How we work with AI

Your project needs a memory that lasts beyond the meeting.

A floor plan records where things go. The conversation behind it holds why: how you want to serve, what your staff needs, what you want a guest to feel. Those reasons matter when the drawing changes.

Across a project, they can spread through emails, sketches, meeting notes and site conversations. Finding a file is one thing. Knowing which decision it carries—and whether that decision still stands—is another.

That is where our work with AI begins.

At KPHD, we’re exploring how a shared record can help carry the thinking between conversations. The aim is to keep the original intention visible as details develop, questions arrive and choices change.

For you, the useful questions are familiar. What did we agree? What changed? Why? What needs a decision now? The system behind the work should help us answer those questions with evidence.

Keeping the reasons with the drawing

A finish can change while the reason for choosing it still matters. A layout can evolve while the way you want to operate remains the same.

We’re studying how to connect those decisions to their sources and to the choices that follow. AI can help read and organize the record. Human judgment determines what a change means for the design.

The first article looks at what becomes visible when project records are drawn as connections: missing links, decisions without a clear history, and work that has quietly moved beyond its original brief.

Read What your project record is not telling you on GitHub

Giving the tools somewhere reliable to start

If you run a small business, you know how much lives in your head. Explaining it all again to a tool can become another task.

Our second article explores the structure underneath useful AI: one place for current decisions, clear names for different kinds of information, and links that show what belongs together.

It also sets a boundary. AI can retrieve, organize and check. The intention and consequential decisions remain with people.

Read The anatomy of a small business that survives AI on GitHub

Making room for the work

The upkeep matters too. A system that needs constant explanation competes for the same attention it was meant to support.

The final article follows the frustrations that shaped this work: repeated explanations, lost intentions and long afternoons repairing connections. It asks how routine care can become part of the record itself, leaving conversation available for thinking.

Read What the field sends back on GitHub

These are working studies from one small practice. The articles distinguish observations from results still to be measured. We’re sharing the questions and methods as they develop.

You can read all three without a technical background. The second and third also include a separate section for AI readers.

The purpose stays close to the design: carrying what matters from the first conversation into the next decision.