Research

What we are still thinking through

Work at the stage before code. Each of these is a written concept — the problem stated properly, the design argued, the risks listed without softening them. Neither has been implemented, and we say so on the page rather than in a footnote.

Why this half of the company is public

A concept presented as a product is a lie that costs a reader their time. A concept presented as a concept is something else: evidence of how the company thinks before it commits. We would rather be judged on that than on a page of adjectives.

Neither of the entries below has a line of code. Some will never be built. That is the normal outcome of research, and pretending otherwise is how software companies end up shipping things nobody thought through.

Concept document · September 2026 · No code

Quorum

The problem. A meeting is an expensive, synchronous mechanism doing two jobs that are not the same: distributing context, and reaching a decision. Neither needs everyone in one room at one time. Worse, asking for a decision live forces a commitment before the person has had time to check it, consult their team, or weigh what it costs them — so the answer you get is often the safest one to say out loud rather than the best one available.

The proposal. An asynchronous deliberation layer for organisations. Material and a question go in with the people who must know and the people who must decide; what comes out is a certified record — a detailed agreement, signed by the participants, with a verifiable log of what was shared and what each person said. That record then becomes the reference the resulting work is checked against.

What we are unsure about. Whether an organisation will accept a decision process it cannot hurry, and whether the certified record is valuable enough on its own to justify the change. The concept document lists these objections rather than answering them prematurely.

Status: concept document, version 0.1, September 2026. Working name, not a settled brand. No implementation commitment.

Concept document · August 2026 · No code

kith.ing

The problem. A social network's value is not its software, which anyone can rebuild, but the social graph it holds. While that graph belongs to the platform, a better product cannot win, because adopting it means losing the people you talk to. That is not a market failure; it is the design.

The proposal. Not another social network — the layer underneath them, so that the experience becomes interchangeable and the data stops belonging to the platform. Four decisions hold it together: identity verified once by an independent verifier while the person stays pseudonymous in public; content stored where its owner chooses; availability through permitted replication among the people entitled to see it; and a client anyone can build over the same data.

What we are unsure about. Verification at scale and distributed synchronisation are both hard problems with a long history of failure, and the concept document says so in those words. The economics of a system that refuses advertising are equally unproven.

Status: concept document, version 1, August 2026. The domain is registered to the company. No implementation commitment.