1. Who We Are
Esek (available at esek.io; "we", "our", "us") is an AI workspace for businesses. It answers phone calls and customer messages, books appointments, manages customer records, and runs sales and marketing campaigns. It connects businesses ("Business Customers") with their customers ("End Users") by phone, WhatsApp and, where enabled, Instagram and Facebook Messenger. We act as a Technology Provider under the Meta Business Platform and process data on behalf of our Business Customers in accordance with Meta's Platform Terms and Business Messaging guidelines. This policy explains how we collect, use, store, share and delete personal data.
2. Data We Collect
We collect data from two audiences: Business Customers who register to use our platform, and End Users who communicate with those businesses by phone, WhatsApp, Instagram, or Facebook Messenger.
From Business Customers
- Verification text messages: When you request an SMS security code, we send your phone number and that code to Twilio for delivery. Supported verification routes use Twilio Verify. We record the time of your request and, where available, the verification receipt with the pending challenge to operate the verification flow and document your request. Verification messages do not enroll you in marketing messages.
- Account data: name, email address, and password hash of registered agents.
- Onboarding data: when a business connects via Meta Embedded Signup, we receive the WhatsApp Business Account ID, phone number ID, business verification status, and a scoped access token. We store only what is necessary to operate the integration.
- Product usage data: events recording how our apps and website are used, such as which screen was opened, which action was taken, and which errors were shown. Screens are recorded as route patterns, so identifiers that appear inside a page address are not collected. Each event carries a randomly generated session identifier that puts one visit in order: in the apps it lasts until the app is closed, and on our website until the browser tab is closed. When a registered agent is signed in we also record the agent identifier once per session, so usage in that session can be associated with that account. We also record that operational events occurred, for example that an order was created or a payment settled, using identifiers, statuses and counts only, never the contents of a message. All of this is stored in our own database. We do not send it to an external analytics provider.
- Operational logs: log entries and error traces generated during normal operation.
- Mobile advertising measurement data: if you allow tracking through Apple's App Tracking Transparency prompt, the iOS app sends installation and activation events, a device identifier, and basic app and device metadata to Meta App Events so we can measure the performance of our Meta advertisements. We have disabled collection of Apple's advertising identifier (IDFA). If you deny or restrict tracking permission, Meta App Events is not initialized and this data is not sent.
- Signup security and provenance data: a one-way, keyed fingerprint of the IP address used when a workspace is first created, the sign-in method, and a coarse client platform. We use this only to prevent abuse, distinguish internal testing from external signups, and diagnose signup problems. We do not retain raw IP addresses or full user-agent strings for this purpose.
From End Users (via WhatsApp conversations)
- Conversation data: WhatsApp phone number, display name, and messages exchanged through the platform.
- Media files: images, videos, documents, and audio messages sent or received via WhatsApp.
- WhatsApp Business app data: If a Business Customer connects the number they use in the WhatsApp Business app, Meta can also share that app's chat history and contact list with us. From the contact list we keep only each contact's WhatsApp number and the name the business saved, and we use them only to recognize people the business already talks to itself, so the Chat Agent does not reply to them until the business hands the chat over. A contact the business removes in the app is deleted here too, and the rest is deleted with the business account.
Connected Instagram and Facebook Messenger accounts
When a Business Customer enables these channels and authorizes a connection, we receive the selected professional Instagram account or Facebook Page identifier, account name, authorizing account identifier, granted permissions, and access credentials. We encrypt the credentials and use them to operate that business's messaging connection.
For customers who message a connected account, we receive a Page-scoped Messenger customer identifier or Instagram-scoped customer identifier, available profile name, message content, supported attachments, and any reply, story, or referral context supplied by Meta. We keep each account's conversations separate. Authorized staff can link a channel identity to an existing customer record. We do not infer that two people are the same person from a matching profile name. Supported attachments may be stored privately so staff can view them after the provider's temporary download link expires.
