Subprocessors
Every service provider that touches your data is on this list
We rely on specialist service providers for parts of running the platform. Below we list each one's purpose, the data it processes and its location. Before adding a new subprocessor, we give our tenants 30 days' notice.
Send a requestPlatform subprocessors
Location shows where data is processed or where the provider is headquartered. Every location outside Türkiye is a cross-border transfer under KVKK Article 9.
| Provider | Purpose | Data processed | Location |
|---|---|---|---|
| Hetzner | Server, storage and backup hosting | All tenant data on the platform | Germany |
| Cloudflare | DNS, content delivery, DDoS protection, firewall, TLS | IP address, request details, traffic in transit | Global edge network, headquartered in the US |
| Mailgun | Transactional email (notifications, sign-in codes, reconciliation letters) | Recipient name and email, email content | Headquartered in the US |
| Turkcell e-Şirket (platform trial account) | Letting tenants that have not connected their own provider try the e-document flow in a test environment; documents are not transmitted to GİB and have no official status | Document content, buyer name, tax number, address, line items and amounts | Türkiye |
| NETGSM | SMS delivery (sign-in codes, reminders) | Phone number, message content | Türkiye |
| OpenRouter | AI model routing gateway: forwards the request to the model provider serving the chosen model; the models in use today are Anthropic's Claude models | Only for companies whose owner has switched AI on and given consent: in AI chat, the message the user writes, the conversation history and the AI agent's instructions (chat does not access panel records); in automated tasks, conversation text from the connected messaging account and the related lead record. Requests are sent so that they are routed only to endpoints with data retention disabled (zero data retention) and that do not use data for training | United States; plus the processing region of the model provider serving the request (e.g. United States) |
| Anthropic (Claude models, via OpenRouter) | Running the AI model. We have no direct contract with Anthropic; OpenRouter forwards the request. We list it separately because it is the party that actually processes the request | Request content coming from OpenRouter (the data in the OpenRouter row above) | United States; the request may be processed on Anthropic's own infrastructure or on the cloud infrastructure serving the model (e.g. Amazon Bedrock, Google Vertex AI); the processing region of that infrastructure may also be outside the United States |
| EVREN AI platform | Open-source language models operated by SSB and hosted in Türkiye | Not active at present. Once switched on: the message written in AI chat, the conversation history and the AI agent's instructions; data does not leave Türkiye | Türkiye |
Access to the AI model
AI features run per company: the company owner switches them on from the AI screen and gives consent; without the owner’s consent no data is sent to a model. The active route today is OpenRouter, and the models in use are Anthropic’s Claude models. We have no direct contract with model providers (such as Anthropic); OpenRouter forwards the request, which may be processed on Anthropic’s own infrastructure or on the cloud infrastructure serving the model (Amazon Bedrock, Google Vertex AI). Requests to OpenRouter are sent so that they are routed only to endpoints with data retention disabled (zero data retention) and that do not use data for training; free (data-collecting) models are not used in automated tasks. Zero data retention concerns the endpoint that processes the request; limited retention for abuse monitoring is subject to the relevant provider’s terms.
The data sent to the model depends on the feature you switch on: in AI chat, the message you write, the conversation history and the AI agent’s instructions (chat does not access your panel records); in the automated task that runs overnight, the previous day’s conversation text from your connected messaging account (WhatsApp, Chatwoot) and the related lead record. The conversation text includes internal (private) team notes in your messaging account. From the lead record only the name, stage and the needs profile the AI previously extracted are sent; custom fields and Conversation Tracking notes are not sent. Document parsing (PDF text and e-invoice XML) runs on the platform’s own servers, and no data is sent to a model for it.
EVREN is operated by SSB and hosted in Türkiye; it is not active at present. Once switched on, data on that route does not leave Türkiye and this page is updated beforehand.
Services that receive no personal data
Some services supply data to the platform but receive no personal data from us, so they are not subprocessors. For example, we get exchange rate data from CollectAPI; no personal data is sent to that service.
Providers for subscription and credit payments (no tenant data)
Card payment for your Müşavir subscription and AI credit packages is taken through iyzico Ödeme Hizmetleri A.Ş. (Türkiye). As a payment institution licensed under Law No. 6493, iyzico processes card data under its own responsibility as an independent data controller; it has no access to tenant data on the platform. We store only iyzico customer and subscription reference codes and the last four digits of the card.
