Privacy policy

1. What is processed and where

Transcript text and confirmed case terms

Opening a file loads it into your browser. Starting AI checking sends the text and confirmed terms through our server to OpenAI’s API for analysis. The app does not create a server-side transcript library. Processing through our server and provider is still a disclosure of that content outside your device.

Local drafts, decisions, and notes

Recovery drafts contain the transcript, filename, findings, edits, decisions, and personal notes. They are stored in this browser’s database on this device, not synchronized to another device. New drafts are listed only for the signed-in account that saved them. Older drafts without an account association are preserved but hidden, not automatically assigned to the next person who signs in. The app does not add its own encryption to drafts or downloaded review backups. Personal review notes are not included in AI review requests.

Recordings in a proofreading review

Attaching a recording, playing it, and lining it up with your transcript’s CAT time codes happen in your browser; the recording is not uploaded. Check against recording is optional: after you confirm, it sends a short clip of the recording (about 6 to 30 seconds around one line) through our server to OpenAI’s API to be written out, and returns the words to your browser to compare with the suggestion. We do not store the clip or the words; the words are kept only in your review draft and backups. The same OpenAI terms and retention described below apply.

Audio-scoping drafts and recordings

Opening an audio-scoping ZIP, matching saved transcription and playing a recording happen locally. Choosing Create reviewed draft after confirming the disclosure sends transcript text and saved speech-recognition text through our server to OpenAI; that step sends text only, not the recording. Scope from the recording, where it is offered, is different: after you confirm, your browser cuts short stretches of the recording (up to about 35 seconds each) and sends only those stretches through our server to OpenAI’s API to be written out. The whole recording is not uploaded, and we do not store the stretches or the words. The same OpenAI terms and retention described below apply. Audio-scoping work is held in the open tab, not autosaved to My transcripts. Save progress downloads a ZIP containing the transcript, findings, edits, decisions and attached recording. The smaller JSON backup excludes the recording. These downloads are not encrypted by ECH Transcript Studio.

Account and activity information

Our hosting sign-in service supplies an account identifier and email for access checks. Our activity database records user identifiers, joined/last-seen dates, review identifiers, counts, progress, decisions totals, and status. It also records the AI token counts and estimated cost of each review, and the number and estimated cost of recording checks, linked to the account. It does not store transcript text, filenames, or recordings. The site owner can see per-account usage totals (reviews, pages, tokens, and estimated AI cost) to run the service and manage costs. Ratings are stored with your account but shown to the site owner without your name.

Feedback and support

Submitted ratings and comments, including thumbs-up or thumbs-down answers from our emails, are stored with a user identifier, optional review identifier, and timestamp. Administrators can read feedback. “Did we miss something?” reports are stored with your account, and the report text, along with any comment, is emailed to the site owners so we can fix the miss. That text stays in our records and inboxes until removed, so anything you type or paste there, including transcript lines, is kept. Support emails contain the information you choose to send and are handled through our email service. Do not include confidential transcript content in reports, feedback, or support messages.

Notices to the site owners

The site owners get an email when an account is created, when a payment is made, and when a check starts, finishes, stalls, or fails. These include the account’s name or email address, page and line counts, and timing, never transcript text or file names.

2. Why information is used

We process information to provide requested reviews, restore local progress, control access, record usage, investigate problems, respond to support, and improve the service from submitted feedback. Personal My notes are part of your review, not a feedback submission.

The application does not include advertising trackers or a data-sale feature. Hosting and sign-in services may use necessary cookies or similar storage and process connection, security, and diagnostic information under their own policies.

Links in our emails can carry a short tag naming which email they came from (for example “email-a”), never who received it. If you follow one, the tag is kept in a cookie on our site for 30 days and, if you create an account, saved as your answer to “How did you hear about us?” Our emails do not contain tracking images, and we do not track whether you open them.

3. Providers and access

ECH Transcript Studio uses OpenAI’s API for AI analysis, Cloudflare for hosting, application infrastructure and activity storage, Clerk for sign-in, Stripe for payments (card details go to Stripe, not to us), and Resend for account email. Support correspondence uses our email provider. These services process information needed to operate their functions. Authorized operators may access server-side records for support and administration; the aggregate dashboard is not a guarantee that operators cannot access underlying records.

We may disclose information when required by applicable law or when reasonably necessary to address abuse or protect the service. This policy does not override a court order, your client’s instructions, or your professional duties.

4. AI storage and model training

Anchor Point AI Consulting LLC has signed OpenAI’s Data Processing Addendum, which covers how OpenAI processes the text we send to its API. Our API requests disable response storage. This is not a zero-retention guarantee. OpenAI documents default abuse-monitoring retention of up to 30 days, with exceptions, and other feature-specific storage. Its API training policy excludes customer data by default unless the account opts in to sharing.

Sharing for model improvement is disabled for the API project used by ECH Transcript Studio. We do not opt your transcript inputs or review outputs into OpenAI’s model-training program. This does not eliminate provider security retention or change how previously shared data was handled. See OpenAI’s data controls for its current policy.

5. Retention and removal

• Browser drafts: retained until you remove them, clear site data, or the browser removes them. There is no automatic expiration. Use Remove in Recent drafts for drafts listed under your account. Clearing site data removes all local recovery, including older unassigned drafts and other accounts’ drafts in that browser profile; back up authorized work first. Signing out alone is not a way to delete browser drafts.
• Downloads: remain wherever you save them, including synced or backed-up folders, until you remove those copies. We cannot delete files from your computer.
• Activity, feedback, and support: the pilot has no automated expiration schedule or self-service server-record deletion. Contact us to request access, correction, or deletion. We may need to verify your identity and explain any records retained for legal, security, or operational reasons.
• Provider records: subject to provider policies and settings. Deleting a draft does not delete provider logs or previously submitted records.

6. Your choices and security

You can decline AI processing and use the prepared sample instead. Use a private, protected device and browser profile. Account-filtered draft lists are not encryption: someone with access to that browser profile or your unlocked tab may still access local content. Close review tabs before signing out or changing accounts. Store exports and backups according to your professional obligations. No internet service or browser recovery copy is guaranteed secure or continuously available.

The live site uses HTTPS. Cloudflare documents encryption at rest for D1, which holds our activity and feedback records—not a transcript archive. These provider controls do not establish that ECH Transcript Studio has its own SOC 2 report. ECH Transcript Studio has not completed an independent security audit, and no claim of HIPAA readiness or US-only processing is made.

Depending on where you live, additional privacy rights may apply. Contact us with a request or concern; please do not send identity documents or sensitive transcripts in the initial message. This pilot is intended for adult professional users, not children.

7. Done For You

This policy describes the self-service pilot. Before any managed-service intake, contact us without attachments to agree on data suitability, human access, transfer method, and retention. The website is not permission to email confidential material or assume a separate confidentiality or healthcare agreement is in place.

8. Contact and updates

Contact Anchor Point AI Consulting LLC (doing business as ECH Transcript Studio) through our contact page for privacy questions or requests. We will post revisions here with a new effective date. Review the policy before submitting material if the service or your requirements change.