Подробное руководство пока доступно на английском языке.
Practical reference
Practical privacy guide for account and writing data
The policy above describes the service. The following explanations help you understand its actual data flows and the choices available in the interface. They do not promise unverified provider retention rules or replace a provider’s own privacy terms.
Understand the different places your data can exist
There are several separate data flows in the application. Information entered in the writing form is used to prepare an AI request. Generated text is available in account chat history and can be exported to your device. Signed-in generation requests and responses are stored automatically in private conversation history; profile preferences are stored when you save them. Downloaded text and PDF files are stored by your browser on your device. These locations have different purposes, so deleting or changing one copy does not automatically change every other copy.
For instance, editing a generated response does not rewrite a PDF you already downloaded or the original response stored in chat history. Copy or export editor changes before leaving. A research conversation stores its questions, response and available source links; this is different from storing the raw resume upload. Extracted attachment text can be included in the saved generation input. When considering privacy, think about each copy you created and who can access that location. Signed-in conversations persist in the database until deleted, so do not assume that closing a browser deletes writing you submitted.
Read privacy →Provide the minimum necessary personal context
A useful writing request normally needs a name or preferred signature, relevant background and opportunity details. It rarely needs a full postal address, a national identity number or detailed financial information. Before using the service, decide which facts the reader needs to assess your application. Replace unrelated personal details with a short description of the experience that matters. If you upload a document, inspect the entire file rather than just its first page, since old pages or comments may contain information you did not intend to share.
Data minimisation is also a writing technique. A shorter brief with strong evidence often produces a more focused letter than a long unfiltered history. In the profile, keep skills and experience relevant to the work you plan to do. In research notes, save the official URL and your verified observations instead of copying another applicant’s documents. In a review, use a public display name and omit confidential application details. These choices remain under your control and reduce how much unnecessary information appears in AI requests, saved writing and public feedback.
Read privacy →Protect the information you choose to enter
Application writing often contains personal information, but it does not need every detail in your records. Provide what the task requires. A relevant skill, project or qualification may be enough; passport numbers, bank account details, medical records and account passwords do not belong in a letter generation brief. If a resume contains information you do not want an AI provider to process, remove that information before uploading or use a concise text summary instead. Check screenshots for private addresses and unrelated messages.
Account access is managed through Neon Auth, and protected account APIs verify the signed user identity before reading or changing stored writing. The server limits document operations to that identity. These controls reduce accidental cross-account access, but no online service can promise perfect security. Use a password you do not reuse elsewhere and sign out on shared devices. Keep downloaded letters somewhere you control. If you are unsure whether another person’s information may be included, obtain their permission or leave it out. Privacy-conscious preparation also makes a writing brief shorter and clearer.
Read privacy →Consider third-party processing separately
The application uses third-party services for hosting, account authentication, database storage, AI processing and crypto payment handling when checkout is enabled. The current generation routes may use Puter and Google Gemini, depending on the available workflow. Neon provides authentication and database infrastructure, while NOWPayments is the integrated crypto payment provider. Sending a writing request to an AI service is different from saving a draft in the chat history. Review the relevant provider’s terms and privacy information if their processing matters to your use case.
Do not assume that our interface controls a provider’s entire retention or training policy. Those practices can depend on the provider, account type and request route. The application should not claim that every provider deletes requests immediately unless that behaviour has been established. If you need a particular data-processing agreement or organisational approval, verify it before submitting sensitive work. You can reduce exposure by providing a concise factual summary instead of a complete document. Provider availability may change; use the current product behaviour and policy rather than relying on an old model name from an earlier planning discussion.
Read privacy →Manage browser-local drafts on shared devices
Browser-local saved sessions help you return to previous writing in the same browser. They are not a private account vault and do not automatically synchronise across computers. Anyone who can use the same browser profile may be able to open locally saved drafts. Private browsing modes or browser cleanup can remove local storage, and changing browsers can make it look as if your drafts have disappeared when they are simply stored elsewhere. Keep an exported copy of work you cannot afford to lose.
