KIRIL DACHEV / YOUR INFORMATION

Privacy,
in plain words.

How personal information is handled on dachev.ai and in the invitation-only Personal Hub.

PRIVACY POLICY

Last updated

Download the PDF

A copy to read or keep. Same policy, same date.

You can browse the public website without a Hub account. Private workspace information is used to provide the features you choose. Connecting an external account or enabling AI processing is optional.

01

Who is responsible

Kiril Dachev operates dachev.ai, including www.dachev.ai and Personal Hub (the Hub). In this policy, "I", "me" and "my" refer to Kiril Dachev. Where applicable data protection law applies, I am the controller of personal information processed to operate this website and the Hub.

For privacy questions, requests for access or deletion, or concerns about information someone has added about you, use the website's contact section or email kiril@dachev.ai. This policy covers visitors, invited account holders and people whose information appears in records provided by an account holder. It does not replace the privacy notices of independent services you connect or visit.

02

Information the site handles

  • Website visits: hosting and security services may receive your IP address, browser and device information, requested pages, timestamps, referrer information and diagnostic logs when delivering and protecting the site.
  • Accounts and contact: email address, profile details you supply, invitation and membership records, authentication and multi-factor authentication information, session information, and correspondence you send me.
  • Private workspace records: information you enter or upload, including financial accounts and transactions, statements and supporting files, budgets, notes, tasks, contacts, relationship records, birthdays, habits, plans, goals and memories. Records can include information about other people, such as payees, contacts and event attendees. Only provide information you are entitled to use.
  • Connected services: account identity, the permissions you grant, connection credentials and selected data from an external service. Depending on the feature you enable, this may include Gmail messages, Google or Apple/iCloud calendars, Google Docs document content and metadata, WHOOP health and fitness metrics, or Strava activity information. Availability depends on provider setup and your authorisation.
  • Processing results: imported or extracted fields, source references, classifications, summaries, your corrections, analysis revisions, sync status and security or processing records. Sources are you, the services you authorise, other account holders who provide relevant records, and technical information generated when you use the site.

03

Why information is used

Information is used to deliver the website; manage invitations, sign-in and account security; store and organise your workspace; carry out imports, syncs and actions you request; provide optional analysis; respond to messages and privacy requests; diagnose faults; and prevent misuse.

Where UK or EU data protection law applies, the lawful basis is legitimate interests in running and protecting the site, providing the private workspace you request, and responding to enquiries, balanced against your rights. Where processing is necessary to perform an agreement with you or take requested steps before an agreement, the basis is performance of that agreement. Processing required by law relies on legal obligation. Where consent is required for an optional use, that use relies on your consent, which you can withdraw.

Health information is sensitive. Optional processing of your health information requires your explicit consent where applicable law requires it; connecting an account does not authorise unrelated uses. You can stop future collection by disconnecting and can request deletion of information already held.

You do not have to create an account or connect a service to visit the public site. Account and security information is needed to provide account access; other workspace content and integrations are optional. Declining a required permission means the related feature may not work. Personal information is not sold or used for targeted advertising.

04

Google accounts: Gmail, Calendar and Google Docs

Each Google account is connected separately with your permission. Google supplies account identity information and access tokens; the Hub does not receive your Google password. The Google consent screen shows the permissions requested.

Gmail features may read message headers, senders and recipients, subjects, message text, labels, dates and thread identifiers to display mail, prepare selected intake and create review cards or reminders. Mail and calendar write permissions support actions you request, such as drafting or sending mail, changing message labels or read status, and creating, editing or deleting calendar events. Calendar data can include calendar names, event titles and descriptions, locations, times, attendees and recurrence details. Messages you send are shared with the recipients you select. Calendar changes are made in the account and calendar you choose and follow that calendar's existing access settings.

When Google Docs features are available and you authorise them, the Hub may access document titles, identifiers, metadata and contents to display documents and carry out document creation or editing you request. The permissions shown by Google determine which documents the Hub can access. Changes are made in the Google account and document you choose and follow the document's existing sharing settings. Disconnecting the Hub does not delete documents held in your Google account.

Authorising access and selecting background intake are separate choices. Intake follows the account, date and filter settings you approve. Reading a message or event on demand can access it outside the background intake selection when you request it.

Use and transfer of information received from Google APIs will adhere to the Google API Services User Data Policy, including its Limited Use requirements. Google user data is used for the requested, visible Hub features, not advertising, data brokerage or credit assessment. Google user data and data derived from it are not used to create, train or improve AI models. People may read Google user data only with your prior affirmative agreement for the specific data, where necessary for security or legal compliance, or for internal operations using aggregated and anonymised data as permitted by Google's policy.

You can disconnect an account in Connections and revoke the Hub's access in your Google Account connections. Removing permission stops future authorised access; deleting copies previously imported into the Hub is addressed below.

05

Optional AI and document processing

Connecting a service does not by itself enable external AI analysis. Optional mail analysis requires an enabled configuration and accepted provider settings. When used, selected message text or excerpts and relevant metadata may be sent to the configured provider, such as OpenAI, Anthropic, or a disclosed upstream provider through OpenRouter, to produce classifications and evidence-based review cards. Screening aims to reduce sensitive content sent to a model but cannot identify every sensitive passage.

Finance processing through Codex is a separate, user-initiated workflow. A scoped grant can allow OpenAI Codex to receive the selected original files, file metadata and available analysis history. These files may contain financial and third-party personal information. Review your selection and the service's settings before starting; you can revoke the grant. Uploading a file alone does not start this processing.

When you enable an optional AI feature and affirmatively accept its disclosed provider settings, you authorise processing for the requested feature and its disclosed security and technical troubleshooting purposes. We ask you to allow temporary retention of the data sent to the AI provider, and its outputs, for up to 60 days for those purposes. This is a precautionary allowance, not a standard storage period: current mail-analysis routes ordinarily limit provider retention to 30 days or less. Any shorter period or stricter control you accept continues to apply.

