Privacy Policy

Version 9 · effective October 27, 2026 · last updated October 27, 2026

This policy explains what Hayatiq Financials (the Service) does with data, where that happens, who else is involved, and what you can ask of us. It is written to be read alongside the Terms of Service; for firms that process their clients' data through the Service, our Data Processing Agreement forms part of the Terms and applies automatically.

Who we are. The Service is operated by HAYATIQ AI SOLUTIONS F.Z.E (trade name as registered on the trade licence and commercial registry: HAYATIQ AI SOLUTIONS - F.Z.E), a free zone establishment registered in Ajman, United Arab Emirates, at B.C. 1306744, Ajman Free Zone C1 Building, Ajman Free Zone, Ajman, United Arab Emirates — the same entity as in the Terms (Hayatiq, we). Contact for anything in this policy, including every request about your data: privacy@hayatiq.ai. Help using the Service: support@hayatiq.ai. Anything else: hello@hayatiq.ai. Hayatiq has not appointed a data protection officer, because none of the conditions that require one applies to its processing: it is not a public entity, it does not monitor individuals' behaviour on a large scale, and it processes no sensitive data. The Service is offered to businesses in any country: the preparer chooses the company's country, and the draft carries that country's local items where the Service models them. Hayatiq has no representative in Saudi Arabia, the European Union, the United Kingdom or any other country, and Hayatiq has not registered on the Kingdom's national register of controllers. Hayatiq's own processing of account, billing, support and website data is subject to the personal data protection law of the United Arab Emirates, where Hayatiq is established (Federal Decree-Law No. 45 of 2021, the UAE PDPL); to the Saudi PDPL for people in the Kingdom; and to the GDPR or the UK GDPR where either applies to it. If a registration or an appointment becomes required, this paragraph will name the role and how to reach it.

Two roles. For your account, the checklist download, billing, support and this website, Hayatiq decides why and how data is processed: we are the controller. For the trial balances, mappings, answers and names you enter into an engagement — data about your business or your clients' businesses — you decide; Hayatiq processes it only on your instructions as your processor, under the Data Processing Agreement. For the guest download, Hayatiq processes the draft you send only to render your file, as your processor on the terms of the Data Processing Agreement.

Where the product stands. Sign-in, saving, export (Word, Excel and PDF), the download offered before signing in and paid plans are live. The assistant runs only when you allow it on an engagement (section 14). Each saved engagement snapshot, its questionnaire answers and the rounded statement figures recorded at export are encrypted by Hayatiq's application before they are stored; the engagement's name, country, period, currency and framework choice, and the metadata records beside them are not encrypted this way (section 13). The no-storage purge runs every night (section 10). Zero data retention with the assistant's model provider is not in place and has not been requested (section 14). Reading last year's issued financial statements or a commercial registration extract is offered only where the Service shows the option to read a document; the sections that describe it say "where the Service offers it".

1. The short version

• The engine runs in your browser. Uploading a trial balance, mapping it, answering the questionnaire, filling the worksheet, generating the statements, the tie-outs and the checklist all happen on your device. Nothing about an engagement is sent to Hayatiq until you take one of four explicit actions: save, ask the assistant, generate note narrative, or export (signed in, or the one download offered before signing in).
• Until you save, an engagement lives in your browser's own storage (IndexedDB), under your control; clearing site data removes it.
• When you save, Hayatiq stores canonical balance rows and a rebuildable snapshot of your choices — never the uploaded file itself — in a database in Frankfurt, Germany, isolated per organisation by row-level security; the snapshot is encrypted by Hayatiq's application before it is stored (section 13).
• Hayatiq never puts client financial data in logs or analytics.
• The assistant receives account names and codes, your answers (including any text you typed) and each note as it prints with its figures, so it can write about them; it never supplies a figure to a note's narrative — every number in a paragraph must match the engine's figures, the note's fixed allow-list or text you entered or confirmed for that note, or the paragraph is refused. What is sent is not used by Anthropic for training (its commercial terms; Hayatiq has not opted in); under the data-retention policy Anthropic publishes for its commercial API customers, Anthropic deletes it, and what it returns, within thirty (30) days of receiving or generating it, and keeps it longer only where that policy says so, such as content flagged under its usage policy, or where the law requires it (section 14). Where the Service offers it, if you choose to have it read last year's issued financial statements or the commercial registration extract, it receives that file whole, for that one reading, and proposes facts and figures from it for you to confirm or change; Hayatiq does not keep the file.
• We rely on seven providers, listed in section 11 with what each receives; six are sub-processors and one (Paddle) is an independent controller for the purchase.
• You can ask us to access, correct, export or delete your data (section 15).