Chat history automatically stores new signed-in generation requests and their responses for reopening on another device. Existing local drafts from an older version are not uploaded or migrated automatically. The current dashboard does not offer a separate saved-session manager. To tidy up a shared browser, use its site-data controls and your device file tools, checking for an exported copy first. Signing out and deleting browser-local writing are separate actions. A downloaded PDF is another separate file. Understanding these distinctions prevents accidental disclosure and mistaken expectations about recovery or synchronisation.
Read privacy →Keep login and payment credentials private
Your account password is used by the authentication service, not as part of a letter prompt. Never place it in custom instructions, a saved document, a review or a support message. The browser obtains a short-lived signed token for protected account requests, and the server verifies that token before accessing account data. A valid identity still does not authorise reading another user’s documents. Resource ownership is checked in the database operations as well as in the interface.
Crypto wallet access details, private keys and recovery phrases are unrelated to the application’s writing features and should never be shared with it. The payment integration creates provider checkout information without asking you to disclose a wallet recovery phrase. If a payment or account error appears, describe the problem with a non-sensitive order reference or screenshot that hides credentials. Public configuration such as the application’s auth service URL is different from a private API key. The application’s provider secrets belong on the server and must not be copied into public content or frontend fields.
Read privacy →Organise conversations and generated responses
A useful conversation history makes generated drafts easy to identify. Give each item a title that includes the organisation or programme and the purpose, such as “Analyst role — autumn application” or “Masters programme — requirements notes.” The sidebar lists private conversations for cover letters, university applications, professional emails and research. Open a conversation to restore its generated response, rename it to identify the opportunity, or continue it with a new request. Older conversations load in pages of twenty; the result page can load earlier messages. There is no separate manual Saved drafts library or Save to account action.
New generation requests automatically save their brief and response to private chat history. Open a conversation from the sidebar to restore a completed response, or continue it with a new request. Existing browser-local copies are not silently uploaded. Research notes should include the source and what you verified so that they remain understandable weeks later. Delete drafts you no longer need, and download an important final document for your own records. A saved draft is a working document rather than proof that an application was sent, accepted or reviewed by an employer. Use your own submission records to track those events.
Manage your writing accountReuse a profile without losing personalisation
The account profile can hold your name, skills, experience, preferred output language and preferred tone. Reusing those details saves repeated typing, especially when you are preparing several applications. Keep the profile factual and current. A useful skills entry describes tools or capabilities you can demonstrate; a useful experience entry gives the setting and your contribution. Avoid putting passwords, identity document numbers, banking information or unrelated confidential material into the profile fields.
A profile is a starting point, not a universal letter. The opportunity still changes from one application to another, and the most relevant example may change with it. Use the profile in the generator, then select the background that fits the current reader. Review stored experience after a course, role or project ends so old information does not accidentally appear as current employment. If you share a device, sign out after using your account and remember that browser-local drafts may remain on that device. Account settings and browser storage serve different functions and should be managed separately.
Manage your writing accountManage the copies you have saved
Chat history provides a delete action for a selected conversation. The server applies that action only to a conversation belonging to the signed-in user and deletes its associated messages. It does not delete a downloaded file or older browser-local data. If you want to remove those copies too, use your browser site-data controls and your device file management tools. A conversation can contain several responses, so check its earlier messages before deleting it. Take care not to delete the only final copy of writing you still need.
Account-wide deletion requests are handled through support rather than an automated account-deletion button in this version. Contact support@ailettergen.com from the account email and describe the request without including your password. Payment records may need different handling from writing drafts, and provider or infrastructure backups are separate from the active library. This page does not promise an immediate universal purge across every third-party system. Ask about the specific information you want removed if you need clarification. Clear storage descriptions are more useful than a blanket claim that every copy disappears when a browser tab closes.
