an afternoon, and everybody has agreed to live with it.

Research desks.

Screening, market monitoring and comparison work that is genuinely repeatable, and is currently repeated by hand.

Committee reporting.

The pack that gets rebuilt from scratch every cycle because the underlying numbers live in six places and none of them agree.

§ 04 — What it is not

What it is not.

  • Not a product. There is no licence, no seat count and nothing to renew. Every build is scoped to one team and one process.
  • Not a support retainer. Handover ends the engagement. If you want it extended next year, that is a new scope, quoted then.
  • Not asset-class agnostic. Real estate investment teams. The engineering would generalise; the judgement about what should be automated does not.

§ 05 — Terms

Terms.

A scoped project fee, quoted as one number before any code is written. No success fee, no licence, no per-seat charge, and nothing that recurs after handover.

You own what I build. That is the point of handing it over.

§ 06 — Next Step

Have a process worth building for?

Tell me what your team does by hand and how often. If it is not worth building, that is a short and cheap conversation.

Five working days · Personal reply from Michal