Privacy Policy

1. Who we are

Requt is operated by Avinu Servicos de Tecnologia LTDA, a company in Brazil (“Requt”, “we”, “us”). We are the controller for the personal data of Requt account holders described in this policy. Contact: comercial@avinu.tech. We answer every privacy question and request at that address, including requests from people who do not have a Requt account (section 6).

EU representative. We have not appointed a representative in the European Union. Our assessment is that the exemption in Article 27(2) of the GDPR applies to Requt at its current scale: our processing is occasional, is unlikely to result in a risk to people’s rights and freedoms, and does not include special categories of data on a large scale. We will revisit this as use of Requt in the EU grows.

Data protection officer (encarregado). We consider Avinu a small processing agent under ANPD Resolution CD/ANPD nº 2/2022, which exempts small processing agents from appointing an encarregado under the LGPD. The address above is our channel for communication with data subjects and with the ANPD.

Candidates entered by a career coach. If you are a career coach and you add a candidate who is not you, you decide what to upload about that person and why. You are the controller for that person’s data, and we process it on your behalf under our Terms of Service (section 5). Section 6 explains how the person you coach can reach us.

2. What we collect

• Account data — your email address and sign-in identity (a password or Google sign-in), managed through Supabase. We keep a copy of your email address from your sign-in token and update it when it changes.
• Profile data — the optional onboarding answers (name, how you use Requt, target role, experience level, interview timeline and how you heard about us).
• Content you provide — job descriptions, candidate details, intake and draft answers, and videos you record or upload. Recordings are sent to our transcription and analysis providers when you submit them for analysis. The recording studio does not stream your microphone audio for live captions.
• Candidate details — a name, the file name and text of an uploaded resume, the structured details we derive from it, and intake answers. If you are preparing your own interview, this is your own data; if you are a coach, it describes the person you coach. We keep the extracted text and details, not the resume file: the file is read in memory and never written to storage. Candidate records do not include an email address.
• Generated data — outlines, transcripts, scores and feedback produced for you.
• Usage and billing data — your credit balance and ledger (including records of past live-caption sessions, a feature we have retired), purchase and subscription records from Paddle, and a minimal hashed identity marker used to block access after an account is closed. Card numbers are handled entirely by Paddle and never reach our servers.
• Browser and technical data — browser storage keeps you signed in and remembers your analytics choice, your recording preferences and, while you finish signing up, the version of the terms you accepted. Our server keeps a one-minute request counter per IP address to limit abuse, and our systems keep technical logs (section 5). Our service providers may also process IP addresses, request information and diagnostic logs to operate and secure the service. Clearing browser storage may sign you out or reset preferences; it does not delete your account data.

We also keep records of your choices and of what we have done:

• Analytics consent record — whether you allowed optional analytics, and when you last said yes.
• Terms-acceptance record — the version of the Terms of Service you accepted when you signed up, and when. Accounts created before we began keeping this record have none, and we do not create one for them.
• Candidate-data attestation — each time you add a candidate record, your confirmation that you may give us that person’s details and that they know Requt processes them with AI, with the version of the text you confirmed and when.
• Revocation marker — when an account is closed, a one-way hash of its sign-in identity, with no name or email address. It is checked on every request so that an old sign-in token cannot reopen the account, and kept until no backup could bring the account back.
• Deletion register — when we erase an account, a restricted record of what was deleted (internal identifiers and storage keys, not the content), so that the same deletions are applied again if we ever restore a backup.
• Request and compliance records — a minimal record of each privacy request and what we did, and our records of copyright notices and security incidents.

3. Why we process it and on what legal basis

The table sets out each purpose, the data it uses and our legal basis under the GDPR (Article 6(1)) and Brazil’s LGPD (article 7).

Candidate records a career coach enters about someone else are not in this table: the coach is the controller for that data and relies on their own legal basis, and we process it on the coach’s behalf (section 1).

Recordings are processed to give coaching feedback, not to identify anyone: nothing in Requt extracts or stores a face or voice template. Please leave health, religious and similar sensitive details out of resumes and recordings; if they are there, we process them only as part of the service you asked for.

