NYANZA

Privacy

Version: July 17, 2026

Your career material lives in your Nyanza account and is not sold. We do not use your content to train models. Below: what we store, which providers process it and what each receives, what the extension does, and how to export or delete everything. These are the same sections as the terms, word for word.

1. What data we store, and where

  • Your career material: profile details, resume text, work history, education, skills, target role, salary preferences, work-authorization / visa status, application answers you teach, voluntary Application disclosure choices you separately save, job applications you track, interview practice results, learning progress, and the persona reflections the product derives from your activity.
  • This is synced to Nyanza's servers (hosted on Vercel, in the United States) tied to your account email, so it follows you across devices. It is your data: nothing in it is shared with other users unless you explicitly publish it (e.g. a public Twin link you create).
  • Provider credentials you save: your selected AI provider, model, and API key stay in this browser's per-user storage and are excluded from Nyanza account sync. Server-managed features use Nyanza's configured provider credentials; browser-interactive BYO requests go directly to the provider you selected.
  • Product usage events: which features are used and coarse outcomes (e.g. "a kit was generated", "an assessment was passed"), used only to improve the product. Not sold, not shared with advertisers - we don't run ads.
  • Email: we send invite/access emails and, only if you enable them, job digests. If you opt into recruiter reply routing, mail sent to your Nyanza alias is processed to suggest tracker updates - you approve every change.
  • Gmail, only if you connect it: read-only access (Google's gmail.readonly permission), offered to invited test accounts while Google reviews it. Nyanza searches the last 30 days for messages likely to come from employers or applicant tracking systems (and from the companies and contacts on your board), reads those to detect an interview invite, an offer or a rejection, and files a suggestion you confirm in the tracker; nothing changes a status on its own. We keep each message's id and what was detected, never the message text, and the access token encrypted. Disconnecting, or deleting your account, revokes the access at Google and deletes what we kept. Nyanza's use of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements: it is used only for this feature, never for advertising, never sold, never used to train models, and not read by people except with your permission or where the law requires.

2. AI processing

  • Text you provide (resume, job descriptions, questions, answers) is sent to an AI model to generate resumes, cover letters, coaching, and similar output - either our managed provider (Anthropic, subject to a daily cap) or a provider you configure with your own key (their terms then apply to that processing). Interactive requests may go directly from your browser; background and other server-side features route relevant content through Nyanza's servers before the provider.
  • Who pays for the AI: on our managed provider, we fund it and apply a per-account daily usage cap. If you configure your own key, that processing runs on your account under your provider's billing and terms - there's no Nyanza-side cap because there's no Nyanza-side cost.
  • Voice sessions: ElevenLabs processes live-session audio and transcripts. Your own saved agent credentials are used by your browser to connect; built-in voice obtains a signed connection through Nyanza's server. Features that save voice history sync the resulting transcript or feedback to your Nyanza account.
  • We do not use your content to train models, and we design the product to ground generated output in your own material - it flags figures it cannot verify rather than inventing them. Still: review everything AI-generated before you send it to an employer. You are responsible for what you submit.

Who processes your data for us, and what each receives:

  • AI models: Anthropic, and OpenAI or OpenRouter where configured, receive the prompts built from your resume, profile, job descriptions, questions, and answers.
  • Embeddings: OpenAI receives the text of your memory items and search queries, to compute the vectors behind recall.
  • Voice: ElevenLabs receives live-session audio and transcripts, speech to transcribe, and text to speak. OpenAI receives live audio when a Talk to Nia session uses its realtime voice.
  • Email delivery: Resend receives your email address and the subject and body of each email we send you.
  • Code practice: Judge0, through RapidAPI, receives the code you submit to run.
  • Avatars: DiceBear receives the avatar seed (your name unless you chose another) when the picture on a Twin you published is loaded.
  • Hosting and database: Vercel runs the application, and our Postgres database host (Neon) stores your account and career data.

3. The browser extension (optional)

  • Stores a local autofill copy in your browser and exchanges profile, kit, selected job, and status data with your Nyanza account. Saving a job or requesting generated content may route relevant page and account content through Nyanza to the configured AI provider. Filling writes your information into the employer or ATS page you opened. The extension never clicks a page-owned Next, Continue, or final Submit control.
  • The extension fills a supported application and shows the source for each answer, then you review and submit directly on the employer page. Your Career Persona, AI-generated answers, and Saved Answers never infer or supply self-identification answers. If you separately save and authorize a supported pronoun, gender, race or ethnicity, veteran-status, or disability-status Application disclosure, the extension may reuse it only from a fresh record on a reviewed, unambiguous control with an exact option; U.S. demographic, veteran-status, and disability choices also require verified U.S. role context. Unsupported, stale, unauthorized, or ambiguous questions stay for you, and legal attestations, history questions, signatures, and consents remain manual.
  • The MCP integration can read a privacy-bounded projection of confirmed Career Persona evidence, Saved Answers, your queue, Tracker, and Application Kit lifecycle metadata; search and compare postings; draft review-first answers; and stage jobs. Prepared Kit artifacts, contact details, work authorization, compensation, self-identification, and secrets stay inside signed-in Nyanza. A personal MCP token grants the stated read and staging access until you rotate or revoke it. MCP cannot fill, navigate, accept terms, or submit an application.

4. Your controls

  • Export everything anytime: Profile → Settings → Your data → Download a backup.
  • Reset your persona from the same place.
  • Delete your account and synced data from Profile → Settings → Your data. If you cannot access the account, email privacy@nyanza.ai from its address for help.

5. Contact

Questions, concerns and deletion requests: privacy@nyanza.ai. The Chrome extension’s own policy is at /privacy/extension.