NOVEMBER 4, 2026 · PRODUCT DIRECTION
Research that changes the world starts with a question.
The question is often the beginning of the work, not something you can resolve with a single answer.
Does this result support the mechanism we expected? Are two studies actually disagreeing, or measuring different things? What observation would make us abandon our favourite explanation?
We’re building aidanna for those moments. A scientific research colleague that helps turn an open question into a focused investigation, initially in biochemistry and molecular life sciences.
Follow the evidence. Keep the uncertainty.
An investigation moves through five stages: framing the question, retrieving literature, comparing hypotheses, inspecting supported data, and bringing the findings together. You can guide each step or choose an autonomous run with explicit boundaries.
Autonomous research can continue in the background when the worker is confirmed as active. Results are saved at checkpoints, so you can return to the work rather than start over. Pausing stops progression; time and usage limits remain in place.
A citation should be a starting point for scrutiny.
Aidanna retrieves literature through PubMed/Europe PMC and Crossref. Some sources provide abstracts, some only metadata, and eligible open-access articles may provide selected passages. Those access levels are visible. A retrieved record is not a claim that a whole paper was read.
New dossiers pair claims with matched excerpts and can compare studies by design, system, intervention and limitations. Checking that an excerpt exists improves traceability. It cannot establish that a model’s interpretation is scientifically correct.
A hypothesis is a proposal. A suggested experiment is future work. A completed AI investigation is not a verified discovery.
Talk through the question. Learn the idea behind it.
Live brings real-time conversation into the research workspace. Discuss saved findings, question assumptions, and explicitly ask the agent to start, pause or resume an investigation. Live audio needs an open browser session, even when research can continue independently.
Learn supports the other part of inquiry: understanding. Highlight a supported research passage and ask for an explanation in context. Start directly, or choose narrative or dialogue when another way of explaining would help.
End with a useful next action.
The aim is to clarify the next decision: which evidence to look for, which assumption to test, or which measurement could distinguish two explanations. A useful proposal also explains what its result could not establish.
Aidanna can inspect CSV files and extract text from supported documents. It does not run biological experiments, replace scientific review, or turn a targeted search into a systematic review.
A place to begin.
The ambition is simple: help people do more thoughtful work at the beginning of research. Bring a real question, examine the evidence critically, and decide what is worth investigating next.
Bring your question →