Scores and feedback are advice to you. Requt takes no decision about anyone: it does not rank people, reports nothing to employers, and nothing it produces has a legal or similarly significant effect on you, so Article 22 of the GDPR does not apply. You can still ask for a person to review any feedback (section 6).

We do not sell personal data.

4. Who processes it for us

We share data with service providers only as needed to run each feature. For each provider we say what it does, what it receives, its role and where it processes data; section 8 covers the safeguards for transfers.

• Contabo — hosting. The server that runs Requt: the database, the private recording store, temporary processing files, our backups and our technical logs. Contabo’s automatic backup service also keeps copies of the whole server, stored away from it, for up to 10 days. Role: processor. Location: France (EU).
• Supabase — sign-in. Your email address, password or Google sign-in identity and your sessions are held by Supabase Auth; our server only verifies your sign-in token. Supabase’s built-in mailer sends the emails that confirm a new account and reset a password. Recordings made before our current recording store went live were kept in Supabase Storage; none remain there, and new recordings never go there. Role: processor. Location: United States (US West, Oregon).
• Google Workspace — email. Hosts our comercial@avinu.tech mailbox: privacy requests, copyright notices and other messages to us arrive there, and we send our replies and inactivity notices from it. It holds what you write to us, your email address and anything you attach; section 5 says how long we keep it. Role: processor. Location: Google’s global infrastructure, including the United States; section 8 describes Google’s data processing terms.
• OpenRouter, with Google and Amazon Web Services — outlines and video analysis. Resume text, job details, intake answers, the candidate’s name, the outline, the transcript and, for an analysis, the video itself are sent to OpenRouter, which passes each request to a cloud service hosting the model: outlines use Anthropic’s Claude models served by Amazon Bedrock or Google’s Vertex AI; reading resumes and analyzing videos use Google’s Gemini models served by Vertex AI. Video analysis goes only to Google’s Vertex AI. No email address or account id is sent. Every request tells OpenRouter to use only endpoints that do not use data for training and keep no copy of the prompt or the output after responding (zero data retention); a request that cannot be routed that way fails rather than falling back. With that setting Anthropic’s own service is not in the path: the models run on Amazon’s or Google’s infrastructure. This controls where a request is processed; OpenRouter’s own handling of requests is governed by its terms and our account settings. Role: OpenRouter is our processor, with Google and Amazon Web Services as its sub-processors. Location: United States — OpenRouter, Google and Amazon are US companies, and a request may be served from their data centres in other regions.
• AssemblyAI — transcription. The video file of a recording you submit for analysis is sent for transcription. After each attempt we ask AssemblyAI to delete the transcript, and we keep retrying until it confirms. The uploaded media copy, and what AssemblyAI may do with it, are governed by AssemblyAI’s own terms and retention rules. Role: processor. Location: United States.
• Apify — Loom imports. When Loom imports are available, the public Loom link you paste is sent to a downloader that runs on Apify to fetch a copy of the video; we then delete the run’s storage and retry until that succeeds. Loom imports are currently disabled, so no link or video is sent to Apify. Role: processor. Location: Apify is based in the Czech Republic (EU).
• PostHog — optional analytics, only if you allow it. Page categories, actions such as recording or checkout, broad campaign-source categories, and analysis and payment outcomes, linked to a random browser identifier or your internal account id. We do not send names, email addresses, videos, transcripts, resumes, scripts, scores, feedback or signed video links. Session replay and automatic interaction capture are off, and so is location lookup from your IP address, although PostHog still sees your IP address when your browser connects. The analytics library loads only after you say yes, and our server checks your choice again before every event it sends. You can change your choice at any time with the Analytics preferences button; withdrawal stops future collection and does not itself erase earlier events, which you can ask us to delete. Analytics browser storage is optional and separate from essential sign-in storage. Role: processor. Location: United States.
• Grafana Labs (Grafana Cloud) — monitoring. Aggregate error counts, operation timings, sampled technical traces and restricted failure logs from our server. We exclude interview content, resumes, user ids, credentials, web addresses and payment details from this telemetry, and your browser reports errors only to our own server, never to Grafana. Role: processor. Location: United States.
• Paddle — payments. Checkout, payment processing, receipts and tax handling as merchant of record. We pass Paddle the price, your email address when we know it and your internal account id; you enter your name, address, tax id and payment method in Paddle’s own checkout, and none of that reaches our servers. Paddle sends us transaction and subscription ids, status and totals. Role: Paddle is an independent controller for its payment, tax and fraud-prevention obligations; its privacy policy explains how and where it handles your data (section 8).
• healthchecks.io — backup monitoring. After each successful nightly backup our server sends it one request, so that we are alerted when a backup does not run. The request contains no personal data.