When authorized staff open a conversation or ask Iska (Copilot) for a customer profile, we may also retrieve the available profile name, Instagram username and profile link, and profile photo URL from Meta. We store an available display name on the channel identity; the additional profile fields are fetched on demand rather than saved to that customer record.
Authorized staff can read and reply to these messages. When the business enables AI assistance and the applicable consent requirements are met, message text and relevant history are processed by the AI providers described below. The retention and deletion rules for these channels are described in section 8.
Connected Zoom accounts
Each appointment host can choose to connect their own Zoom account. We receive the host's Zoom account and user identifiers, account email, granted permissions, and authorization tokens. We encrypt the tokens and use the connection to create, reschedule and delete meetings for confirmed online appointments. Meeting requests contain the business and service name, scheduled time and duration, a booking reference, a generated passcode and security settings. They do not contain the customer's phone number, email or an attendee list.
We store the connection identity, encrypted tokens, meeting identifiers, participant join URLs and passcodes, and synchronization status. To recover an interrupted meeting creation without duplicating it, we read scheduled meeting listings and the matching meeting's details. Unrelated listing data is not saved as Esek meeting records. This integration does not access recordings, transcripts, AI Companion summaries, participant rosters or real-time media.
The host or an authorized administrator can disconnect Zoom in Esek. Disconnecting removes the stored connection and retained Zoom meeting access data. When Zoom notifies us of removal or deauthorization, we clear authorization tokens, account identity, granted permissions, meeting identifiers, join URLs and passcodes. Existing Esek appointment and business records retain the periods in section 8. Meetings already created may remain in the host's Zoom account. Privacy and deletion requests can be sent to privacy@esek.io.
From End Users (via AI phone calls)
When a Business Customer turns on the AI phone assistant, calls forwarded to the phone number we provide are answered by an AI voice assistant on that business's behalf. Business Customers configure the opening greeting. For each such call we collect:
- Caller identity: the caller's phone number, as delivered by the telephone network or by WhatsApp. We also use that number to create or match a contact record in the Business Customer's own customer list.
- Call transcript: a written record of what was said during the call, produced by speech-to-text.
- Call summary and messages: a short written summary of the call, and any message the caller asked us to leave for the business.
- Call metadata: when the call started and ended, how long it lasted, the number that was dialled, and whether it was answered.
AI calls and voicemail. During AI answering, call audio is processed live. We keep the written transcript, summary and message as the record of the call. Where call recording is enabled, we keep a private recording of the call audio, both the caller's side and the assistant's side, for up to 30 days, or longer where section 8 says so. Authorized workspace members can play it back. We also use it to investigate problems and check quality. It is then deleted automatically. It is not used to train AI models and is not shared for advertising. Depending on where the call comes from, callers may hear at the start of the call that it is recorded. Erasing a call, or deleting the workspace, deletes this recording immediately, except as section 8 says. When the business has insufficient minutes and balance, direct phone calls switch to voicemail: callers hear a greeting asking them to leave a message after the tone, and we privately store up to two minutes of their message. Authorized workspace members can play it. Erasing the caller’s record also deletes this recording.
Calls answered by the AI phone assistant end automatically after ten minutes.
Optional device permissions
- Microphone (mobile apps): our iOS and Android apps may request microphone access when you choose to record a voice message in chat. Audio is not captured in the background or outside the action you started. Audio from Copilot voice sessions is described in section 5.
3. Legal Basis for Processing
- Contractual necessity: processing Business Customer data is required to provide the platform service under our terms.
- Legitimate interest: processing End User conversation and phone call data is necessary for Business Customers to communicate with their customers and for us to maintain platform reliability. We rely on legitimate interest also to measure how our own product and advertising perform, so that we can improve them.
- Consent: where required by applicable law (e.g. optional microphone access), we rely on your explicit consent, which you may withdraw at any time.
4. How We Use Data
- To provide the messaging and booking service to connected businesses.
