Privacy Policy
Effective September 29, 2026 · Last updated October 29, 2026
At a glance
- This policy covers the parlor study web app and the Nevel Trade company website. An iPhone app is planned for later.
- Your saved notebooks, conversations, practice and reading records are stored on our server. Some preferences and unfinished drafts stay in your browser.
- AI features share relevant content with OpenAI and, for lecture notes processing and handwritten page images, Anthropic after the applicable sharing permission.
- Demo accounts work like other accounts but run AI features on a lighter OpenAI model, with a weekly limit and an end date. Referral links let students invite friends; a referrer sees counts, not who joined.
- Instructors can give a section one class code to join. Instructors don’t see students’ work in parlor study; we may tell an instructor class-level totals, never names or study content.
- Signing in (with Google or a code sent to your email), Help and Feedback involve separate services, described below. Help and portal records are not removed automatically by deleting your parlor study account.
- We do not sell your personal information or use it for targeted advertising. parlor study is for people aged 13 and older.
Who we are and what this covers
parlor study is a browser-based reading and study service operated by Nevel Trade Inc, a Florida corporation in the United States (“we”, “us”). This policy covers parlorstudy.com, its supporting server, and neveltrade.com. It describes the web service currently being tested, including optional features when enabled. A future iPhone release will have disclosures appropriate to that app.
For privacy questions or requests, contact info@neveltrade.com.
Sign-in and account information
When Google sign-in is offered, Google verifies your identity and we receive your Google account identifier, verified email address and display name. We keep these with your account creation and sign-in times, invitation information, and confirmation that you are at least 13. We do not receive your Google password or access to your Gmail or Drive. Our sign-in integration does not send your notes, answers or reading history to Google.
Email codes: you can instead sign in with any email address. We send a 6-digit code to that address through our email delivery provider, Resend, which receives the address and the message containing the code. We only send codes to addresses that already have an account or a pending invitation, and our reply on the sign-in page is the same either way. Codes expire after 10 minutes and are stored only in a scrambled (hashed) form, which is deleted once the code is used, replaced or expired. To limit repeated requests we keep a scrambled record of each request for one day; the address itself is not stored in these records. Code emails come from signin@parlorstudy.com, which does not receive replies; write to info@neveltrade.com instead.
The code request is protected by Cloudflare Turnstile, which checks for automated abuse. Cloudflare receives information about your browser and connection for that check, and our server sends Cloudflare the check result token and your IP address to confirm it. If you sign in with Google and with an email code for the same address, both reach the same account where the address is confirmed. You can set your own display name in Settings; it replaces the name from Google within parlor study and is included when you open Feedback.
Invitations are assigned to an email address and expire after 30 days if unused. Unused, expired and revoked invitations remain admission records until removed. Redeemed invitation details are included with the account's export and deletion. Authorized administrators can see account and usage information to admit readers, provide support and operate the service.
Suspension: if we suspend an account for misuse under our Terms of Use, we keep the date and our reason with the account, end its sessions and refuse sign-in until we restore it. The reason is a note for our administrators; ask us at info@neveltrade.com to see it. It is removed when the account is restored or deleted.
Invitation requests: the form at parlorstudy.com/request-invitation asks for your email address, confirmation that you are at least 13, and how you heard about parlor study, and optionally your role (such as student or teacher) and where exactly you heard of us. With it we record the first parlor study page you opened in that browser tab, any link tags in its address (utm_source, utm_medium, utm_campaign), and the other website you arrived from, if your browser reports it. We use your email address only to reply about your invitation, and the other answers to understand which channels bring people to parlor study. We do not store your IP address with a request; it is used briefly in memory to limit repeated submissions. Authorized administrators can see requests. Requests are deleted 12 months after your last one, and with your account if you later join using the same email address.
Demo accounts: if you join with a demo link or a demo invitation, your account is a demo account. We keep that, the date your demo ends, the demo link you joined with, and whether you asked for full access, with your account. A demo account stores the same kinds of study data as any other account, and its AI features all run on a lighter OpenAI model (see AI processing). When the demo ends your saved work stays until you delete it or your account; we don't delete ended demo accounts automatically. If we switch your account to full access, everything you saved stays with it. A request for full access sends our operators a short Telegram notice without your email address.