5. Storage and retention

Your content. Videos you submit for analysis are saved in private storage for playback in your account; playback uses temporary signed links that expire after one hour. Saved jobs, candidates, drafts, outlines, analyses and transcripts stay available until you delete them or your account is erased. Deleting an analysis deletes its stored recording first and the analysis record second; if the recording cannot be deleted, the record is kept so that the deletion can be retried. Deleting a job or candidate also deletes the outlines and drafts generated from it, but not existing analyses, transcripts or recordings — delete those analyses separately. Generated content may repeat information from the original resume or job description, and is erased with your account.

Processing queue. A video waiting for analysis is removed from the queue once its analysis finishes or fails for good. A saved video you have not analyzed yet, or a failed analysis you can retry, keeps its file for up to 24 hours. Temporary processing files are removed after processing, and swept after one hour if a job is interrupted. The queue’s records of finished jobs — status, file name and timestamps, not the video — are deleted after 90 days.

Your account. Your account, profile and the records in section 2 are kept for as long as your account exists. You can ask us to erase your account at any time (section 6).

Inactive accounts. If you do not sign in for 12 months, we email you from comercial@avinu.tech. If you still have not signed in 30 days after that email, we close your account and erase it the same way as on request, with the same exceptions for financial and compliance records. Signing in at any time before then keeps your account open. An account with an active subscription or unused purchased credits is not closed for inactivity. Candidate records a coach entered stay in the coach’s account until the coach deletes them or the coach’s account is erased.

Financial records. Purchase and credit-ledger records, Paddle transaction and subscription ids and refund records are kept for 5 years after the end of the fiscal year they relate to — the limitation period for Brazilian tax and accounting records — including after your account is erased. After erasure they are linked only to an internal account record that holds no name or email address (it keeps the version of the terms you accepted), and they are deleted with it at the end of that period. Paddle keeps its own transaction records under its own obligations.

Compliance records. The deletion register, our log of privacy requests, and our records of copyright notices and security incidents are kept for 5 years after the request, notice or incident is closed, then deleted. They hold identifiers and a record of what we did, not the content itself.

Email to us. Messages you send to comercial@avinu.tech stay in our mailbox (section 4, Google Workspace). A privacy request or a copyright notice is kept for 5 years, together with our record of how we handled it; any other support email is deleted after 2 years.

Technical logs. These are the logs Requt keeps itself; providers keep their own. None of them contains your recordings, transcripts or resume text.

• Per-IP request counters: one minute, in the server’s memory only.
• Application and worker logs: operational messages that can include your internal account id and provider ids when a step fails, but no IP addresses. Each container keeps at most about 30 MB of logs, dropping the oldest first, and the logs are discarded whenever the container is replaced.
• Reverse-proxy log: the proxy in front of Requt logs requests that fail, with the client IP address, the address requested and the request headers. That log is not yet size-limited; it is cleared whenever the proxy is redeployed.
• Backup log: the names and counts of the files each nightly backup writes, any errors and, when copying a recording fails, its storage key, which contains an internal account id. Kept for 30 days.
• Sign-in logs: sign-in events and IP addresses, kept by Supabase under its own retention settings.

Backups. Every night we back up the database and the recordings to our own server. Backups are kept for 30 days and used only for recovery. If backups start failing, we keep the newest database backup and recording snapshot beyond 30 days so that there is always one copy to restore from, and we review that copy by hand if backups have not recovered within 7 days. Separately, Contabo’s automatic backup service keeps up to 10 daily copies of the whole server, deleting the oldest as each new one is made.

