Energy & industrial
Tools where the maths is the product. Configurators, geospatial calculation and estimation software, built alongside the physicists and engineers who own the underlying model.
- 01Domain logic that lives with experts rather than in a specification
- 02Calculations heavy enough that a round trip to the server ruins the experience
- 03Sales teams needing an answer during the conversation, not the next day
- 04Geospatial data from providers that each model the world differently
- 05Estimates that have to be defensible, because someone will build on them
In-browser geospatial
Calculation optimised to run client-side, so drawing a shape gives an answer immediately.
Configurators
Interfaces where a non-expert makes expert decisions correctly, with the constraints enforced by the tool.
Working with domain experts
Building the algorithm alongside the people who own the physics, rather than from a written handover.
Estimation and yield modelling
Calculations that are traceable, so a number can be explained and defended.
Map integration
Satellite imagery and geometry via Turf.js and map providers, handling the inconsistencies between them.
When the domain expert is not an engineer
The defining feature of this work is that the hardest knowledge sits with someone who does not write code. A physicist knows how irradiance translates to yield. A process engineer knows why a configuration is invalid. Neither of them can hand you a specification that survives contact with an implementation.
So we build it with them, in short cycles, checking outputs against numbers they already trust. A written handover of a domain model is where these projects go wrong.
What we have built
Helios is a web-based solar design tool for Autarco. Dealers and customers trace their roof on a satellite map and immediately see how many panels fit and how much energy the system would generate. The geospatial calculations were optimised to run client-side using Turf.js, because a configurator that pauses for a server round trip stops feeling like a tool and starts feeling like a form.
The algorithms were built together with the company's physics experts rather than delivered from a specification.
Why in-browser matters
A salesperson sitting with a customer has about two seconds before the conversation moves on. Drawing a roof outline and watching the panel layout appear changes the nature of that conversation. The same calculation delivered as a report the following morning is a different, much weaker product.
That constraint drives the architecture: as much computation as possible on the client, with the server reserved for the things it genuinely has to own.
Tell us the one thing it has to do.
If it's sharp enough to prove in a week, we'll say so and give you a date. If it isn't, we'll tell you that too, along with what we'd build first instead.