RabbitSoftware
Back to project overview
The story / Why this exists
Software for people who want to check the work
Research software that people can inspect, question, and control — not simply trust blindly.
The problem
Research is hard to explore, and the tools that make it easier often ask for something in return: your files, your data, or your trust. AI answers arrive without sources. A scientific name, an AI answer or a blockchain record can sound authoritative without giving you any way to examine it.
The idea
RabbitSoftware exists to make research easier to explore without making privacy, transparency or access an afterthought. Start from public sources, keep every answer tied to where it came from, keep your own records on your own device, and make the system’s memory of what changed something anyone can check.
Evidence should be traceable
Records keep their source identifiers. Answers cite them. A record, a remembered experience and a verified finding are different things, and the software treats them differently.
Uncertainty should be visible
AI summaries are labeled unverified. Experimental ideas are labeled experimental. Research prototypes are not presented as clinical capabilities.
People choose what happens to their information
Local tools stay local. Sharing waits for consent. Personal data never goes on the public chain.
What exists today
- RabbitOS, an account-free workspace at app.rabbitsoftware.io: public research records with their source links, a brain and body explorer, an open data hub, and AI chat where you approve what is sent.
- A research pipeline that retrieves public records (PubMed, Europe PMC, ClinicalTrials.gov, ClinVar, NIH RePORTER and others), keeps their identifiers, and proposes what to add. A person approves what is stored.
- A public, tamper-evident record of what changed, copied to several places, with a proof-of-work chain and Bitcoin timestamps. See the chain, explained.
- Local-first tools you can run on your own computer, including a read-only assistant and a local file viewer.
- A throwaway cloud terminal for trying AI coding assistants with your own key, inside a locked-down sandbox.
- Status you can watch at the status page: what is answering, whether the chain copies agree, and when the research agents last reported.
Who is behind it
RabbitSoftware is created and maintained by Chase Allen Ringquist. The work grew out of a long habit of asking how information relates to human experience: what can be measured, what the measurements mean, and what they cannot tell us. His own account of that path is on the author page, written in his words and marked as a personal reflection rather than verified research.
What we are careful about
- This is research tooling, not clinical software. It does not diagnose, recommend treatment or edit DNA in a body.
- Experimental blockchain and biomedical descriptions are research goals, not established capabilities. Not every experiment becomes a working capability, and the project says so.
- A record on the chain is a record, not a verdict. A timestamp shows when something existed, not that it is true.
- AI-assisted development helps build and test the system; it does not replace evidence, consent or scientific validation.
How you can help
- Use it and tell us what breaks. Open RabbitOS, or read the known gaps.
- Check our work. Verify the chain yourself with public files and standard tools, and watch the status page.
- Report a problem or ask a question. See reporting contacts or write to [email protected].