What deletion means for backups. Deleting something in the app, or erasing your account, removes the data from the live service; it does not erase it from backups at once. Deleted data leaves Contabo’s copies within 10 days and our own backups within 30 days, subject to the exception above. If we ever restore a backup, we re-apply every account erasure and every copyright removal made since that backup was taken before the service reopens.

Provider copies. We ask AssemblyAI to delete each transcript after processing and retry until it confirms; this does not cover the uploaded media copy. For Loom imports, we delete the downloader run’s storage and retry failed cleanup. When your account is erased we also delete your analytics events at PostHog and your sign-in identity at Supabase. Otherwise, provider retention depends on each provider’s service, settings and legal obligations; contact us about provider-held copies.

6. Your rights

You can delete jobs, candidates, outlines and analyses directly in the app, and delete your whole account on the Account page. For anything else, email comercial@avinu.tech. You can ask us to:

• Access — confirm whether we process your data and give you a copy of it.
• Correct — fix data that is wrong or incomplete. You can also change profile fields in the app.
• Delete — erase your account and content. You can do this yourself on the Account page: you confirm by typing DELETE, and if you last signed in more than 10 minutes ago we ask you to sign in again first. Your account closes at once and the erasure usually finishes within minutes; an active Job Hunt plan is canceled at the same time (or we ask you to cancel it first). If you cannot sign in, email us. We keep the financial records, the revocation marker and the compliance records described in section 5, backup copies age out as section 5 describes, and Paddle keeps its own records. The Account page, or our reply to your email, tells you what we keep and why.
• Export — receive your profile, content, transcripts, feedback and credit history in a machine-readable file, and your recording files if you ask for them.
• Restrict — limit what we do with your data while a dispute about it is settled.
• Object — object to processing we base on legitimate interests (section 3).
• Withdraw consent — for optional analytics, at any time, with the Analytics preferences button at the bottom of every page; withdrawal does not affect processing that happened before it.
• Human review — ask for a person to review any score or feedback Requt produced for you.

If you are in Brazil, article 18 of the LGPD also gives you the right to confirmation that we process your data; access; correction of incomplete, inaccurate or out-of-date data; anonymization, blocking or deletion of unnecessary or excessive data, or of data processed unlawfully; portability to another provider; deletion of data processed with your consent; information about the public and private entities we share data with; information about your option not to consent and what happens if you do not; and revocation of consent. Article 20 lets you ask for a review of decisions taken solely by automated processing; Requt takes no such decisions (section 3), and we offer a human review of feedback anyway.

How to ask. Email us from the address on your Requt account. Before acting, we confirm that you control the account, through a signed-in session or a verification message to the account’s email address. We do not accept a user id or a different email address as proof. If you have lost access to your account, use account recovery first.

If you do not have a Requt account. A career coach may have entered your details, or you may appear in a video someone uploaded. Write to comercial@avinu.tech with the coach’s name (or the email address of the coach’s Requt account) and the name used for you in the record; we cannot search recordings by face or voice. Because the record belongs to the coach’s account and the coach is responsible for it, we confirm your request with the coach before we act on it. You can also ask the coach directly to change or delete the record.

If a coach gave us your data. If a coach added you as a candidate, we hold what the coach entered — your name, the text of your resume and the details derived from it, and intake answers — and any recordings of you the coach submitted, with their transcripts, scores and feedback. We received this from the coach, not from you. It is used only to provide Requt to the coach, processed as described in sections 3 to 5 (including by the providers in section 4), and kept in the coach’s account until the coach deletes it or the account is erased. Coaches confirm, when they add a candidate, that the person knows they use Requt.

How long we take. We answer within one month of receiving your request. If a request is complex, or we receive many at once, we may take up to two more months; if so, we tell you within the first month and explain why. Under the LGPD, a request to confirm or access your data gets a simplified reply straight away and, if you ask for it, a complete statement within 15 days.

Complaints. You can complain to a data protection authority: in Brazil, the ANPD (Autoridade Nacional de Proteção de Dados); in the European Union or EEA, the authority of the country where you live or work. We have no establishment in the EU, so no single EU authority leads on Requt. We would be glad to hear from you first.