Referral links: when we turn referral links on for your account, you get a personal link and code to invite friends. A friend who signs up with it joins directly instead of with an invitation. We record which account joined through whose link and when, whether the friend has used parlor study enough to qualify (counted from their AI use and study time totals, not their content), and any reward credits given. As a referrer you see how many friends joined and are studying, never who they are; our administrators can see both accounts. A new sign-up with a referral link sends our operators a Telegram notice without email addresses. To stop the same person collecting rewards with several addresses, we compare a new address with existing accounts in a normalized form (for example ignoring “+” tags and dots in Gmail addresses), and keep one-way scrambled (hashed) forms of a deleted account's addresses so they can't rejoin through a referral link. If we offer a share of friends' purchases as money, choosing it asks you to confirm you are 18 or older and live in the United States and to give a PayPal email or Venmo username; we use these only to send payouts and keep records of payouts made. We may ask for tax information before paying larger amounts.
Classes: an instructor's section can join with a class code or class link instead of an invitation per student. A class has a name, an optional instructor name, a number of seats, an optional end date and, optionally, a school email domain (for example college.edu); for a domain-limited class we check that the address you signed in with is in that domain. When you join a class, at sign-up or later in Settings → Classes, we record which class, when you joined, and whether you joined at sign-up or from Settings. If you leave a class, we keep the record with the date you left, so your seat is kept if you rejoin; the class disappears from your Settings and from the class's members and totals.
Instructors don't have access to parlor study's records: joining a class doesn't share your notes, answers, feedback, conversations, reading or study time with your instructor or with classmates, and students can't see who else is in a class. Study time is never counted for a class. Our administrators can see a class's members (email address, name, when and how each joined, last sign-in, and how many paid AI actions each has used since joining) and class totals (members, and how many signed in and used AI recently), to run the class, keep it within its seats and answer the instructor's questions. We may tell the instructor class-level totals, such as how many students joined and how many used parlor study in a given week. We don't tell instructors who joined, who was active, or anything you studied, and we don't share a total for a group so small that it would point to particular students.
Some private previews use a shared password instead of personal accounts. Those previews share a workspace and should not be treated as private from other people given the same access.
What we save and why
| Information | Purpose | Retention |
|---|---|---|
| Saved lecture and reading text, course details, questions, answers, feedback, review dates and source references; lecture bookmarks, highlights and notes | Your notebook and study history | Until you remove the records or delete your account |
| Book conversations and their recaps, reading positions, bookmarks, book highlights and notes, saved ideas and reflections, and recorded comprehension results | Resume reading, discussion, practice and reflection | Until removed through the relevant controls or account deletion |
| Account information (email address, name, sign-in methods) and redeemed invitation information | Sign-in, access and support | Until account deletion, except the deletion marker below |
| Saved settings: reading and discussion preferences, reader layout, and which AI-sharing and feedback notices you agreed to (by wording version) | Bring your choices back on any browser after you sign in | Until you change them or delete your account |
| Demo account details: that the account is a demo, its end date, the demo link used and any full-access request | Apply the demo's limits and handle requests for full access | Until account deletion |
| Referral records: your referral code and link settings, which account joined through whose link and when, qualification status and reward credits | Run referral links and give rewards | Until deletion; if one side deletes their account, the record stays for the other side without the deleted account |
| A scrambled (hashed) form of a random browser ID, with first and last sign-in (not recorded when you choose “This is a shared computer”) | Only to hold a referral for our review when the friend signed up in a browser the referrer signed in from | 90 days after that browser's last sign-in, or account deletion |
| Class memberships: which classes you joined, when and how (at sign-up or from Settings), and when you left one | Group class members, apply seats and end dates, and give the class totals described above | Until account deletion; a class keeps only its count of seats used |
| One-way scrambled (hashed) forms of a deleted account's email addresses | Prevent rejoining through a referral link after deletion | Kept with the deletion marker below |
| Referral money choice, age and country confirmations, PayPal email or Venmo username, earnings and payout requests (when offered) | Send payouts you ask for and keep payment records | Until deletion; payouts already made are kept as payment records without your account or payout handle |
