Trust isn't a feature. It's the architecture.
Axial is neuro-symbolic by design. A language model handles perception; a deterministic engine makes every structural decision. The two are kept apart on purpose — because a life-safety calculation can never be a probability.
One pipeline. A hard line down the middle.
Everything to the left of the line is perception — reading the messy reality of a project. Everything to the right is determination — exact, reproducible Eurocode mathematics. Nothing crosses.
The maths is code, not a model.
Each Eurocode clause is implemented as hard-coded Python against the Blue Book section data and BS EN 10025 material properties. There is no temperature, no sampling, no learned weights in the calculation path — so the result is reproducible and auditable, line by line.
The eight checks, in full.
Every steel member is taken through the complete ultimate- and serviceability-limit-state suite, to EN 1993-1-1 and the UK National Annex.
Each verdict seals the last.
Every entry stores the hash of the one before it. Change a single value anywhere in the history and every subsequent hash breaks — so tampering is not just discouraged, it's mathematically evident.
Grounded in the codes that govern.
Axial doesn't invent rules. It implements the published standards exactly — and produces the evidence a Category-3 independent verifier and the regulator expect.