7. Security

Data is encrypted in transit. Beyond that, these are the procedures and controls we rely on:

• Access to production systems is limited to the company’s owner and a small number of people who need it to run Requt, all bound by written confidentiality obligations.
• Recordings live in a private storage bucket with public access switched off; nothing in it can be fetched without a signed link.
• Signed playback links are created only for the account that owns the recording and expire after one hour. Anyone who has a link can use it until it expires, so do not share it.
• Every read and delete of a job, candidate, outline or analysis is limited to the signed-in owner; no account can see another account’s data.
• When an account is closed, a hashed revocation marker is checked on every request, so an old sign-in token cannot get back in.
• Requests are rate-limited per IP address, and account reads are also limited per account.
• Analytics, when you allow them, never record your screen or keystrokes: session replay and automatic capture are off.
• Payment notifications from Paddle are checked against a signature before any credit is granted.
• Backups stay on our server, in a directory only its administrator account can read.

No method of storage or transmission is completely secure. If a security incident affects your data, we follow our incident procedure. Under the GDPR, we notify the competent supervisory authority within 72 hours of becoming aware of the breach where the law requires it, and we tell you without undue delay when it is likely to result in a high risk to you. Under the LGPD, we notify the ANPD and the people affected within 3 business days when the incident may cause them relevant risk or damage.

8. International transfers

We are a company in Brazil and our server is in France (EU). Providers based in the United States also receive data: OpenRouter and the cloud services running the models behind it (Google and Amazon Web Services), AssemblyAI, Supabase, PostHog, Grafana Labs and Google Workspace. These transfers rely on each provider’s contractual safeguards: the EU–US Data Privacy Framework where the provider is certified under it, and otherwise the Standard Contractual Clauses in the provider’s data processing terms. For the LGPD, we rely on the ANPD’s standard contractual clauses where the provider offers them. Ask us at comercial@avinu.tech which safeguard applies to a given provider.

For Google Workspace, Google’s Cloud Data Processing Addendum states that it is incorporated into Google’s agreements with its customers and that Google processes customer data as a processor, on the customer’s instructions. Google states that Google LLC is certified under the EU–US Data Privacy Framework, which it relies on for European data transferred to the United States; the addendum adds Standard Contractual Clauses for transfers the framework does not cover, and Google publishes standard contractual clauses approved by the ANPD for transfers under Brazilian law.

Paddle acts as an independent controller and describes its own transfers in its privacy policy. See OpenRouter’s provider data policies, AssemblyAI’s privacy policy, Google’s Cloud Data Processing Addendum and Paddle’s privacy policy. These policies describe provider practices and do not replace your rights against Requt.

9. Children

Requt is not directed at children under 16 and we do not knowingly collect their data. If you believe a child under 16 has an account, tell us and we will erase it.

10. Changes and version history

We may update this policy as the service evolves. Material changes will be announced in the app or by email. Every version is dated, and we keep every previous text; ask us if you would like a copy.

• 2026-09-01 — first published version.
• 2026-09-10 — provider list, including Grafana Cloud and Contabo; backups; the analysis worker; deletion of AssemblyAI transcripts.
• 2026-09-11 — consent-based analytics (PostHog).
• 2026-09-15 — our position on an EU representative and an encarregado (section 1); legal bases (section 3); each provider’s role and location (section 4); retention periods, including inactive accounts, financial and compliance records and technical logs (section 5); your rights, response times and the route for people without an account (section 6); security procedures and incident notification (section 7); transfer safeguards (section 8); this history (section 10).
• 2026-09-16 — this version: deleting your account yourself on the Account page (section 6).

Earlier texts were corrected after publication without a new date: the 2026-09-10 text on 2026-09-11 (live captions), and the 2026-09-11 text on 2026-09-12 (the Apify entry for Loom imports), 2026-09-13 (the backup exception) and 2026-09-15 (Loom imports disabled). From the 2026-09-15 version on, every change to this policy gets a new date.

11. Contact

Privacy questions and requests, including from people without a Requt account: comercial@avinu.tech. This is also our channel for communication with data subjects and with the ANPD. To report content that infringes your copyright, see our Copyright page.