| Sign-in codes and code-request records, stored only in scrambled form | Email sign-in and preventing abuse | Codes: until used, replaced or 10 minutes; request records: one day; both are removed with the account |
| Reports sent with “This feedback seems wrong,” including the previewed answer, question, feedback, excerpt and optional comment | Human review of reported feedback | 90 days, or earlier through account deletion |
| Active study time: daily totals per activity (reading, discussion, reading practice, lecture notes, lecture practice, maps) and per book, lecture or reading material | Show you your study time | Until you delete it in Settings or delete your account |
| Reading events, such as book/chapter opens and glossary terms selected | Reading insights and service operation | 180 days; saved reading positions and history have the retention above |
| AI allowance, credits, redemptions, request counts, costs and retry fingerprints, and which book or lecture each AI request was for | Show your allowance and how you have used it, apply limits and avoid duplicate charges | Until account deletion; separate daily installation usage counters are kept for 35 days |
| Non-content operational totals and sanitized error records | Reliability, security and capacity planning | Kept as operational records; they do not contain your study text |
Study time: parlor study counts active study time only while its tab is open, in front, and you are reading or working (a recent key press, click, scroll or selection; read-aloud; or waiting for a reply you asked for). It is stored as daily totals, not a timeline. Only you can see it: it is never shown to instructors, including through class codes, and never used for feedback, grading or credits. You can turn it off in Settings → Study time, which stops recording on every device, and delete your history there. Time spent on a lecture you later delete stays in your totals as “Removed material”. Our administrators see only totals across all students, never yours. Your study time is included in your data export and deleted with your account.
Pilot statistics: to see whether parlor study works, we combine the records above into totals across students: how many joined and how (class code, referral, invitation), how many practised in their first week and came back in later weeks, average AI use and AI cost per active student, median study time, and, from later lecture checks answered without help, how often the feedback found the explanation supported by the notes (the app's own feedback, not a grade). A figure for a group of fewer than five students is not shown. We may share these totals with schools, partners or prospective investors, never with names, email addresses, study content or any one student's figures. Our administrators can leave an account out of these totals (for example a tester's); demo accounts are never included. When you delete your account, your records are deleted; totals already worked out for past weeks are kept as numbers only.
Saved browser content is linked to your account or preview workspace. Deleting a lecture does not necessarily delete a separately submitted feedback report or a separately saved reflection. Full account deletion covers the associated reader records held by parlor study.
Browser storage, cookies and imported files
A necessary sign-in cookie keeps your session active for up to seven days. If you choose “This is a shared computer” when signing in, the cookie lasts only until the browser closes, and the session also ends after 2 hours without use. Signing out revokes that session; deleting an account revokes its sessions. Signing in also sets a cookie with a random browser ID (up to 400 days); our server keeps only a scrambled form of it, as described in the table above, and uses it only to review referrals. Your reading and discussion preferences, reader layout and AI-sharing choices are saved with your account, so they come back when you sign in on any browser (see the table above); the browser keeps a working copy, which signing out removes. The light or dark appearance, voice and speaking rate stay in the browser. Practice drafts, new-lecture drafts and unsent feedback reports may also be stored locally; some are limited to the current tab. These unsaved drafts are not all part of the server export. Save or copy important unfinished work before clearing browser data.
Original lecture PDFs, photos of pages and page images stay in the browser during source review rather than being uploaded as complete files. Saving a lecture sends the accepted text and source/figure information included in the saved notebook to our server, including a small image of each figure you keep, cut from its page so it shows with your notes. Improve Formulas and handwriting transcription separately send each selected rendered page image to an AI provider only after its sharing permission. Refreshing or leaving source review can lose the original file and unfinished review.
The public parlor study pages (the home page, guides, FAQ and invitation form, not the signed-in app) may use Cloudflare Web Analytics, which counts page views, pages, referring websites and browser and device types without cookies or other storage on your device. Those pages also keep the details the invitation form records about how you arrived in your browser's session storage for that tab; they are sent only if you submit the form.
