Skip to content
RECAP RAVEN

Privacy Policy

Last updated: 2026-07-22 · v1.3

This Privacy Policy explains what personal data Recap Raven collects, how we use it, and your rights. It forms part of our Terms of Service.

1. Who we are

Recap Raven is a service operated by Agile Outpost (“Agile Outpost”, “we”), which is the controller of the personal data described here. You can reach us on our Discord.

2. What we collect

  • Account data: your name, email address and authentication data.
  • Session content you provide: the session audio, plus campaign notes, names and spelling-dictionary entries. Audio reaches us in one of two ways — our Discord bot records the session’s voice channel directly (our recommended path), or you give us a recording link or pasted transcript. Audio is only ever processed to produce your transcript and recap (see Retention below).
  • Discord data (when you use our bot): the Discord user IDs and usernames of participants present in the recorded voice channel, the channel/server identifiers, and the in-channel consent announcement record. You may also map a participant’s Discord identity to their player character so the recap reads naturally.
  • Optional character portraits: artwork you choose to attach to a Cast member, plus a player-safe visual description and minimal safety-check metadata derived from it. We do not keep the original upload or its filename: we keep a metadata-stripped, re-encoded copy after it passes the safety check.
  • Generated outputs: transcripts, recaps, GM notes, engagement stats and, only when you separately enable the feature, one session illustration for an eligible recap.
  • Usage & technical data: limited logs, security events and operational telemetry needed to run, debug and secure the Service. We avoid storing transcript content or secrets in logs.

3. How we use it

We use your data only to provide and operate the Service — authenticating you, downloading and transcribing your recording, generating recaps and notes, metering fair-use credits, and keeping the Service secure. We do not sell your data, and we never use your content to train AI models.

When you provide campaign notes (lore) to build your campaign spelling dictionary, those notes are processed by our AI gateway to extract the names, places and terms to spell consistently. They are held only transiently for that purpose and are not stored as campaign content; we keep only the resulting spelling-dictionary entries, which you can edit or delete.

When you upload an optional character portrait, we canonicalise it, check it for prohibited content and ask a visual AI model for a bounded description of visible character appearance. If you later enable automatic session artwork, we may send player-safe recap excerpts and relevant approved portraits to the named AI processors to create one illustration. The artwork feature remains off by default and has its own just-in-time notice.

Optional public Discord forums. A campaign owner can explicitly switch on publication of player recaps to a forum in a linked Discord server. When enabled, we send only the player-facing recap, its unlisted share link and any recap image to a forum post visible to everyone who can access that forum in Discord. We may add session participants as thread members, which can make their association with that session visible in Discord; adding them does not make the thread private. The post content does not include an attendee list or participant IDs. Transcripts and GM-only reports are not posted to the public forum; they continue through their existing private or owner-only delivery paths.

4. Legal bases (UK/EU GDPR)

We process data to perform our contract with you, including optional Service features you choose to activate; on the basis of consent where we expressly ask for GDPR consent; and for our legitimate interests in securing and improving the Service. A feature switch or rights confirmation is not described as GDPR consent unless its notice expressly says that it is.

5. Processors and third parties

We use a small number of processors to deliver the Service, each under appropriate data-processing terms. Our AI providers may briefly retain a request to operate and protect their service, but none of them use your content to train AI models, and we do not sell your data.

  • Requesty (requesty.ai) — our AI gateway, accessed over its EU endpoint. It routes your transcript and prompts to the large-language-model providers that generate your recaps, notes and chat answers. Those providers currently include Google, OpenAI, Mistral and Anthropic; the exact set may change as we tune quality, but Requesty remains the gateway and all of them are bound by no-training terms.
  • Groq (groq.com) and DeepInfra (deepinfra.com) — cloud transcription: the voiced segments of your recording are sent for speech-to-text over an encrypted connection. DeepInfra also hosts a model we use to rank search results when you ask questions about your campaign, and visual models that safety-check and describe an optional character portrait. Before automatic session artwork is made available, this notice will identify any additional image-generation processor. We delete the audio on our side within minutes of transcription (see Retention).
  • Discord (discord.com) — the platform our recording bot runs on. When you use the bot, participants’ voice audio passes through Discord to reach our bot, and we may use Discord to sign you in. Discord is operated independently under its own terms.
  • Brevo (brevo.com) — transactional email: sends verification and account/job notification emails to your address.
  • OVH — hosting and infrastructure (in France), with Cloudflare providing CDN and web-application-firewall (WAF) protection and Backblaze holding encrypted off-site database backups.