2. What happens in your browser, and what reaches Hayatiq

The header of every engagement screen carries a "Where your data is" indicator that reads "In this browser only" until a save has happened, and links back to this policy.

3. The save boundary: what a saved engagement holds

When you save an engagement, the Service sends Hayatiq a rebuildable snapshot of your work, together with the version numbers of the engine, the line-item schema, the chart of accounts and the framework pack it was built with, so that the same figures can be rebuilt later. The snapshot holds:

• the engagement's details: entity name, reporting period and currency;
• canonical balance rows for both years — account code, account name and balance — together with the parse report (row counts, warnings);
• the mapping — each account's statement line, status and traits, and who set it: the role of the user who confirmed it, or the draft, where it applied the mapping as an assumption (the assistant's per-account suggestions are not stored; they are re-derived on load);
• questionnaire answers and the cash-flow policies you chose;
• the cash-flow worksheet, including confirmed derived figures, the prior-year cash-flow statement, note worksheet tables, and any equity movements you classified;
• the note narratives you kept or edited, whether you or the assistant drafted them;
• the assumptions and the release: which values the draft applied as assumptions (each one's kind and source, with no figure or name) and, once a user releases the draft for export, the release record — that user's account e-mail address, the time, and the counts of assumptions, items to be confirmed and flagged lines;
• the no-storage flag, if you set it (section 10).

The snapshot never holds the uploaded file, a document the assistant read, the assistant's per-account suggestions or a generated statement set. Facts and figures you confirm from a reading are saved as your answers and entries, like anything you type. When you export a saved Engagement, Hayatiq records that version's rounded statements, checklist results and warning codes (the rounded statements encrypted, section 13; for a draft saved in no-storage mode, only the checklist results and warning codes, so the recorded version holds no statement figures), and for each file its format, size and a hash — never the file itself, which is streamed to your browser and not stored. For a released draft, the recorded version also keeps the release's counts, never the releasing user's address. The snapshot also names each uploaded file by its file name, size and a cryptographic hash of its contents, with the layout detected and row counts; Hayatiq never receives the file itself.

Names of people can appear in a snapshot where they appear in your data: an account name, a related-party answer, a director named in a note table. That is why the snapshot, the questionnaire answers and the rounded statement figures recorded at export are encrypted by Hayatiq's application before they are stored (section 13), and why the Data Processing Agreement treats the snapshot as your personal data processed on your instructions.

4. Account and sign-in data

Sign-in to Hayatiq ID is by a one-time link sent to your e-mail address. We hold your e-mail address, the organisation you belong to and your role in it (owner, preparer, reviewer), the time of sign-in, and the session that keeps you signed in. Hayatiq stores no passwords. When you release a draft for export, your e-mail address is printed in the release statement of the files exported from it, and kept in the release record of an engagement saved after the release (sections 2 and 3). Printing the address is part of the processing your organisation instructs when a draft is released: from then on, the address in the files and in the saved release record is part of that engagement's data, which your organisation controls and Hayatiq processes for it under the Data Processing Agreement. Whoever receives the files receives the address; Hayatiq cannot recall or change a file already downloaded. Legal basis (where the GDPR applies): performance of the contract with you; PDPL basis: necessary to perform the agreement to which you are party.

An organisation's owner can invite members by e-mail address on the Members screen, within the seats of the organisation's plan. An invitation does not make anyone a member: it stays pending, and the Members screen shows it as waiting for acceptance, until the invited person, signed in with the invited address, accepts or declines it on their Engagements page. Someone who has no account yet first signs in with that address, which gives them an organisation of their own, and then accepts or declines in the same way. Hayatiq sends no invitation e-mail; the owner lets the person know. The reply to an invitation does not say whether the address is registered, and an invitation is refused when every seat is in use, whoever it names. The Members screen shows an owner each member's e-mail address, role and joining date, and each pending invitation's address and role; a member who is not an owner sees only their own role.