The company website at neveltrade.com has no advertising or sign-in cookies. Cloudflare delivers it, may use necessary security cookies, and counts visits with Cloudflare Web Analytics (page views, pages, referring websites and browser and device types) without cookies or other storage on your device. Its pages and the study app serve their fonts themselves, with no request to a font service. Google sign-in, the Cloudflare Turnstile check on the sign-in page and the feedback service may use their own cookies and storage when used.
AI processing and your choices
AI requests go through our server. The study app asks for the applicable permission before sharing your content. Settings → Privacy lets you change the book/reading permission and remembered lecture-sharing choices; they are saved with your account and apply wherever you sign in. Choosing not to use AI does not stop you from reading or keeping a notebook. Withdrawing permission stops future covered requests; it cannot recall a request already sent.
| Feature | Content shared | Recipient |
|---|---|---|
| Book discussion and conversation recaps | Your messages, relevant recent conversation and book context; a recap uses up to the latest 40 saved messages | OpenAI |
| Reading questions, answer checks, summaries and Revisit | Relevant source text, questions, answers, saved ideas or reflections; chapter summaries may include your book highlights | OpenAI |
| Lecture questions and feedback | Reviewed section text and questions/answers; a targeted later question also uses the earlier question, answer and feedback | OpenAI; Anthropic processes section text to build a notes map |
| Lecture overview map and Split by ideas | An overview map sends the text of every part of the lecture; Split by ideas sends long pasted notes before you save them, and only where each part starts and its title come back | Anthropic for the overview map; OpenAI for Split by ideas |
| Improve Formulas and page transcription | Each selected page image, re-encoded for transcription; you can choose several handwritten pages at once | OpenAI for printed or typed pages; Anthropic for pages you mark handwritten |
Demo accounts: every AI feature in a demo account sends the content described above to OpenAI, using a lighter model, including the uses listed for Anthropic, except where the full model's notes map for the same text already exists and is reused. Handwritten page transcription isn't available in demo accounts. Answers and feedback from the lighter model can contain more mistakes.
Lecture text-sharing and page-image-sharing choices are separate. A remembered permission can skip a later confirmation; you can ask to be prompted again. Public-book-only prompts and glossary requests do not need your personal writing. Source processing may keep a temporary notes map in server memory for up to a day. An available notes-checking route also uses OpenAI, although the current interface does not call it.
We use business APIs. OpenAI and Anthropic state that API content is not used for model training by default. Their retention depends on the service, configuration and legal or safety exceptions; ordinary API handling can include retention for up to 30 days. Deleting content in parlor study does not automatically delete provider records. See OpenAI’s API data controls and Anthropic’s API retention information.
Safety checks: submitted discussion and practice text may also be checked by our server and OpenAI for indications of intended self-harm, to offer support resources instead of ordinary feedback. We keep operational counts of these interventions, not a separate archive of the safety-check text. Content you save normally follows the retention rules above.
AI feedback can be wrong. It is not an official teacher's grade or a decision about eligibility for education or other services.
Help, feedback and email
Help: sending a Help message sends the text you enter and an opaque reader reference through our server to our Dify-based support service and OpenAI. We do not automatically attach your notebook, answers, current page, name or email. The current conversation is held in the browser view and clears on refresh or New chat; Dify and OpenAI may retain separate support records. New chat and parlor study account deletion do not delete those records. Contact us for a separate support-history request.
Feedback: opening Feedback, after a notice, shares your account ID, email address and name with our feedback service, Quackback, which we run ourselves at feedback.neveltrade.com on a server rented from Hetzner. It stores the feedback, votes, comments or files you choose to submit there. We do not automatically attach study content or screenshots. Posts may be visible to other people with access to the feedback portal. Its records and sessions are separate: signing out of parlor study does not sign you out of a separately opened portal, and parlor study export/deletion does not include feedback-portal records. Contact us to request their export or removal separately.
Feedback reports inside practice are different from portal posts: their previewed content goes to parlor study for human review and follows the 90-day retention above, without an AI-provider request.