- To generate AI-assisted replies, signals, suggestions, images, and booking confirmations on behalf of the Business Customer. This includes the customer-facing AI surfaces: the Chat Agent (which drafts or sends replies to End Users on the business's behalf), the AI phone assistant (which answers forwarded phone calls on the business's behalf, can book appointments and take messages during the call, and writes up what was said afterwards), and the silent Signal Capture pass that extracts quotes, commitments, and handoff brief fields from human-replied chats for the merchant's later review; and the merchant-facing surfaces: Copilot (in-app assistant, including real-time voice when enabled), Ads suggestions (campaign name and creative copy), and image generation for business assets (logo, catalog photos, ad creative).
- To send transactional email notifications (e.g. new conversation alerts).
- To send call summaries and balance alerts on WhatsApp to your phone. To stop them, write so in that chat.
- To improve platform reliability and diagnose technical issues.
- With your App Tracking Transparency permission, to measure iOS app installations and activations attributable to our Meta advertisements.
We use WhatsApp, Instagram, and Facebook Messenger data strictly for the purposes described above. We do not use End User data for advertising, profiling, or any purpose unrelated to providing the messaging service.
5. Data Sharing
We do not sell personal data. We do not sell or share mobile numbers or SMS opt-in data with third parties for their marketing (see Text messages). Data may be shared with the following third-party processors:
- Meta (WhatsApp, Instagram, and Facebook Messenger): account authorization and messages are processed through the official API for the connected channel.
- Meta (App Events, iOS only): after you allow tracking through Apple's App Tracking Transparency prompt, installation and activation events, a device identifier, and basic app and device metadata are transmitted for advertising measurement. The SDK is configured not to collect IDFA and remains inactive if permission is denied or restricted.
- Telephony: calls to the phone numbers we provide for the AI phone assistant are routed by DIDWW, which receives the call's metadata, such as the caller's phone number and the number that was dialled. It does not handle the call audio. The audio is carried by Twilio, which therefore processes the caller's phone number and what is said while the call is in progress. Twilio also handles call recordings, SMS one-time sign-in codes, caller number lookups, and the documents needed to register a phone number with regulators.
- Text-to-speech providers: when Esek speaks a message aloud, such as a spoken briefing, a Copilot voice reply or a voicemail greeting, the text is sent to Google Gemini to turn it into speech. Google Gemini also marks Hebrew vowels in Copilot voice input.
-
Third-party AI providers: we use third-party AI APIs to power the customer-facing Chat Agent, the AI phone assistant, the silent Signal Capture pass on human-replied chats, the merchant-facing Copilot (including its voice mode), Ads suggestions (campaign name and creative copy), and image generation. The current providers are OpenAI, Anthropic and Google Gemini, with the provider depending on the feature. AI text features, such as chat replies, the Copilot assistant and summaries, run on either OpenAI or Anthropic. We have used Anthropic since 15 October 2026, and we move between the two without notifying you each time. Call audio, speech transcription and image generation stay on OpenAI. We may add or substitute other providers in the future. If we add a provider other than OpenAI, Anthropic or Google Gemini for Business Customer data, we will update this policy and re-prompt for consent in-app before routing that data to the new provider. Moving between OpenAI and Anthropic is not such a change, and we do not re-prompt for it.
The following categories of data are transmitted to the AI provider strictly to generate the requested response:- The text of messages being processed (End User WhatsApp, Instagram, or Facebook Messenger messages, or the Business Customer's prompt to Copilot, or a text prompt for image generation).
- Relevant conversation history needed for the model to produce a coherent reply.
- Business context the assistant needs to answer (e.g. the Business Customer's catalog, hours, pricing, the customer thread the merchant is viewing, lead status, and the merchant-authored notes document used to tune the Chat Agent, most of which is information already publicly available about the business).
- When voice mode is enabled in the merchant-facing Copilot, the microphone audio captured during an active voice session is streamed to the provider's realtime API for transcription and reply generation. Audio is not captured outside of an active voice session. We also keep a private copy of the audio from voice sessions in our own storage for up to 30 days, to investigate problems and check quality. It is then deleted automatically. It is not used to train AI models and is not shared for advertising.