5. The checklist download

The landing page's disclosure-checklist download asks for an e-mail address only — no name, no company — and stores it, lower-cased, with the time of the request and the word "lead-magnet". It goes into the table that held the waitlist before the waitlist form was withdrawn; the addresses that form collected are kept there under the word "landing" on the same terms. No mail has been sent to these addresses; if Hayatiq writes to them about the product, each mail will say how to stop. We keep an address until you sign up or ask to be removed, or for twelve (12) months after the request, whichever is sooner. Legal basis: your consent, given by requesting the download; withdraw it at any time by writing to privacy@hayatiq.ai.

6. Billing data

Purchases are made from Paddle, our merchant of record. Paddle collects your name, e-mail address, billing address, tax number where relevant, and payment details, and issues receipts; Paddle is an independent controller of that data under its own privacy notice (section 11). Hayatiq receives from Paddle the events needed to run your subscription — a customer and subscription identifier, the plan, the billing period and the status — and stores only the fields it acts on. Hayatiq never receives or stores card numbers, and its record of a Paddle event holds no name, e-mail address, postal address or amount; the full record stays in Paddle.

7. Support and correspondence

If you write to us — support@hayatiq.ai for help with the Service, hello@hayatiq.ai for anything else — we keep the correspondence and the details in it for as long as needed to deal with your request and for twenty-four (24) months afterwards for reference. Please do not send trial balances or client data by e-mail; the Service is built so that you never need to.

8. Analytics and logs — no client financial data

Hayatiq measures how the product is used with Vercel Analytics, which processes those events for Hayatiq as our processor. The events it sends are named steps — upload, mapped, generated, exported, checkout — with counts, category values and true/false flags only. No entity name, account name, balance or file name is ever sent to analytics; this is a hard rule of the codebase enforced by a review gate. The page-view beacon reports the page path, which on an engagement screen contains the engagement's random identifier (an opaque code), never a client value. Vercel Analytics does not use cookies and does not track you across sites.

Server logs record reason codes and counts — "parse failed, 3 rows" — never a row from a trial balance, an account name, a balance or a difference. Code that could log anything else is blocked by an automated lint rule with a single, reviewed exception that accepts codes and counts only. Hayatiq keeps no log store of its own: server logs exist only within the log retention of the hosting and database providers' plans (a matter of days, stated in their documentation, section 11) and are not exported or archived by Hayatiq; the providers keep their own connection logs (IP address, time, URL, user agent) for security and operations, for the periods in their notices.

Error reports from your browser. The engine runs in your browser, so when a page, the engine or a part of the draft fails there, the page sends a short report to Hayatiq's own server; no third-party error service is used. A report holds fixed codes only: where the error came from, its kind (such as "TypeError"), the page's route pattern without its identifiers, the build it came from, the browser family and a count. The error's message and stack, the page's address with its identifiers, and anything you typed or uploaded are never sent. Each report becomes one line in the hosting provider's function log and is kept nowhere else.

Uptime check. Scheduled checks regularly ask the Service's health address whether it is running: a job that runs on a machine Hayatiq operates (at the effective date, once an hour) and, where Hayatiq has set one up, an external monitoring service. The answer carries a build number and a status, and no personal data.

Reports of the browser's script policy. Each page tells your browser which scripts it may run and asks it to report, without blocking anything, a script that a stricter policy would refuse. Your browser sends such a report to Hayatiq's own server, which reduces it to fixed codes (the policy rule, the kind of script, whether it is one of the site framework's own, and whether the rule was enforced), writes one line in the hosting provider's function log for each different combination of those codes, with a count of how often it occurred, and drops the rest of the report, including the page's address, the script's address and any sample of its text.

Request limits. To slow down abuse and protect the Service, the server counts requests in its own memory: by IP address for sign-in links, the guest download, the checklist download, error reports and script-policy reports; by e-mail address for sign-in links, held only as a SHA-256 digest of the address; and by opaque identifier for document reading, where the Service offers it, and for the assistant, whose requests are counted per user and per organisation, with a daily number for each. These counts are held only in the memory of the server instance that answered the request, never written to the database, a log or analytics, and are lost when that instance stops; the store is bounded, so the oldest entries are dropped first. Where Hayatiq has set a rule for sign-in links, the guest download or the assistant, the hosting provider's firewall can also count requests by IP address at its edge; those counts are the provider's (section 11).