Services the tenant connects
The services below only run if you connect them. Credentials are stored encrypted with AES-256-GCM; data goes to these providers during an action you start.
| Service | Purpose | Data sent |
|---|---|---|
| İmzala | E-signing the contract when a proposal is accepted | Signer name, contact details, contract document |
| PayTR | Accepting online payments | Payment amount, customer name and contact details |
| Paraşüt (Paraşüt Yazılım A.Ş., Türkiye) | Sending invoice and account records to your Paraşüt account; when you connect it, transmitting the e-invoices and e-archive invoices you issue to GİB through Paraşüt | Invoice content, buyer name, tax number, address, line items and amounts; account details |
| Your own Turkcell e-Şirket account | When you connect your own Turkcell e-Şirket account, transmitting the e-invoices, e-archive invoices, e-dispatch notes and e-SMM you issue to GİB | Document content, buyer name, tax number, address, line items and amounts |
| WhatsApp and Chatwoot | Inbox conversations; if the AI automated task is on, updating the lead record based on conversations | Message metadata and preview; if the automated task is on, conversation text is read and sent to the AI provider |
| WhatsApp (Meta Platforms; the line you connect yourself) | Delivering to the recipient the notification and reply messages you start through the panel, the API or an automation. Müşavir carries out this delivery as a data processor on your behalf | The recipient’s phone number or WhatsApp group ID, message content. Message content is not kept in our records; the activity log holds only the time of sending, the target type and the last four digits of the number. Meta Platforms is headquartered in the USA (in Europe, Meta Platforms Ireland); this is a transfer abroad |
| Telegram (your own Telegram bot) | Delivering the AI task report to your team | Numeric summary; no person names or phone numbers are sent |
| Meta (your own ad account, Conversions API) | Reporting a conversion event to your company’s Meta dataset when a quote is sent to, or accepted by, a person who came from an ad | Only for people with an ad trace: hashed email and phone, event type, quote amount. Hashed data can still be personal data; it is transferred to Meta (abroad, mainly the USA) |
If you do not connect one of these services, the related task either runs on the platform’s own infrastructure or does not run; connecting them is optional. E-invoices, e-archive invoices, e-dispatch notes and e-SMM are transmitted to GİB only once you connect your own provider account: you issue the document and your provider transmits it; Müşavir is not a private integrator and does not transmit documents on your behalf through its own account.
Change notification
When a new subprocessor is added, or the scope of data an existing one processes expands:
- The change is announced at least 30 days in advance on this page and by email to tenant administrators.
- Tenants with a data processing agreement can send a reasoned objection through the contact page within that period.
- For changes made for urgent security reasons, notice is given as soon as possible, together with the reason.
- This notice applies to the platform subprocessors table above. Services you connect (the tenant-connected services table) are enabled by your choice and under your own agreement with the provider; when we add a new function to one of them we announce it, but no data is sent until you connect the service.
Frequently asked questions
Will I be told if a new subprocessor is added?
Yes. We announce a new subprocessor to tenants by email and on this page at least 30 days before it goes live. Tenants with a data processing agreement may raise a reasoned objection within that period.
Why are the services I connect myself not on this list?
Accounts such as e-signature, online payments, accounting export, your e-invoice provider (Turkcell e-Şirket or Paraşüt), WhatsApp, the inbox and Telegram are connected by you, and you choose which provider to work with. That relationship is between you and them; we list them separately below.
Do you have contracts with subprocessors?
We have a contractual relationship with every subprocessor we work with directly, and we aim to keep it consistent with the data protection obligations we take on towards you. We have no direct contract with model providers (such as Anthropic); the request is forwarded through OpenRouter and under OpenRouter's terms.
Does the AI provider store my data?
AI features run per company, once the company owner switches them on and gives consent. The active route today is Anthropic's Claude models via OpenRouter; EVREN is not active at present. Requests via OpenRouter are sent so that they are routed only to model providers with data retention disabled (zero data retention) where offered and that do not use data for training; limited retention for abuse monitoring is subject to the relevant provider's terms. Data sent to EVREN stays in Türkiye.
Related
Trust Center →
Security, KVKK and GDPR, subprocessors, backups and the compliance roadmap in one place.
KVKK and GDPR →
Data subject rights, how to apply, cross-border transfer and AI processing.
Data Processing →
Controller and processor roles, data categories per module and retention.
Security →
Infrastructure, tenant isolation, encryption, access control and audit trail.
Integrations →
Your email, SMS, e-signature, payment and messaging accounts, connected and encrypted.
Let us attach the subprocessor list to your contract
We send the current list as an annex together with your data processing agreement request.
Send a request