Where supported and enabled for the selected service and account, we use zero data retention. It is not available for every provider or feature. Provider-specific retention terms must be disclosed before you enable processing. Legal obligations and exceptional safety or abuse investigations may require a provider to retain some information beyond the ordinary 30- or 60-day period; the allowance is not a guarantee that every copy is deleted within 60 days.

Saved Hub records, outputs and corrections follow the retention and deletion rules below. Files and task content transferred to Codex follow that product's account settings and retention terms separately; revoking or expiring a Hub access grant does not delete copies already transferred. The temporary AI-provider allowance does not set a 60-day deletion deadline for these saved records.

This notice does not itself enable AI processing or expand a user's existing acceptance. Material changes to purposes or retention that require consent will be presented for a separate affirmative choice. AI use remains limited to the disclosed purposes and your choices, including the Google-data restrictions above. The Hub does not use private workspace content to train a general-purpose AI model. AI outputs are suggestions for your review; the Hub does not make solely automated decisions producing legal or similarly significant effects about you.

06

Service providers and international processing

Vercel provides website hosting and delivery. Supabase provides authentication, database, file storage and backend services. Google Fonts supplies the website's fonts and receives the network information needed to deliver them. Finance exchange-rate requests to Frankfurter contain currency codes and date ranges, rather than your account balances or transaction contents, alongside normal network information.

Connected services receive requests and data necessary for the features you authorise. An AI service receives selected information only when its processing workflow is used as described above. Information may also be disclosed where necessary to meet a legal obligation, protect rights or investigate security incidents. Private Hub records are not published merely because you create a contact, group or account.

Providers may process information outside your country, including in the United States. The Hub's Supabase project is hosted in the United States. Where applicable law requires protection for a restricted international transfer, appropriate arrangements are required, such as an applicable adequacy decision or contractual safeguards, including relevant UK provisions. Contact me for information about the arrangements applicable to your data; this site does not promise UK-only or EU-only storage.

07

Storage and access

The Hub uses authenticated access, multi-factor authentication, per-account database permissions, private file storage and encrypted connection credentials to protect private information. Authorised hosting and backend providers, and the operator using privileged systems, may technically be able to access server-readable information when needed to operate, support or protect the service. The Google-specific limits above apply to Google user data. This is not an end-to-end encrypted vault.

Keep your devices, password and authenticator secure, and sign out on shared devices. No online service can guarantee absolute security. Downloaded files and copies you send to another service are subject to the protection and deletion controls of that destination.

08

Retention, disconnection and deletion

Account details and workspace records are kept while needed to provide your account and the records you choose to retain. Enquiries, security records and diagnostic information are kept according to the purpose, the time needed to resolve issues or protect the service, and applicable legal requirements. There is no single retention period for all information.

  • Mail: unsaved imported records normally expire 90 days after import. Cards you explicitly save remain until deleted. Deleting a mail card can retain a minimal message identifier to prevent it being imported again. Disconnecting a Gmail connection removes its locally stored intake settings and mail records, including saved cards.
  • Health and activity: imported WHOOP metrics use a rolling 93-day window. Disconnect removes the connection's imported records. Strava activities are read live and are not kept as a Hub activity archive.
  • Calendars and connections: disconnect stops future syncing and removes connection credentials and saved calendar selections. External calendar events are read when requested; Hub-created plans remain until separately deleted. It does not delete the source account or its records at the provider. Provider token revocation is attempted; if it cannot be confirmed, revoke access at the provider as well.
  • Finance and other saved content: uploaded originals, accepted statements and analysis history can be retained independently of a service connection. Some linked records and statement revisions cannot be deleted through the ordinary interface. Contact me to request deletion of these records or your account.
  • Backups and other copies: removal from active systems may not immediately remove retained backups, security records, legal records, or copies already downloaded or sent to a provider. These are subject to their applicable retention and deletion processes. A deletion request will be assessed across the relevant records; no immediate deletion from every backup is promised.

09

Cookies, browser storage and external links

The site does not currently include advertising trackers, session replay or a third-party visitor analytics script. Necessary hosting or security technologies may process technical information to deliver and protect the site.

The Hub uses browser session storage for authentication and temporary sign-in state. The public site and Hub use local storage to remember your motion preference. Session information is ordinarily limited to the browser tab session, although browsers may restore sessions. Motion preferences remain until you change them or clear site storage. Blocking or clearing storage may sign you out or reset preferences.

Fonts are requested from Google when pages load. Links to other websites take you to services with their own privacy practices; their handling of data is governed by their own notices.

10

Your choices and rights

Depending on the law that applies and the circumstances, you may ask to access a copy of your personal information, correct it, erase it, restrict its use, receive portable data, or object to processing based on legitimate interests. Where processing relies on consent, you may withdraw consent at any time without affecting the lawfulness of processing before withdrawal.

To make a request, use the website's contact section or email kiril@dachev.ai, and identify the account or information concerned. You do not need a Hub account to make a request. I may need proportionate information to verify your identity and locate the records. Requests will be handled within the period required by applicable law, normally one month under UK or EU data protection law, subject to permitted extensions or exceptions.

You may raise a concern with me or complain to your local data protection authority. In the UK, you can contact the Information Commissioner's Office.

11

Children

The Hub is intended for adult invited users and is not directed at children. If you believe a child has provided information through an account without appropriate authorisation, contact me so I can investigate and address it.

12

Changes to this policy

This policy may change as the site develops. The date at the top identifies the current version; the downloadable PDF carries the same date. Material changes will be explained where appropriate, and new consent will be requested where required before information is used for a new purpose.