- When the AI phone assistant answers a call, the audio of that call is streamed live to the provider's realtime API so the assistant can understand the caller and speak back, and the provider returns the transcript to us as text. We also send a short summary of that caller's most recent earlier contact with the business, where one exists, so the assistant does not make them repeat themselves. We also keep a private recording of the call in our own storage for up to 30 days, as described in section 2.
The AI provider processes this data as our sub-processor under their API Data Processing Addendum. Per the providers' API terms in effect as of the "Last updated" date above, data submitted via the API is not used to train the providers' models. Each provider keeps it under its own retention rules, which are separate from the retention periods in this policy. By default, Anthropic deletes API inputs and outputs within 30 days. If its automated trust and safety systems flag content, Anthropic may keep those inputs and outputs for up to 2 years and the related classification scores for up to 7 years. OpenAI keeps abuse-monitoring logs, which can include prompts and responses, for up to 30 days, or longer where the law requires it or where it is reasonably necessary to prevent harm. Google logs prompts and responses for a limited period to detect violations of its Prohibited Use Policy. Anthropic may also keep data longer where the law requires it. - ChatGPT connection: when a Business Customer explicitly connects an Esek business to ChatGPT, requested tool results are sent to OpenAI through that customer's ChatGPT account. Results can include business settings and activity, customer names and contact details, messages, notes, appointments, catalog, orders, payments and other records the connected staff member is allowed to access. The connection is limited to the business and permissions approved during linking. These copies are processed under ChatGPT's own terms, privacy policy and account data controls; the API-provider assurances above do not determine how the customer's ChatGPT account retains or uses them. Disconnecting Esek in ChatGPT, losing business membership or disabling the Esek account stops future access. Disconnecting does not erase copies already held in ChatGPT. For an Esek privacy or deletion request, contact privacy@esek.io.
- Infrastructure providers: Google Cloud Platform (US region, Iowa) hosts our servers, databases and file storage, and Firebase Cloud Messaging delivers push notifications.
- Email: Resend sends our transactional email.
- Other Google services: Google Cloud Translation receives message text to translate it, Google Web Risk receives links found in messages to check them for safety, and Google Maps Platform receives business addresses to look them up.
- Billing and sign-in: Paddle, the merchant of record for top-ups bought on our website, receives billing details and the customer's email address. Apple and Google handle Sign in with Apple, Google Sign-In, and purchases made in their app stores.
- Meta (Conversions API): to measure our own advertising, we send Meta signup events with a hashed email address or phone number and a hashed internal account identifier. For visitors who accepted marketing cookies, or who are outside the EU, the EEA and the United Kingdom and have not declined them (a Global Privacy Control signal counts as declining), the events also carry Meta's ad click identifier and browser identifier and the address of the signup page. When a lead reaches our own workspace from one of our click-to-WhatsApp ads, we also send Meta a lead event with the ad click identifier, our WhatsApp Business Account identifier and the contact's hashed phone number.
- Integrations a Business Customer connects: when a Business Customer connects its own account with Morning (Green Invoice), iCount, Tranzila, Cardcom, Grow (Meshulam), Easybizy, Google Drive, Google Calendar or Microsoft Outlook Calendar, we pass data to that provider at the Business Customer's direction. Before outbound sales calls, phone numbers are checked against Israel's national do-not-call registry.
The full list of sub-processors, with what each one receives and where it processes data, is on our Sub-processors page.
Microsoft Outlook Calendar. Each appointment holder can choose to connect their own Microsoft account. We receive the account email and identifier and read event times and availability status to prevent overlapping bookings for that person. Event titles, descriptions and attendees are not requested or stored. With the holder’s permission, we also create a separate Esek calendar and write the business’s booking details there. Without that permission, the connection is read-only and checks availability only. We store authorization tokens for ongoing synchronization. A holder or business administrator can disconnect the calendar in booking settings; this deletes our stored connection and tokens.
