IWhat we collect
Bouquin & Ink collects only the data necessary to provide our writing tools: your account email, manuscripts you create, and basic usage analytics. We do not sell your data to third parties. We do not enrich it with information from other companies, nor do we share it with advertisers.
Account data — email, password hash, plan tier, billing identifiers — is stored to operate your account. Manuscript data is stored to provide the editing service. Usage data: feature use, error logs, performance metrics, are all stored to improve the product.
IIYour manuscripts
Your writing belongs to you. We store it on our servers by design in the editor, when you delete a manuscript it's gone from our active database immediately. It may persist briefly in encrypted backups for up to seven days before being overwritten. After that, it's unrecoverable, including by us.
When you close your account, the entire archive goes with it on the same schedule.
We do not train AI models on your creative work without explicit opt-in consent. If we ever offer a research opt-in, it will be clearly labeled and reversible at any time. The default is off, unless you choose otherwise.
IIIAI features and third-party processing
Some of Bouquin & Ink's intelligence features run on third-party AI infrastructure. When you use those features, the relevant portion of your text is sent to a third-party provider, processed, and the result is returned.
Voiceprint scoring is the exception. The feature that builds your writing fingerprint and scores chapters against it runs on our own servers. It's the most sensitive thing we do, therefore it stays in-house.
Everything else uses third-party AI processing. Continuity checking, Corkboard synopses, Muse Board prompts, image generation, and world bible entity extraction all send relevant text — usually a single chapter or a small excerpt, never your entire manuscript at once — to a third-party provider to be processed. Your text is not used to train any AI model. It is retained by the provider only as long as their data retention policies require, then deleted. After the result comes back, we don't keep the text on our side either, it's processed and discarded.
Image generation is a special case. When you request a Muse Board image, a short manuscript excerpt is sent to a text model first to distill a scene prompt. Only that distilled prompt, about 150 characters, gets sent to the image model. The image model never sees your raw prose.
Things we have decided are off the table.
- Generate prose anywhere in the editor — there is no generation feature currently in Bouquin & Ink, and we will not introduce one without explicitly notifying you and giving you the choice to use it or not.
- Retain your text after analysis is complete.
- Use your manuscript to train any model, ours or anyone else's, without explicit opt-in consent.
- Share your text with any third party for any reason beyond the AI processing described above.
- Sell, lease, or rent any data we collect.
Provider responsibilities. We choose providers whose terms meet our privacy bar — including no training on user content. We monitor their terms and will reassess our use of any provider whose terms change in a way that doesn't meet that bar.
A note on pasted content. Bouquin & Ink can't tell where text in your editor comes from. If you paste content generated by another tool, Voiceprint will score it against your fingerprint like any other passage — it can flag the inconsistency, but it can't determine how the content was created. Your provenance certificate documents your process inside Bouquin & Ink; what you choose to bring into the editor is your responsibility.
If you use our BYOK (Bring Your Own Key) plan, your AI requests are passed through to the third-party provider you've chosen, using your own API key. We don't see, retain, or store the content of those requests on our servers — but you should review your provider's data handling policies.
IVBeta calibration
While Bouquin & Ink is in beta, we use anonymized usage signals — voice match scores, consistency patterns, audit accuracy, false-positive frequency — to calibrate how Voiceprint scores writing against the fingerprint it builds for each user. This is not AI training. We are not feeding your manuscripts into a generative model. We are tuning the scoring system so it works better for the writers who use it.
Your manuscript text is sent to our processing layer when you run an analysis, scored, and discarded once the result is returned. Nothing is retained beyond the moment of the audit. Only the analytical metrics produced by Voiceprint are studied — not your prose.
Your feedback shapes the calibration. During beta we actively want to hear from you when Voiceprint flags something it shouldn't have, or misses something it should have caught. Beta participants can submit feedback at any time. We use those reports to improve the scoring engine for everyone.
What changes when beta ends. Calibration is ongoing — the AI and publishing landscape shifts, and Voiceprint needs periodic re-tuning to stay accurate. After beta, calibration becomes opt-in. You'll be invited to join an anonymized signal program if you want to keep contributing to how Voiceprint evolves; if you don't, your data stays out of that work entirely.
Beta participation is free, and this calibration is how we ensure our product is ready for launch.
VIYour rights
You can export your manuscripts and project data at any time, in standard formats, from inside the editor. You can delete your account and all associated data at any time, from your account settings. We honor data-access and data-deletion requests promptly. If you're in a jurisdiction that grants additional rights (GDPR, CCPA), those rights apply to you.
VIIContact
Questions about your data? Reach us at hello@bouquinandink.com. We answer.