Legal basis for sections 6 to 9 (where the GDPR applies). For billing (section 6) and support (section 7), performance of the contract with you and, for billing records, Hayatiq's legal obligations; for the analytics, logs, error reports, script-policy reports, uptime check and request limits in this section, Hayatiq's legitimate interest in running, measuring and securing the Service, which they serve with codes, counts and counters only; for the cookies and browser storage in section 9, which the Service needs to work, performance of the contract with you.

9. Cookies and browser storage

The Service uses only what it needs to work:

• a locale preference cookie set by the site framework (NEXT_LOCALE) so that pages open in your language;
• session cookies that keep you signed in to Hayatiq ID, set by the authentication provider and readable only by the Service;
• IndexedDB, the browser's local database, which holds each engagement draft on your device until you save it or delete it (section 1);
• a local storage entry, hayatiq-theme, which remembers on this device whether you chose the light or the dark theme, and holds nothing else;
• Paddle's checkout cookies: when you open a checkout on the billing page, the page loads Paddle's checkout script (Paddle.js), and Paddle sets the cookies its checkout needs to take the payment and prevent fraud, as an independent controller under its own notice (section 11). The script is not loaded until you open a checkout.

Hayatiq sets no advertising cookies and no third-party tracking cookies, and does no fingerprinting. We show no cookie banner because Hayatiq sets no cookie that needs consent; Paddle's checkout cookies are governed by Paddle's notice.

10. Retention, deletion and no-storage mode

No-storage mode. When saving, you can flag an engagement for no-storage mode. A purge job runs every night at 03:00 UTC and, for every flagged engagement, deletes its saved snapshots and their working rows: the mapping, the questionnaire answers, the cash-flow worksheet and the review notes. It keeps the engagement record, the export records and the rounded statement figures recorded by an export made before the engagement was flagged, until you delete the engagement. Next year's roll-forward reads a saved snapshot, which the purge deletes; a mapping template you downloaded is a file on your own device, which the purge never reaches. An export made from a draft saved with the flag records no statement figures; an export from a browser that still holds an earlier copy of the draft, saved without the flag, can record them. A snapshot saved with the flag is kept only until the next nightly run.

11. Providers

Hayatiq relies on the following providers. Six of these are sub-processors; Paddle is an independent controller for the purchase. Each sub-processor receives only what the "What it receives" column describes, under a contract that binds it to process that data only for Hayatiq, except what a provider's terms let it keep for its own safety review or because the law requires it (for Anthropic, section 14). Hayatiq gives customers at least thirty (30) days' notice under the Data Processing Agreement before adding or replacing a sub-processor.

Each provider publishes its own privacy and security documentation: Supabase, Vercel, Anthropic, Resend, GitHub, Paddle, Cloudflare.

12. International transfers

United Arab Emirates (UAE PDPL). Hayatiq is established in the United Arab Emirates and stores data in Germany, so data of people in the UAE that you enter into the Service — your own account details and any individual named in an engagement — is processed outside the UAE, in Frankfurt and by the providers in section 11. For data you enter into an engagement, the transfer rests on the contractual safeguard in section 12.4 of the Data Processing Agreement: you instruct Hayatiq to store the data in Germany and to use the providers listed there, and Hayatiq binds each of them by written terms to protect it to at least the standard that agreement requires. For your account details, which Hayatiq controls, it rests on the written terms that bind each provider in section 11 to process the data only for Hayatiq. In both cases the data is encrypted in transit and at rest.