6. Google Account Data
Google connections are optional. Esek requests permission through Google's consent screen when you choose a feature that needs it. Google Sign-In does not itself grant access to your files, calendars or meetings. You can use Esek without connecting Google; features that import or synchronize Google data require the relevant permission.
- Google Sign-In. Using identity permissions (
openid,emailandprofile, as applicable), we receive your Google account identifier, name, email address and available profile photo. We store these with your Esek account to authenticate you and display your profile. - Google Drive and Docs. The
drive.filepermission lets us read files you select in Google's file picker and create business documents when you ask. It does not grant general access to your Drive. We receive the selected file's identifier, name, type, revision information and content. Catalog spreadsheet imports use a temporary access token to read the file and retain imported catalog records and the selected file identifier; the import token is discarded. Linked business documents and batches of selected files use an encrypted refresh token for ongoing synchronization. We store imported content, extracted text and derived knowledge records so authorized staff and the business's assistants can use them. A file you link can be read again until you disable or remove its source. - Google Calendar. Each appointment holder can connect their own account. We receive the account email and identifier and use
calendar.freebusyto read busy/free intervals and prevent overlapping bookings. We do not read titles, descriptions or attendees from existing calendars. With the holder's separate permission (calendar.app.created), we create a separate Esek calendar and create, update or remove the business's booking events there. Event data can include appointment times, service and customer details, and a Google Meet link when requested. We store encrypted authorization tokens, connection settings, calendar and event identifiers, and synchronization status. - Google Meet session documentation. When an assigned meeting host explicitly connects this feature,
meetings.space.readonlyanddrive.meet.readonlyallow us to retrieve the linked meeting's conference identifiers and start/end times and an available Google-generated summary document. Imports require recorded participant consent and authorized staff action. We retain imported session documentation in the customer's confidential Esek record. This connection does not itself start recording or transcribing a meeting.
Storage and sharing. Google account data is processed by our backend and stored in our access-controlled Google Cloud infrastructure described in sections 5 and 7. Data travels over TLS; stored integration tokens are encrypted. Authorized business staff can access imported records according to their workspace permissions. When you enable AI features and give the applicable in-app consent, selected document content and derived knowledge can be sent to the AI processors named in section 5 to classify business information, generate responses or produce requested drafts. Google-generated session summaries are imported without an Esek AI call; generating a draft from a staff-supplied transcript is a separate action requiring AI consent. We share Google data only as needed for these user-facing features with consent, for security, or as required by law. We do not sell Google user data, use it for advertising, or use it to train generalized AI models.
Retention and deletion. Sign-in identity is retained while your Esek account exists. Integration credentials remain while the connection exists; disconnecting a calendar deletes its stored connection and tokens. Removing a linked document source deletes that source, its stored credentials and its imported knowledge content. Disabling synchronization stops future reads but keeps the source. Imported catalog records, bookings and session documentation remain workspace records until separately deleted or the workspace is deleted, under section 8. Disconnecting or revoking Google access does not erase those records or delete the original Google files and calendars. Workspace deletion removes associated data from the live service and from recovery backups within 30 days. To request access, correction or deletion, contact privacy@esek.io or use our data deletion instructions.
Your controls. You can remove linked sources in business settings, disconnect a calendar in booking settings, or revoke Esek's Google permissions at your Google Account permissions page. Revocation stops future authorized access; reconnecting requires consent again.
Esek's use of information received from Google APIs follows the Google API Services User Data Policy, including its Limited Use requirements. This also applies to data derived from that information. Our personnel may read Google user data only with your affirmative agreement to view specific data, when necessary for security or legal compliance, or when aggregated for internal operations as permitted by that policy.
