AI, Browser Assistance, and Voice Disclosures

Configured AI providers may receive the resume, profile, job, application, or interview context needed for a requested task after the applicable consent. AI output can be wrong, incomplete, biased, or unsuitable. Users must review it before use.

Interview feedback is candidate-facing. The application does not produce an overall hireability score and must not score accent, identity, emotion, disability, deception, or protected characteristics.

Remote Browserbase application assistance and connected job-board sessions are retired. Existing provider references may be retained only long enough to complete account export and deletion duties.

Supported submissions use either connected Chrome or the hosted cloud browser, according to the selected route and availability. Workday runs through connected Chrome only and may use a Job Finder-generated, encrypted candidate-account password; verification and separate legal commitments are handed back to the user.

Job Finder prepares an application bundle from the approved form and the signed-in user’s own data. The AI model may receive that user’s profile, application package, tailored resume and cover letter text, confirmed Answer Vault entries, job posting, location and preferences to prepare ordinary answers. The browser applies the prepared answers and files through the supported provider controls; it does not give the model open-ended control of the browser. Only that user’s own data can enter that context. The model does not receive cookies, passwords, browser history, unrelated tabs, screenshots, or access to the user’s computer. Job Finder checks the model’s form references and file ownership before anything is filled. Short-lived, site-bound tokens authorize the extension’s requests. LinkedIn and Indeed pages are never modified. The extension requests no debugger access, does not solve CAPTCHA, and does not claim support for every employer form.

Auto-Apply is off by default and requires the account holder to record versioned consent before it can be enabled. In per-job mode the user chooses each application; in autonomous mode the service may select eligible scored matches on the platforms the user chose. Prepared applications use connected Chrome or a hosted cloud browser according to the selected route and availability, with Workday limited to connected Chrome. Applications are submitted at most up to the user’s daily cap.

The AI model answers employer questions from the user’s own Job Finder data and its own judgment about the application, including choosing from the options a form offers when no exact stored answer exists. Auto-Apply does not solve CAPTCHA, bypass access controls, access another user’s data, or bypass duplicate-submission protection. The account holder remains responsible for the resulting application.

Each Auto-Apply submission is recorded with the job, the time, the resume sent, where it was sent from, and the confirmation observed on the employer page, under the same audit trail as assisted submissions. Turning Auto-Apply off stops new submissions immediately; applications already submitted cannot be recalled by this service.

Before first voice use under a policy version, the user separately approves microphone capture and sending audio directly to OpenAI for live speech-to-speech processing. The browser separately controls device permission for the site. Transcript text is stored as the interview progresses. After the interview, the user can review and correct each saved answer; correction history is preserved, earlier feedback is marked out of date, and the user can request updated feedback.

Text-to-speech playback sends the finalized server-owned question to OpenAI after a user gesture. Visible question text and captions remain available, and playback can be stopped immediately.

The application stores no raw voice audio. It stores conversation transcript text, finalized answer revisions, consent versions, operation timing/status, usage settlement, and safe cost metadata. This does not promise that a speech provider retains nothing; provider terms and settings require legal and vendor review.

Universal ATS support, employer-facing interview scoring, and marketing automation are not current capabilities. Automatic submission exists only as Auto-Apply, only on the supported one-click platforms, and only after the user enables it and records consent, using the execution routes described above. Voice practice is available only when its server and provider configuration is enabled. A future policy update cannot enable a capability by itself; runtime controls and separate approval are required.