Saudi Arabia (PDPL). Hayatiq is established in the United Arab Emirates, outside the Kingdom, and stores data in Germany, so data of people in Saudi Arabia that you enter into the Service — your own account details and any individual named in an engagement — is transferred outside the Kingdom. Hayatiq's position is that such transfers take place under the Standard Contractual Clauses issued by the Saudi Data & Artificial Intelligence Authority (SDAIA) in accordance with the Regulation on Personal Data Transfer Outside the Kingdom, supported by a transfer risk assessment, and that the Data Processing Agreement incorporates those clauses for customers who are controllers in the Kingdom. The clauses are the Standard Contractual Clauses for Personal Data Transfer published by the Saudi Data & Artificial Intelligence Authority on 1 September 2024, in their controller-to-processor form (or their processor-to-processor form where you are yourself a processor for your clients); every such customer enters into them with Hayatiq by accepting the Terms, which incorporate the Data Processing Agreement and its Annex 5. Hayatiq maintains a transfer risk assessment for the transfer, written for you to adopt as controller, and provides it on request at privacy@hayatiq.ai. Hayatiq has not registered on the Kingdom's national register of controllers or appointed a representative in the Kingdom; if either becomes required, this section will name it. The PDPL basis for the processing described in sections 4, 6 and 7 is that it implements the agreement you are party to, and for the usage measurement and logs in section 8 it is Hayatiq's legitimate interest in running and securing the Service, which involves no sensitive data; for the data you enter into an engagement, the basis is yours as controller — normally the law that requires the financial statements, or your agreement with your client. Financial statements are largely commercial rather than personal data, which narrows what falls under the PDPL; names of directors, shareholders, related parties and employees are the usual personal data in an engagement, and you should enter only what the financial statements need. Where the Service offers document reading, a commercial registration extract can carry owners' identification numbers: the assistant is instructed not to return them, Hayatiq discards any it detects, and Hayatiq does not keep them; the provider's handling is as section 14 states.

European Union and United Kingdom (GDPR). Data is stored in Frankfurt. For the transfer from you to Hayatiq, which is established in the United Arab Emirates, the Data Processing Agreement incorporates the European Commission's Standard Contractual Clauses and the UK International Data Transfer Addendum as its Annex 6 (section 12.3). Where a provider outside the EEA or the UK receives personal data — Anthropic, Resend, Cloudflare and GitHub in the United States; Vercel (edge network global, functions in Frankfurt); Paddle in the United Kingdom — the transfer relies on the mechanism that provider offers: each sub-processor's data processing addendum incorporates the European Commission's Standard Contractual Clauses and the UK International Data Transfer Addendum, Paddle as an independent controller relies on its own, and Annex 3 of the Data Processing Agreement records the mechanism for each sub-processor. Hayatiq has no representative in the EU or the UK (see 'Who we are').

Other countries. For people elsewhere, including the United States, Hayatiq handles data as this policy describes, with the same providers and safeguards. Hayatiq does not sell personal data, and does not share it for targeted advertising.

13. Security

Implemented at the effective date:

• Row-level security on every database table, so that a signed-in user can read and write only rows belonging to organisations they are a member of; an automated suite proves cross-organisation reads and writes are denied on every change, and the public database role holds no table grants.
• Client data stays on your device until you act (section 2); the assistant's mapping request carries a fixed set of fields and scrubs figure-shaped values before sending; where the Service offers document reading, a document is sent only when you choose Read for that file (these controls ship with the reading).
• Encryption in transit (HTTPS) between your browser, Hayatiq and every sub-processor; encryption at rest on the database disks as provided by Supabase's infrastructure.
• Column-level encryption in the application. Each saved engagement snapshot, each questionnaire answer, and the rounded statement figures recorded at export are encrypted on Hayatiq's server with AES-256-GCM before they are written to the database, with a key held in the hosting provider's encrypted configuration and never in the database. Each value is bound to its engagement and its row, so a value copied elsewhere does not decrypt, and the database refuses to store a value in those columns that is not in this encrypted form, so every copy or backup of the database taken after that rule took effect holds them only as ciphertext (rounded statement figures recorded before the rule was added for them were encrypted when it was added, and an older backup is removed within the retention in section 10). The engagement's name, country, period, currency and framework choice, and the metadata records beside it (export records, checklist results and warning codes), are not encrypted this way; row-level security and the disk encryption above protect them.
• Request limits on sign-in links, the guest download, the public forms, error reports, script-policy reports, the assistant and, where the Service offers it, document reading, kept in the server's memory (section 8).
• No client financial data in logs or analytics (section 8), and Paddle event records limited to the fields acted on.
• Secrets never in source code; credentials held only in the encrypted configuration of the provider that uses them (the hosting provider for the application, the backup runner for the backup), the backup decryption key by Hayatiq's owner outside every provider, and a tooling guard that blocks any automated change to secret files.
• Sign-in by one-time e-mailed link with no password store, and the save boundary described in section 3 — live at the effective date.
• Nightly encrypted backups of the database (section 10), with one restore verified against the backup's own manifest; the verification is repeated after the first real client data is saved. Recovery rests on these nightly encrypted dumps: point-in-time recovery is not in place, so a restore returns the database to the previous night's copy.
• No development database holds data. The pre-launch development project was retired on 2026-09-27: preview and development deployments carry no database, and nothing is stored there.