7. International Data Transfers
Your data may be transferred to and processed in countries outside your country of residence, including the United States, where our servers are hosted and where Meta and most of our sub-processors operate. Each sub-processor and the place where it processes data are listed on our Sub-processors page. For transfers from the EU, the EEA and the UK, we rely on the EU-US Data Privacy Framework (including its UK Extension) where the recipient is certified under it, and on Standard Contractual Clauses (with the UK Addendum for UK data), together with data processing agreements with each processor.
8. Data Retention
- ChatGPT connection records: we retain connection identifiers, consent and permission metadata, and hashed OAuth credentials for access control and security. These records have no fixed deletion schedule. Access tokens expire after one hour and approved grants after 30 days; revocation stops access sooner. Business records returned to ChatGPT retain the Esek retention periods below, while copies in ChatGPT follow that account's data controls.
- WhatsApp conversation messages and media: retained for up to 12 months from the date of the last message in the conversation, or until the Business Customer deletes the conversation, whichever is sooner.
- Instagram and Messenger records: messages, channel identifiers, stored display names, and message-event records remain as workspace records until erased or the workspace is deleted. We do not automatically delete these records after 12 months. Moving a conversation to trash allows recovery for seven days; permanent deletion is performed by a weekly cleanup after that period, or sooner if the workspace owner permanently deletes it. Social attachments in our private storage expire 12 months after the individual message was sent, or sooner when the associated conversation or social data is permanently erased. This attachment expiry does not erase the message text or customer record.
- Business Customer account data: retained for as long as the account is active.
- IP-derived signup fingerprint: retained for up to 90 days, then deleted. The remaining coarse signup context does not contain the IP address.
- Product usage data: retained for as long as it is useful for measuring and improving the service. We do not delete it on a fixed schedule. It contains no conversation content and no message media.
- AI phone call records: the transcript, summary, caller number, and call metadata are kept as business records of the workspace they belong to. We do not delete them on a fixed schedule. They are retained for as long as that workspace exists. When the workspace is deleted they are removed from the live service and held in a workspace backup for 30 days, in case the deletion is reversed, and then permanently deleted.
- Call audio and Copilot voice audio: recordings of AI phone calls and of Copilot voice sessions are kept in private storage for up to 30 days and then deleted automatically. Erasing a call deletes its recording immediately. Deleting the workspace deletes all of them immediately. They are not copied into the workspace backup. Where a Business Customer has told us that its calls fall under the call-recording duty in the Israeli Consumer Protection Law, recordings of its calls are kept for up to 25 months, as that law requires, and erasing a call does not delete its recording during that time.
- Upon workspace deletion: associated workspace data, including conversations, media, and connection credentials, is removed from the live service and permanently removed from the workspace recovery backup within 30 days. Deleting a staff account removes that person’s profile and access. A workspace solely owned by that account is also deleted; records in other workspaces remain.
Whose record a call is. A record of a call to a business is that business's own customer record, in the same way a written note of the call would be. It is not part of the personal account data of the staff member who works there. Two consequences follow, and we would rather state them plainly than leave them to be discovered:
- Deleting an account removes that person's profile, credentials, and access. It also deletes workspaces for which that person is the sole owner, including their business call records. Records in workspaces with another owner remain.
- Removing a caller from a Business Customer's contact list does not by itself erase the call records already logged against that caller.
If you called a business that uses Esek and you want the record of your call erased, ask the business, or write to us at privacy@esek.io. Erasing a call clears the caller's phone number, the transcript, the summary, any message taken, and any recording of the call we still hold, other than one section 8 says must be kept, and marks the call as erased on request. What remains is how long the call lasted and what it cost the business, which says nothing about who called or what was said. We keep that much because it is the evidence for a charge already made.
Disconnecting a social account and deleting its data. Disconnecting Instagram or Messenger stops new messaging through that connection and clears its stored credentials. It does not by itself erase existing conversations or records the business already holds. To request erasure of social-platform data, contact privacy@esek.io with the connected account or Page and the business concerned. We verify the request and its scope, including records shared with other channels, and confirm the outcome. A request that still needs verification or review is not treated as a completed deletion. Customers who do not have an Esek account can use the same contact address.
