Leysa
ley·sa [ˈlay·sa] Icel. — to solve, to untangle
Tools that automate work with data on energy, land and buildings
Leysa is an engineering practice run by an architect who is also a software developer. It knows the industry from the inside, so it builds tools that fit the way people work, not the other way round. From PV farms controlled by energy prices, through plot analysis, to data flowing between BIM models.
What Leysa does
1. Energy. PV farms that switch themselves off at negative prices. Production plans sent to the grid operator with no one at the keyboard. Price forecasts given as ranges, not a single number.
2. GIS. Public land, zoning and weather data retrieved automatically. Plot selection for PV and wind farms. 3D site plans as the starting point of a project.
3. Construction. BIM is a flow of information, not just a 3D model. Revit add-ins, IFC, bulk model validation and CI/CD for design, so data entered once ends up where it is needed.
4. Software and AI. Custom software, technical oversight of other vendors’ implementations, and deliberate adoption of AI tools in teams, including local models.
5. Case studies. Selected projects, each told as problem, solution, result.
In figures
| Result | Context |
|---|---|
| 40 h down to 10 min | of weekly manual work at a design office replaced by scripts (plus 2 h of review) |
| 3 days down to 3 h | R&D process at a steel-structure general contractor |
| 30–40% lower | daily error of PV production plans than naive methods, backtested on many farms (2022–2026) |
| 40% lower | day-ahead price forecast error than “same as yesterday” |
| 0.93 | ROC-AUC for detecting quarter-hours with negative prices |
| about 200 | IFC models validated at once by a single pipeline |
Why Leysa
- Industry and software expertise in one person. An architect and BIM coordinator who also writes and maintains the software.
- Solutions run in production. PV farms are monitored and controlled automatically, and production plans reach the operator every day.
- Results are measured. Every forecasting model is backtested and compared with simple baselines.
- No vendor lock-in. Open source, with the option to host on the client’s servers. Code and data can belong to the client.
- Delivered end to end. From process analysis through prototype and rollout to maintenance and team training.