Craig is not our processor

Craig (craig.chat) is a separate, independent recording tool that some game masters choose to use. If you record with Craig, you do so under Craig’s own terms and privacy policy — we are not involved in that recording and Craig is not our processor or sub-processor. We offer Craig support only as a convenience for GMs who already use Craig and want to keep their own copy of the audio: when you give us a Craig download link, we simply download that recording to process it, which saves you downloading and re-uploading it yourself. Our own Discord bot is built independently of Craig and hosted by us.

6. Where we process your data (international transfers)

We are based in the UK and host the Service in the EU (France). Some of our processors — notably our transcription and AI-model providers — operate in the United States, so your data is processed in both the EU and the USA. Where data is transferred outside the UK/EEA, we rely on appropriate safeguards (such as the UK Addendum and EU Standard Contractual Clauses) and data-processing agreements that include no-training commitments.

7. Retention

  • Recordings (raw audio / extracted tracks): kept only long enough to transcribe them, then deleted within minutes. If our transcription queue is busy, or a session is waiting to be resubmitted after a transient failure, your audio may wait for up to about three hours before it is transcribed and deleted; anything not processed within that window is deleted. We never retain your recordings beyond this, and there is no way to download audio back out of the Service.
  • Everything else (account data, transcripts, recaps, GM notes, spelling-dictionary entries and campaign data): retained until you delete it or delete your account.
  • Character portraits and descriptions: retained until you remove or replace the portrait, remove the Cast member, delete the campaign, or delete your account. Removal deletes our serving copy. If encrypted media backup copies are introduced, they will expire under the disclosed backup-retention schedule rather than remain live or publicly accessible.
  • Generated session artwork: retained until you remove it or delete the related recap, campaign, or account. Artwork is delivered through an unlisted link. Anyone who obtains that link can view or forward it, and copies held by recipients, browsers, Discord, or third-party caches may remain after we remove our copy.
  • Artwork-setting evidence: while your account exists, we retain a content-minimal record of who enabled or disabled the feature, when, and which legal and availability versions applied. Campaign or account deletion removes the live setting and relational links. A separate content-minimal security/legal audit record may retain pseudonymous user and campaign identifiers, timestamps and version facts for up to six years, solely to demonstrate compliance and establish, exercise or defend legal claims; it contains no campaign text, portrait, generated image or asset link.
  • Optional Discord forum posts: switching the campaign setting off stops future public forum posts but does not remove posts already sent to Discord. When you edit a player recap, revoke its share link, delete its session or campaign, or delete your account, we queue the corresponding forum post for update or deletion. That operation can be delayed while Discord is unavailable, and Discord controls any copies retained under its own terms.

8. Your rights

Subject to applicable law, you can access, correct, export or delete your data, object to or restrict certain processing, and withdraw consent. You can delete jobs and your account in the app, or contact us. You may also complain to your local data-protection authority.

9. Security

We use industry-standard measures including encrypted transport, HttpOnly cookie sessions, CSRF protection and access controls. No system is completely secure, but we work to protect your data.

10. Cookies

We use strictly necessary cookies for authentication and security, such as the HttpOnly session cookie and the CSRF cookie. We do not use advertising cookies, Google Analytics, Microsoft Clarity, or optional analytics cookies.

11. Children

The Service is strictly for adults and for adult TTRPG games. It is not directed to children, and we do not knowingly collect or process the personal data of anyone under 18. You must not upload, submit or process any recording or content that includes the voice or personal data of a minor. Doing so is a serious breach of our Terms: offending accounts will be suspended or permanently banned, and we will cooperate with law enforcement and any relevant authorities in connection with child-safety matters. Ensuring that every recorded participant is a consenting adult is entirely the responsibility of the account holder (the GM). If you believe a minor’s data has been provided to us, contact us on our Discord so we can remove it.

12. Changes

We may update this policy; material changes update the version and date above. Continued use means you accept the updated policy.

13. Contact

Privacy questions? Get in touch with us on our Discord.