When Meta sends us a verified authorization-revocation notice, we stop the affected connection and clear its credentials. When Meta sends a data-deletion request, we also begin erasure of the associated social data and return a confirmation code and status-page link. Shared records, exports, recovery backups, or AI history may require additional review before the request can be marked complete. Removing a disconnected channel from the connection list only hides that connection; it does not erase its history.
9. Your Rights
Depending on your jurisdiction, you may have the right to access, correct, delete, or port your personal data, or to object to or restrict certain processing. To exercise any of these rights, contact us or see our Data Deletion page for the deletion request process. We will respond within 30 days.
10. Children's Data
Our service is not directed to individuals under the age of 16. We do not knowingly collect personal data from children. If we become aware that a child has provided us with personal data, we will delete it promptly.
11. Cookies and Local Storage
The web application uses a cookie to keep you signed in and a short-lived cookie that protects connecting a Google account. It keeps your sign-in, your settings and your cookie choice in your browser's local storage. If you arrive through a referral link, a cookie remembers the referral for 60 days. The application loads Google's libraries only when you connect Google Drive, loads Paddle's checkout only when you pay, and stores a notification token only if you turn on browser notifications.
Our marketing website and signup flow also store, in your browser, a randomly generated identifier and the campaign parameters from the address you arrived on. We use them to measure how the site performs and to attribute a signup to the advertisement that led to it. They are set by us and stay in our own systems, except for advertising measurement described below: for visitors who accepted marketing cookies, or who are outside the EU, the EEA and the United Kingdom and have not declined them, the Meta ad click identifier from that address is sent to Meta with the signup, as described under Meta (Conversions API) in section 5.
Our marketing website additionally carries advertising tags from Meta, Google Ads and OpenAI Ads. These tags set cookies and report page views and conversion events to their providers so we can measure and attribute our advertising. They load only after consent. Visitors in the EU, the EEA and the United Kingdom are asked before they load; elsewhere they load without a prompt. A browser that sends a Global Privacy Control signal is treated as having declined, wherever the visitor is. Advertising tags do not load automatically on ordinary application screens. A completed web signup can load the Google Ads and OpenAI tags to report that conversion under the same consent controls.
OpenAI Ads measurement uses the ad click reference (oppref), a browser reference, the site origin, event type and time, and a random signup event identifier. Its first-party cookies request lifetimes of 30 days for the click reference and 365 days for the browser reference. Where automatic advanced matching is enabled, the OpenAI Pixel can detect supported customer details on the website, normalize and hash them in the browser, and send the hashes with conversion events. Opening WhatsApp is recorded as an intent event, not as a received enquiry. We record an incoming source-labelled WhatsApp message in our own analytics; we do not send its text or the sender's phone number to OpenAI Ads through this integration. Declining marketing measurement prevents future OpenAI events and removes its measurement cookies.
12. Security
Connections between your device and Esek, and between Esek and external providers, use TLS. Since October 30, 2026, connections between our services, database and cache inside our production environment also use TLS with verified certificates. Passwords are hashed with bcrypt. Database access is restricted to authenticated services inside our private network. Access tokens received during WhatsApp Embedded Signup or Instagram and Messenger authorization are stored encrypted and scoped to the minimum required permissions.
13. Changes to This Policy
We may update this policy from time to time. Material changes will be communicated via email to registered Business Customers. Continued use of the platform after changes constitutes acceptance of the updated policy.
14. Contact
Questions about this policy? Contact us at privacy@esek.io.
15. Who we are
The controller of the personal data described in this policy is Zak Goichman, trading as Esek, a sole proprietor (עוסק מורשה) registered in Israel. For customer messaging data processed on a Business Customer’s behalf, that business determines the purposes of processing and Esek acts as its processor.
- Registered address: Herzl 100, Tel Aviv, Israel
- Email: privacy@esek.io
Our Israeli business registration number is available on request from legal@esek.io.