Every building deserves its best design.
Physical AI for structural engineering. Built to learn how buildings really behave in earthquakes, so you can compare structural options on cost, carbon, damage and time back in use.
Why it matters
A building can meet the code and still be lost.
The code asks whether the structure will stand. It does not ask what the damage will cost, or how long the building will be closed. Christchurch showed what that gap looks like.
The quickest design wins
The fastest design that passes is the one that gets built. Heavier than it needs to be, and no more resilient for it.
Better options stay unexplored
Braces, walls, isolation, dampers. Each behaves differently, and comparing them properly takes analysis few projects can afford.
The budget is locked early
Clients approve a budget on a preliminary design. Everything decided after that costs more to change.
What we are building
Software that understands buildings, not only checks them.
Most structural software answers whether a design passes. DatumBlocks is designed to answer three more questions.
Native physical AI
Some AI drives existing software faster. Ours is being built around the physics: how real structures move, bend and sustain damage under real forces. That is what makes a better option possible, not only a faster click.
- 01
Understand
How the building would move in an earthquake, where the damage would be, and what it would cost.
- 02
Propose
Different systems and materials that would behave better. Not just bigger members.
- 03
Recommend
The option that fits the project best, with the numbers on the table.
What you will see
Every option, measured the same way.
Four numbers, side by side, for each structural option. Before the money is committed.
- Build cost
- What the structure costs to construct
- Embodied carbon
- What the structure costs the climate
- Expected damage
- What an earthquake would cost to repair
- Time back in use
- How long the building would be closed
Proposed by AI. Checked by analysis. Decided by engineers.
Every proposal is meant to be checked independently before anyone relies on it. The engineer keeps the final say.
Where we are today
A private preview, in development. Starting with steel buildings and earthquakes in New Zealand. Independent accuracy results come before any performance claim.
Read our approach →Who it is for
Built for everyone who carries the cost of a building.
Engineering firms
Justify the better system without a specialist on every project.
Developers and architects
Choose the right structural option while the budget is still open.
Building owners
Know what low damage design buys before paying for it.
Insurers and councils
Expected loss, and which buildings to strengthen first. A later product.
Resources
All posts →1 June 2026
Introducing Structural Design
What we are building: design and analysis for steel eccentrically braced frames, with a published scope and nothing claimed before it is proven.
20 May 2026
Why structural design is an optimization problem
The slowest part of structural engineering isn't drawing. It's the search for a sizing that satisfies the standards without being wasteful. That's a problem software is good at.
Follow what we are building
Development updates and engineering notes as the preview progresses. No spam.