Design software that shows its working.

Drop a pin on a site, draw the array on satellite imagery, lay out modules and racking, wire it to inverters, simulate a year of production against real weather, and produce a report and a financial proposal - in one tool, from one model of the site.

Every module is modelled, not multiplied

Most tools simulate one module and multiply up. That works until something casts a shadow - and then the number it gives you for mismatch is a factor somebody chose, not a result. Our engine solves the array as a circuit: hour by hour, every module gets its own irradiance and its own cell temperature, the inverter sets a voltage, and each module answers with the current its own curve allows. Mismatch is what falls out of that solve.

The fifteen-step loss waterfall on every report is a decomposition of that solve, not a chain of multiplications. It is the difference between a report that tells you a number and one that tells you where the number went.

We build the workflow, not the physics

Open scientific resources already exist for irradiance, for algorithms, for component characterisations and for ground validation. Where one exists we integrate it or re-implement it against that reference - we do not invent it. Irradiance comes from PVGIS for the site's own grid cell; every model step is checked against a published reference implementation.

That matters because a yield figure is the basis of a financing decision. A report that cannot be traced to the engine build that produced it is not worth much to a bank, so every page of every report we generate stamps its provenance.

What we will not do

An engineering tool earns trust by being exact about its own limits, so these are stated rather than buried:

• We do not present unvalidated figures as validated. Component data carries where it came from and whether anybody has checked it, and a row nobody has reviewed says "never reviewed" rather than showing a date it did not earn.
• We do not fill a gap with a plausible number. A step the model did not apply reports zero and stays on the report; a value nobody measured renders as a dash. "We did not model this" and "this came out at nothing" have to be distinguishable.
• We do not put a country's assumptions into shared code. Currency, grid code, wiring standard, tax and paperwork all come from a region profile, so a second market is a data pack rather than a rewrite.

This deployment

Project data is hosted in this deployment's own region and processed under the privacy law that applies there.