Standards
Everything we claim, you can check.
Most software companies ask you to trust them. We'd rather show our working. Every number we publish traces to a primary source you can open yourself, and every system we build is checked by something more reliable than confidence. This is the standard behind both.
What we publish
Every figure, traced to its source.
The analysis on this site isn't opinion dressed as fact. Every statistic is linked, at the point we use it, to the primary source it came from — the government dataset, the peer-reviewed paper, the regulator's own report — so you can check us in a click. We read the source before we cite it, note when it was published and who paid for it, and keep an archived copy in case it moves or disappears.
And when a number can't be traced to a primary — the famous statistics that circulate for years without a home — we say so, and we don't use it. We spend as much effort explaining why a popular figure is wrong as most people spend repeating it. The result is slower to produce and harder to argue with. Every figure on this site has its working shown.
How the work is made
AI does the volume. People decide what's true.
We build with AI, and we'll tell you exactly how. A person decides what's worth writing, and what a claim needs to stand up. AI does the heavy lifting — the drafting, the first pass through the data. Then every figure is checked against its primary source, and a person signs off before a word is published.
The machine is fast. The judgement is human. That division of labour is the whole method.
The same standard, in code
The rigour you can read is the rigour we build with.
This discipline isn't a marketing exercise bolted onto the writing — it's how we work, and it's what you're buying. When we build your software, AI does the volume work and deterministic checks — plain arithmetic and rules, not a second opinion from another model — verify it before a human ever sees it. That's why the systems we build can read thousands of documents overnight and surface only the one that doesn't add up.
A claim you can check and a calculation that checks itself are the same instinct, applied to words and to code. If you want to see how far it goes in practice, the case study is one system built entirely this way.
The lines we hold
What we won't do.
- We won't invent a number. If we can't source it, it doesn't ship — not on this site, and not in the reports we build you.
- We won't pass a machine's word off as verified. If AI produced it, something deterministic has checked it before it counts as true.
- We won't sell you software you don't need. Where an off-the-shelf tool fits, we'll tell you to keep it — the honest answer is worth more to us than one more build.
Start here
Hold us to it.
A free process audit is thirty minutes. Walk us through the part of the business that wastes the most time, and you'll leave with a written map of what could run itself — held to the same standard as everything else here.
No pitch deck, no obligation. The map is yours either way.