Not in place, and not requested:

• Zero data retention with the assistant's model provider: not in place; Hayatiq has not requested a zero-data-retention arrangement from Anthropic. The assistant runs on the terms set out in section 14, and Anthropic deletes what is sent within thirty (30) days of receiving or generating it, or keeps it longer where the data-retention policy it publishes for its commercial API customers says so (section 14).

No system is perfectly secure. If Hayatiq becomes aware of a breach affecting your personal data, we will tell you without undue delay and, where the law requires, notify the competent authority within the period it sets; the Data Processing Agreement states the notice terms for customers.

14. The assistant and zero data retention

The assistant uses a language model hosted by Anthropic (section 11). By design the assistant decides nothing: its note text is prose only, every figure in a note's statements and tables is placed by Hayatiq's engine from the trial balances and the inputs the draft uses (those you entered or confirmed, and the assumptions the Service applied), a suggestion or proposal reaches the draft as an assumption you can change or keep, or once you confirm it, and a draft that rests on an assumption is exported, signed in, only once it is released; the guest download is not released and carries every assumption mark and the schedule of assumptions instead (Terms of Service, section 2); a fact or figure read from a document, where the Service offers it, is used only when you confirm it. What is sent — account names and codes, your answers (including any text you typed), and each note as it prints with its figures, so the assistant can write about them: for a mapping suggestion, account codes and names (with figure-shaped text stripped), the category label you named on the upload step and the statement group that label points to (current or non-current assets or liabilities, equity, income or expenses); for a proposed questionnaire answer, worksheet answer or note input, account names under opaque references with the statement line each is mapped to — never a balance or the entity-name field; for narrative, each note as it prints with its figures and your answers to the questions that note draws on, including any text you typed for them — never the trial balance or the entity-name field, though a name that appears in those answers or in the note as it prints can be sent as it appears. Every mapping, proposal and narrative request also carries the company's country, as you chose it, and the framework the draft uses, which frame the assistant's instructions; neither is personal data. The assistant never supplies a figure to a note's narrative: every number in a paragraph must match the engine's figures, the note's fixed allow-list or text you entered or confirmed for that note, or the paragraph is refused.

Reading last year's documents. Where the Service offers it (it shows the option to read a document), when you choose Read on last year's issued financial statements or the commercial registration extract, the file is sent whole to the model provider for that reading. The assistant proposes facts (the registration details, the owners (each name as printed, and whether each is Saudi or GCC, never the nationality), the counterparties, and the terms and statuses the statements record) and last year's figures as printed. It does not copy the statements' sentences; it is not asked for, and is instructed not to return, any person's identification number, date of birth, nationality or signature, and Hayatiq's server discards any such number it detects. Each proposal names the page it was read from and is used only when you confirm it. A figure that last year's trial balance can check is checked before it is offered, and one it cannot check is marked as resting on your confirmation. Hayatiq does not store the file.

The assistant runs on the terms set out in this section. Under its commercial terms, Anthropic does not use what is sent to train its models unless the customer permits it; Hayatiq has not given that permission and sends Anthropic no feedback.

How long Anthropic keeps what is sent. Every request the assistant makes — a mapping suggestion, a proposed questionnaire answer, worksheet answer or note input, note narrative and, where the Service offers it, a document reading — is covered by the data-retention policy Anthropic publishes for its commercial API customers: Anthropic deletes what is sent, and what it returns, within thirty (30) days of receiving or generating it, and keeps it longer only where that policy says so, such as content its automated safety systems flag under its usage policy (which it may keep for up to two years; the safety scores it assigns may be kept longer), or where the law or a legal hold requires it. The retention setting of Hayatiq's organisation with Anthropic shows the same thirty (30) days, and the organisation has not joined Anthropic's development partner programme or allowed user feedback (read 2026-09-27). Anthropic's documentation describes shorter handling for some uses; Hayatiq states the thirty-day period because it is the longest that applies in the ordinary case, and does not rely on anything shorter. Zero data retention is not in place, and Hayatiq has not requested it. The thirty-day period above is the position for every request. The product calls only models under Anthropic's standard retention, never a model Anthropic names as carrying a retention requirement of its own, as Hayatiq last read its list, and a test in the code holds it to that.

