Research

Every claim behind the house is published here, with a date on it.

Kult designs houses in software and verifies them before anything is cut. The verification is only worth something if it can be examined, so we publish the research the same way we run it: each paper states a finding, the method and the numbers with their assumptions, and every paper ends with what the result does not show.

Why the research is open

No Kult house has been built yet, so we have no testimonials to offer. What we can offer at this stage is checkable work. Open papers let a structural engineer, a grant reviewer or a sceptical family member go one level deeper than the marketing copy and see whether the claim survives.

External sources outrank our own claims: where a number can rest on a published standard, a manufacturer's handbook or a third party's price list, it does. Where a result comes from our own engine, the paper says so, names the method, and states the limits.

How to read the papers

Every paper has the same skeleton: finding (the headline is the result, not the topic) → methodnumbers, with units and assumptions → what this does not showsources. Negative and inconvenient results are published on the same terms as convenient ones. One of the seven papers below reports an idea of ours failing on every test house.

All papers dated. Nothing here is gated behind an email address.

Reference photo — not a Kult build A newly raised timber frame in solid wood — reference photo, not a Kult build.
Reference photo — not a Kult build. The subject of every paper below: a small timber frame a family can raise.

Working papers

Seven papers, seven findings

  • Fit and assembly order can be verified in software before a single part is cut

    Five geometric mate checks and a per-part insertability sweep run on the same carved solids a CNC would cut. They have caught real errors that force-based checks are blind to, among them a pilot hole emitted 67.5 mm outside the timber.

    Read how the checks work →

  • Statics and moisture margin set the wall thickness, not the energy code

    A per-building-part U-value engine (SS-EN ISO 6946) shows the reference house passes the Swedish requirement already at 170 mm wall insulation. The 220 mm wall stands for other, named reasons.

    See the 170–245 mm sweep →

  • Labour takes 45–60 % of a turnkey small house, which makes materials the cheap part

    A line-item bill of materials prices a ~30 m² structural shell at ~96,500 SEK against turnkey prices of 800,000–1,400,000 SEK, then states plainly which slice self-build moves and which it does not.

    Trace the cost stack →

  • A one-board storm wall fails the racking check on every house we generate

    A storm-rated gypsum board as the sole bracing layer passes 0 of 14 corpus houses at roughly a third of OSB's shear capacity. The version that survives the numbers — storm board outside, structural gypsum inside, both faces credited — has since shipped as the house's wall, at capacity parity with the OSB it deleted.

    Read the 14-house verification →

  • Assembly by untrained hands is tested in practice

    Every part is capped at 30 kg by a check in the engine, and the first full walkthrough of the build instructions logged 14 problems. Each became a tracked design change, and several have already shipped.

    Read the assembly-test loop →

  • The tool names the rule behind every refusal

    Design freedom inside a closed catalogue of 19 test-covered junction types: measured span and pitch envelopes, six pinned refusals, and a space that grows by adding verified parts rather than loosening checks.

    Read the design-freedom paper →

  • Five staged prototypes, each with a published kill criterion

    The five-stage proof ladder runs from a 1:10 printed model to an instrumented, novice-built room module. The paper sets out what each rung tests, in what order, and which result would kill the idea.

    Read the proof ladder →

What is not on this page

These papers cover what has been verified in software and what the physical programme will test. The complete risk register, milestone plan and TRL assessment is maintained for funding reviewers in an unabridged version that includes the risks with no paper yet. That full ledger lives with the grant applications.

Read the grant-board pages →