About
Research software, built with the people using it
StarshipAI builds OrbitLab: a Windows desktop suite that takes a study from question to write-up without the dataset ever leaving the researcher's machine.
Most statistical tooling asks a researcher to choose between two bad options: a general-purpose package that will happily run the wrong test, or a hosted service that wants the dataset uploaded first. OrbitLab exists because neither is a good answer for clinical and academic work.
The design constraint is simple. The analysis runs on your machine, it runs deterministically, and the app is explicit about what it cannot do. AI assists with the parts where judgement helps — framing a proposal, proposing variables, structuring an extraction sheet — and stays out of the computation entirely.
OrbitLab is in Commercial Beta. It is developed release by release with the researchers using it, and the changelog is public precisely so you can see what was broken and what was fixed.
Principles
What we optimise for
Deterministic before clever
Anything that produces a number a researcher will publish is computed by a deterministic engine. AI is used to help plan a study and structure data — never to produce a p-value.
Local before cloud
Participant data belongs on the researcher's machine. Building a product around uploading datasets would be easier to operate and worse for the people using it.
Honest before impressive
When a sample is too small, the app says so. When PDF export cannot render a script, it tells you instead of printing question marks. Software that hides its limits produces papers that inherit them.
Using OrbitLab for real research?
Beta feedback shapes what gets fixed next. If something is wrong, slow, or missing, we want to hear about it.