In which a few things are asked & answered.

For the best AI extraction, include: an overview (title, description, eligibility, funding), key dates (open date, deadline, notification date), application requirements (what applicants must submit), evaluation criteria (how submissions will be scored), and contact information. The more detail you provide, the better the auto-generated form and rubric will be.

Only when an AI feature is actually used. Proposal text then goes to OpenRouter, which routes it to the models listed in our Privacy Policy — for the reviewer assistant, decision emails, translation, and AI eligibility pre-screening when an organizer has switched it on. Every request we send carries the setting provider.data_collection: "deny", so OpenRouter only uses providers that do not log or train on what they receive. SourceVerify receives individual reference citations, never the full proposal. We do not sell or share your data with third parties.

When enabled, reviewers see proposal content but not the applicant's identity. Fields you mark as "identity fields" (name, email, organization) are automatically hidden from reviewers. You can configure which fields are identity fields in the application form builder.

Yes. On a No AI plan you can add your own OpenRouter API key to a specific call, and that call's AI usage bills to your own OpenRouter account. Keys are stored per call, encrypted with AES-256-GCM, and only decrypted server-side when a request is made. You can also add a SourceVerify key for reference checking. CallSherpa still chooses which models are called — bringing your own key changes who pays, not how your data is handled.

Yes, active calls can be edited at any time. Need to communicate a change? Post an addendum that's visible to all applicants on the public call page. Once a call has submissions, form fields and rubric structure are locked to ensure fairness, but title, description, and dates remain editable.

SourceVerify checks academic references in proposals, flagging missing, retracted, or fabricated sources before reviewers start. Useful for research grants, academic papers, and any call with scholarly citations. Available on AI Included plans or with your own SourceVerify API key. Learn more about SourceVerify ↗

All your data is deleted: proposals, uploaded files, reviews, API keys, and any calls you own (including their submissions and reviews). This is irreversible.

No model we use trains on your proposals. All AI runs through OpenRouter on a fixed set of models: OpenAI's GPT-5.4 nano, GPT-5 nano and GPT-5 mini for anything you are waiting on, and Z.ai's GLM-5.3 Flash for background work such as pre-screening, translation and the organizer's daily memo. Every request sets provider.data_collection: "deny", which restricts routing to providers that neither log nor train on inputs. The current list, with what each model is used for, is published in our Privacy Policy and is kept in step with the code that calls them.