Privacy

Who operates FinderK

FinderK is operated by Finderk LLC. For service, billing or privacy questions, contact noreply@finderk.com or use Help & support in your account.

Your account and work

FinderK stores your name, email, encrypted password hash, session identifiers, preferences, simulation work, submissions, completion records and support requests to operate the service. Passwords are never stored as plaintext.

Payments and providers

Stripe processes card details. FinderK stores payment references and subscription status. Email is delivered through our configured email provider. Their processing terms apply to their services.

Activity and email preferences

FinderK records account and learning milestones, such as onboarding, simulation starts, submissions, report views and membership changes. These updates sync to our self-hosted Mautic system to understand the learner journey and send relevant learning emails when you opt in. Private coworker conversations, assignment contents, passwords and payment-card details are not included in these activity records.

Sign-in security emails include your IP address, time and an approximate browser and device description. You can turn learning emails off in Settings or unsubscribe from a campaign email. Essential account and service emails remain available.

Analytics and session replay

With your permission, Google Analytics and PostHog record how the site is used, and PostHog records masked session replays, to help us improve FinderK. Nothing is recorded until you allow it, and we honour Do Not Track and Global Privacy Control. Page addresses are sent without their query strings, so private links stay private. Google Analytics is set up without advertising features or Google signals. Form inputs and page text are masked; audio, video, documents and embedded workspace screens are not recorded by this analytics feature. Operational logs contain request outcomes and account identifiers, never passwords or request bodies.

Optional AI features

Optional AI features retrieve resources from external providers, including Hugging Face and MLC. These providers receive ordinary network information, such as your IP address, during activation. Conversations you save are stored with your simulation on FinderK’s server. Microphone access is requested only when you enable voice.

Workspace evidence and AI assessment

FinderK captures available saved records and changes inside your simulation workspace, including documents, conversations, email, app records, comments and available file history. Capture is periodic and may be incomplete. This is separate from optional analytics replay and does not record your device outside the simulation.

For AI assessment, your submitted work, rubric and relevant captured simulation evidence are sent to FinderK’s configured AI provider. To suggest your next steps, this can also include your active goals and your results from earlier simulations. They do not affect how your work is judged. FinderK currently uses DeepSeek for AI assessment; OpenAI may also be used where configured. Studio AI reviews send scenario drafts, supporting source materials and review instructions to DeepSeek. When DeepSeek is unavailable, the same DeepSeek model hosted by DeepInfra in the United States answers instead, for assessment and for Studio reviews. Under DeepInfra’s terms for this service, it does not keep this data or use it for training. These providers process that information under their applicable service and privacy terms. The resulting feedback and supporting evidence are stored with your report or Studio review and are available to authorised reviewers. AI feedback can be wrong and does not establish real-world professional competence. Account export and deletion controls also apply to these records.

Learning-data collection and optional uses

FinderK records versioned operational metadata about account and team milestones, simulation attempts and actions, submission fingerprints and mechanical checks, workspace capture coverage, AI and human review decisions, reviewer failures and retries, and AI usage and cost. This first-party PostgreSQL collection is separate from optional browser analytics. It does not copy passwords, authorization tokens, private reviewer instructions, payment-card details, recordings or arbitrary request bodies. Identifiers and work fingerprints are not anonymous.

Learning research, assessment benchmarking and model improvement each require a separate, optional choice in Settings → Data & account. These choices are off by default and do not affect access, grades or feedback. With permission, saved-work edits from built-in practice attempts started after the relevant purpose was granted continuously, simulation actions, saved-state observations, submissions and frozen assessment evidence can be copied into a restricted dataset. Organization-owned scenarios, invited assessments and author previews are excluded; no older work is backfilled. Only opt in for fictional work you have the right to share. Copied content expires after 180 days by default and operational metadata after two years. Withdrawal removes that purpose from existing copies and deletes copies with no remaining permission; your ordinary reports and work remain. Account deletion erases linked collection records immediately. No model-training job or external dataset export is enabled by these choices.

External assistants connected to Studio

If you connect ChatGPT, Claude or another MCP-compatible assistant, that service can receive the public authoring guidance, scenario drafts, design plans, source documents, review findings and other tool results you allow it to request in the selected Studio. It can read and edit all scenarios in that Studio and create private author previews, but the MCP connection cannot publish, change prices or access learner submissions. Do not put confidential third-party information or real learner data in scenario drafts.

Your assistant provider’s own privacy, retention and account settings apply to information it receives. FinderK stores connection and authorization records and hashed credentials, not your ChatGPT or Claude password. OAuth access tokens last one hour and can renew automatically for up to 30 days after consent when your client supports refresh. Revoke access in Studio at any time. Revocation stops future access; it does not delete information already shared with an external assistant. Request removal of those copies through that provider.

Sharing and control

Organization owners see member identities and completion counts. Public certificates expose the name and completion details on the certificate; you can make them private. Export your account data or request deletion in Settings. Deletion immediately disables access, then private work is erased after 30 days. Required financial records are retained with an anonymized account. Contact support about access, corrections, retention and deletion.