Read privacy →Turn experience into useful evidence
A strong letter explains what you did, the situation in which you did it and why it matters to this reader. Instead of “I have excellent communication skills,” describe how you answered recurring customer questions, organised a handover or explained a technical issue to a colleague. Numbers help when you can verify them, but they are optional. A precise description without a percentage is more credible than a dramatic metric you cannot support. Projects, volunteering, coursework and responsibilities outside paid work can also provide relevant examples.
Try making a three-column note before generating: requirement, evidence, connection. For a role asking for attention to detail, your evidence might be checking weekly inventory records; the connection is that you can identify inconsistencies before they reach a customer. Include two or three strong connections rather than every task you have ever completed. After generation, check that the draft preserves the scale of your contribution. “Helped prepare” must not quietly become “led the department.” You should recognise your own experience in every paragraph.
Read guides →Use only information you are entitled to share
A resume, screenshot or programme page can contain information about other people. You should have the right to provide the material you use and should avoid uploading confidential employer records, private messages or another applicant’s documents. If a useful example comes from collaborative work, describe your own role without claiming sole ownership of the whole result. If you quote a requirement from a vacancy, use it to explain the application rather than reproduce a large copyrighted document unnecessarily.
The same principle applies to generated writing. Review any named person, organisation, quotation or factual claim before sending it. The model may produce a plausible detail that was never present in your source material. Remove it unless you can establish that it is correct and appropriate to use. For academic applications, follow the institution’s rules about outside assistance and AI use. The service is intended to support honest writing from information you provide; it should not be used to impersonate another person or misrepresent an application.
Read terms →Check university research against official sources
University information can change between application cycles. Deadlines, tuition, scholarship conditions, document requirements and programme availability should be confirmed on the institution’s own website. An AI research summary is a convenient starting point for organising information, but it is not an official admissions decision. Open the source links, check the academic year and note whether the information applies to your nationality, qualification level and study mode. If a summary has no source for an important claim, treat that claim as unresolved.
Use research notes to record the programme URL, the question you were trying to answer and the information you verified. Keep a distinction between a published fact and your own interpretation. “The page lists a language test requirement” is different from “I meet that requirement.” If two official pages conflict, contact the admissions office rather than asking an AI model to choose one. Saved account notes help you revisit the evidence later, but you must check it again before relying on a deadline or fee. Research access shown in a planned paid package should not be mistaken for an admissions guarantee.
Read features →Choose between PDF, text and copied content
Use a text download when you want to continue editing in another application or paste the wording into a form. Use the copy action when the recipient expects text in an email or portal. Use PDF when the application accepts a document attachment and you want a stable visual layout. The available export does not decide what the recipient requires. Check the employer or university instructions before choosing a format, and preview the downloaded file rather than assuming it matches the on-screen appearance perfectly.
For long letters, inspect page breaks and confirm that the closing and signature remain readable. For non-Latin languages, check the characters in the actual PDF and make sure nothing is missing. Keep a sensible filename that identifies the application without exposing unnecessary personal information. Do not upload a file with another employer’s name in it. A downloaded copy is separate from the browser draft and the account copy; later changes to the editor do not automatically change a file you already downloaded. Export again after your final edits and attach the correct version.
Open the writing workspaceExpect product details to be checked over time
Product availability, model routes, provider pricing and university information can change. The application should distinguish a current feature from a planned one, and comparison pages should identify the selected plans and source links. A historical research note is useful context, but it is not proof that a price or requirement remains unchanged. Before relying on a comparison for a purchase or a research summary for an application, open the current official source and confirm the relevant detail.
The privacy policy and terms may also be updated to reflect changes in storage, account features or service delivery. Review those pages when a new feature changes the way you use the application. If a description appears inconsistent with the actual interface, ask support for clarification and avoid submitting unnecessary sensitive information while the issue is unresolved. This is a practical way to use an evolving service without treating old screenshots or planning discussions as a permanent product specification. Saved drafts remain your working text; they should be rechecked when the application context changes.
Read about →