Hayatiq has authorised the assistant on production; it runs for an engagement only once you allow it there, a choice kept for that engagement in your browser. On a deployment where Hayatiq has not authorised it, its controls are shown disabled, and Hayatiq's own tests run the model only against synthetic fixture data.

15. Your rights and how to exercise them

Depending on where you are, you have rights to access the personal data we hold about you, to correct it, to have it deleted, to restrict or object to its processing, to receive a copy in a portable form, to withdraw consent where consent is the basis, and to complain to a supervisory authority — in the United Arab Emirates, the UAE Data Office, where it accepts complaints; in Saudi Arabia, SDAIA; in the EU or UK, your local data protection authority; elsewhere, the authority where you live, if there is one.

To exercise a right, e-mail privacy@hayatiq.ai from the address on your account. Sign-in is by that address, so it is how we verify you: we reply to it, restating what we will do, and act once you confirm from it. A request sent from any other address is confirmed the same way, through the account's address, and until then we say nothing about whether that address has an account. We answer within thirty (30) days of the confirmed request, and tell you if we need longer for a complex one. Most of what we hold you can also see or change yourself in the Service: your organisation, its members, and each engagement, which you can delete at any time.

If you are unhappy with how we handled your data or a request, write to us first at privacy@hayatiq.ai: we look into every complaint and answer within thirty (30) days. You can complain to a supervisory authority at any time, whether or not you have written to us.

Closing your account, and erasure. The Service has no button to close an account: write to privacy@hayatiq.ai as above. A closed account is deleted thirty (30) days after the confirmed request, so that your organisation can first ask for its saved engagements as exported files (Terms of Service, section 17), or sooner if you ask for that in writing; a request to erase your data is met within the thirty days above. When your account is deleted, an organisation in which you are the only member is deleted with everything in it except its billing records; in an organisation with other members, your membership ends and the work you created there stays with that organisation, without you recorded as its author. If you are the only owner of an organisation that has other members, we first ask you to make another member an owner or to close the organisation, and an organisation whose paid subscription is still running is not deleted until that subscription is cancelled: an owner cancels it from the billing portal, where it ends at the close of the period already paid for (Terms of Service, section 6), or asks us and we cancel it in Paddle. An owner can ask us in the same way to close an organisation: after the same thirty days it is deleted with everything in it except its billing records, and its members keep their own accounts. Invitations to your address and any checklist-download entry for it are deleted with your account. What is kept: the records Paddle holds as merchant of record (section 6), which Hayatiq cannot delete; the billing records in section 10, for their retention; the backups, until the last copy holding your data is removed (section 10); what the assistant sent to Anthropic, until Anthropic deletes it (section 14); our correspondence with you (section 7); and, in an organisation you leave, your address in the release records of engagements saved after you released them (section 10), which that organisation controls.

What the assistant sent to Anthropic is deleted by Anthropic on its own schedule (section 14); Hayatiq cannot have it deleted sooner.

If your data is in an engagement that one of our customers saved — you are a director or related party named in their financial statements — that customer is the controller. Please write to them; if you write to us instead, we will pass your request on and help them answer it, as the Data Processing Agreement requires.

16. Children

The Service is for business use by adults. We do not knowingly collect data from anyone under eighteen (18); if you believe we have, contact us and we will delete it.

17. Changes to this policy

We will update this policy when the product or the law changes. Material changes are notified by e-mail or in the Service; the current version, with its date, is always at financials.hayatiq.ai/en/privacy.

18. Contact

HAYATIQ AI SOLUTIONS F.Z.E, B.C. 1306744, Ajman Free Zone C1 Building, Ajman Free Zone, Ajman, United Arab Emirates · privacy@hayatiq.ai (data and privacy) · hello@hayatiq.ai (general). Effective date: 27 October 2026.