Email: messages to info@neveltrade.com are routed through Cloudflare Email Routing to our Google-hosted mailbox. We receive the sender address, message and attachments you send, and use them to handle your inquiry. Email and external support/portal records do not currently have a single verified automatic deletion period; ask us about retention or removal before sending sensitive information.
Hosting and other service providers
- DigitalOcean: hosts the study application server and managed PostgreSQL database in the United States. It processes the saved account and study data needed to run the service, along with connection and infrastructure information.
- Cloudflare: hosts and protects the Nevel Trade company website, sits in front of parlorstudy.com, routes our contact email, runs the Turnstile check on email-code requests, and provides the cookieless Web Analytics described above. It can process connection information such as IP addresses and request details.
- Google: provides sign-in when offered and our email mailbox. Sign-in shares identity information, not access to your Google files.
- OpenAI and Anthropic: process the AI content described above.
- Resend: delivers sign-in code emails. It receives your email address and the message containing the code.
- Dify-based support: handles optional Help conversations as described above.
- Hetzner: hosts the server running our own feedback service (Quackback) described above.
- Project Gutenberg: supplies public books and covers. Direct cover requests from your browser disclose ordinary connection information to Gutenberg.
- Telegram: receives service alerts for our operators, such as errors and spending totals, without your study text or reader identifier. It also receives a short notice of each new invitation request with the role, how the person heard of us and the link source, of each sign-up through a referral or demo link, of each class join (the class name, its number of members and seats used, without who joined), each referral held for review, each full-access request and each referral payout request, without email addresses or payout details.
Hosting and external providers can retain security logs or backups separately from live application records. Deleting live data does not instantly erase backup copies. We have not established one retention period covering every infrastructure log, backup, support system and provider record, so we do not promise immediate deletion from all of them.
Read-aloud features
Read aloud and spoken replies use your browser or operating system's speech service. Depending on the chosen voice, that service may process the text locally or through its provider. parlor study does not receive an audio recording from this read-aloud feature. Text entered using your device's own dictation follows that device provider's settings and privacy terms.
Access, export and deletion
- Download: Settings → Your data → Download my data provides the account and reader data included in the web export, including your saved settings, demo account details, your class memberships, your referral code, your referrals as status and dates (not who joined), your rewards, the browsers you signed in from (as dates) and any referral payout details. It does not include all unfinished browser drafts, original files, or external Help/portal history.
- Clear selected data: Settings offers separate controls for conversations, reading history and study time; these do not delete your whole account.
- Delete account and data: where personal accounts are enabled, this removes your account and associated reader records, notes, practice, history and credits, and signs out your sessions. During the pilot, the same Google account or email address used for email-code sign-in cannot register again after deletion.
- Invitation requests: to have a request deleted before its 12 months are up, email us from the address you used.
- Ask us: email info@neveltrade.com for access, correction, deletion or help with records held separately. We may ask for information needed to verify ownership; do not send passwords or sign-in tokens.
We retain an opaque deletion marker to reject delayed writes and prevent re-enrollment under the deleted identity during the pilot. Deleting browser storage or signing out does not delete server data. Depending on your location, you may also have rights to portability, restriction or objection, withdrawal of consent, and complaint to a data-protection authority.
Age
parlor study is intended for people aged 13 and older, not children under 13. If you believe a child under 13 has provided personal information, contact us so we can investigate and remove it. Additional local age or parental-consent requirements may apply.
Security and international processing
We use encrypted connections, authenticated sessions and account ownership checks to protect access. Application diagnostics are designed to exclude submitted study content; infrastructure providers can still handle ordinary connection logs. No service can guarantee absolute security.
We operate in the United States. Our hosting, AI, identity, email, support and feedback providers may process information in the United States and other countries. Using the service from elsewhere can involve international transfers. Contact us for details about the providers or safeguards relevant to your request.
Changes and contact
We will update this page and its date when the service's data handling changes. Changes to AI-sharing disclosures may require renewed permission in the app.
Use of parlor study is also governed by our Terms of Use.
Nevel Trade Inc · info@neveltrade.com