Data statement
What we send where when we generate a report. Last updated 21 September 2026.
A report is built from what you type into the form, the documents you choose to upload, and publicly available publications and web sources. This page lists each outside service that takes part, what it receives, and where your data lives. The Privacy Policy covers your rights and our obligations; the Terms of Use cover the service itself.
1. What you give us
- Account details: your name, institutional email and institution, held for as long as your account exists.
- Report inputs: the research topic, researcher names, keywords, organisation, period, country, writing style and any guidance text. These are stored with the report so you can see what you asked for.
- Uploaded documents (Professional reports): used only to generate that report and deleted from our servers when the report finishes. The text extracted from them is not kept.
- Proposal text (grant adaptations): sent to the language model for that one adaptation and never stored: no database row, no log, no copy.
- Team plan membership: the organisation you belong to and your role, so the shared pool and the Team tab work.
2. Which services receive what
Every provider below is used through a paid API under terms that exclude using customer inputs to train their models. Only what a step needs is sent to it.
| Service | Used for | What it receives | Hosted |
|---|---|---|---|
| Perplexity | Web research: finding public evidence of impact | Topic, researcher names, keywords, organisation and the research questions built from them | United States |
| OpenAI | Writing and reviewing the report sections | Your inputs, the retrieved publications and web evidence, the text of uploaded documents, and proposal text for an adaptation | United States |
| Google Gemini | Reading uploaded documents, scoring publication relevance, review passes | The same material as above, for the steps it runs | United States |
| Semantic Scholar, OpenAlex, PubMed | Publication, citation and funding lookups | Researcher names, keywords and publication identifiers. No document text | United States |
Publication databases are public services; the queries we send them are the same ones a researcher would type into their websites.
3. Where your data is stored
- Reports and their working files are generated and stored on our application server on cloud infrastructure in Australia. The download ZIP stays available from your dashboard.
- Accounts, credits, Team plan membership and the list of your reports are held in a managed cloud database (Supabase) hosted in Australia.
- Uploaded documents are deleted from the server when the report that used them finishes, whether it succeeds or fails.
- Proposal text for a grant adaptation is never written to disk or to the database.
4. What we do not do
- We do not use your inputs, documents or reports to train models of our own, and our providers' terms exclude training on them.
- We do not sell or share your data, and we do not publish your reports.
- Nobody outside your account can download your report, except the owner of your organisation on a Team plan.
5. If your institution needs more
Data residency requirements, a data processing agreement, or a restriction on sending pre-submission material to a language model: write to us before you submit and we will work through it with you. For a grant adaptation you can also leave the proposal text out; the adaptation then draws only on the report.