Legal
Subprocessors
Last updated: October 16, 2026
To run its AI agents (voice calls, WhatsApp, website chat, SMS and email) and its Rep Parallel Dialer, AIforge Tech Private Limited (operating as Finn) runs its own models on its own infrastructure (shown as “Finn” below) and engages the third-party subprocessors listed below. Each processes personal data under its written data processing terms with Finn, and only to provide the function listed. Stored data rests in the region agreed for your workspace (India by default); Rep Parallel Dialer recordings rest with our telephony provider and error logs with our monitoring provider, in the locations listed below. Enterprise customers can request additional commitments, such as zero data retention or in-region processing, as part of their agreement. Each call or conversation uses only the vendors in its agent’s configuration; the list below is the complete set a configuration can draw from.
- No training on your data
- Zero retention on Enterprise where offered
- Stored data stays in your region (dialer recordings and error logs excepted)
- 14-day notice of new vendors
Pick a region to see where each vendor processes data for it.
| Subprocessor | Purpose | Countries |
|---|---|---|
| Cloud infrastructure | ||
| Microsoft Azure | Hosting, storage and AI models | Your workspace region |
| AI models | ||
| Finn | AI models (self-hosted) | Your workspace region |
| Google Cloud | AI models | Your workspace region for supported models; other models outside it |
| Anthropic Claude (via Google Cloud) | Claude AI models, run by Google | USA, EU or Singapore |
| OpenAI models (via Microsoft Azure) | OpenAI AI models, run by Microsoft | Your workspace region for supported models; other models outside it |
| OpenAI | OpenAI AI models (direct API) | USA, UAE or EU |
| Anthropic (Claude) | Claude AI models (direct API) | USA |
| Deepgram | AI models | USA, EU or India |
| ElevenLabs | AI models | USA, EU, India or Singapore |
| Cartesia | AI models | USA, EU, UK or India |
| Sarvam AI | AI models for Indian languages | India |
| Inworld | AI models | USA, EU or India |
| Smallest AI | AI models | India or USA |
| Telephony | ||
| Plivo | Calls and SMS; Rep Parallel Dialer | USA or India |
| Exotel | Calls and SMS for Indian numbers | India or Singapore |
| Twilio | Calls and SMS | USA or Ireland (EU) |
| Telnyx | Calls and SMS | USA, Germany (EU) or UAE |
| RingCentral | Calls and SMS | USA, UK or Germany (EU) |
| Messaging & email | ||
| Meta Platforms (WhatsApp) | WhatsApp messaging | Your workspace region |
| Google Workspace | Email (Gmail) | USA or EU |
| Payments | ||
| Razorpay | Payments in India | India |
| Lemon Squeezy | International payments | USA |
| Analytics, monitoring & data | ||
| Sentry | Error monitoring | EU |
| PostHog | Website and product analytics | USA |
| Cloudflare | Bot protection and network security | Global |
| ContactOut | Contact enrichment for uploaded audiences | USA |
| gtm.dev | Company research for dialer lead briefs | USA |
Enterprise adds in-region and zero-data-retention configurations where the provider offers them; with zero data retention the provider keeps neither the prompt nor the output after responding.
Notes
- Finn’s own AI models run on Finn-operated infrastructure in your workspace region. Data they process is not sent to the models’ developers.
- Telephony providers listed above connect calls and messages. The downstream networks they route traffic over carry it in transit only and are not Finn subprocessors.
- Our chat inbox is self-hosted on Finn-operated infrastructure. No conversation data goes to the software’s maker.
- Rep Parallel Dialer recordings are stored by the telephony provider (Plivo) and error logs by the monitoring provider (Sentry), in the locations shown for those rows.
- When you connect your own WhatsApp Business account, CRM, calendar or other integration, that provider processes the data under your own agreement with it.
- Where data is processed outside your workspace region, the transfer is covered by appropriate safeguards, such as the Standard Contractual Clauses, as set out in our Data Processing Addendum.
- Exact processing locations for any subprocessor are available on request. For Enterprise customers, the subprocessor schedule in your agreement takes precedence over this page.
Changes to this list
We update this page when we add or remove a subprocessor and give at least 14 days’ notice of any new one. To subscribe to notices, object to a change or ask a question, contact [email protected].
Version history
- Rebuilt as a list you can filter by workspace region and plan. Added Finn (self-hosted, shown as first party), OpenAI and Anthropic (direct); removed Resend and Dodo Payments; the UAE replaces Australia; locations given at country level; zero data retention defined; Dialer recordings and error logs noted as resting with the telephony and monitoring providers.
- Removed Intercom: website chat on hirefinn.ai now runs on our own Platform.
- Noted the open-source text-to-speech models we run (Qwen, Kokoro, Kyutai).
- Covers chat AI agents and the parallel dialer.
- Region-specific data-location views; vendor locations confirmed (payments, enrichment, Plivo in India); AI model regions in plain language.
- Noted zero data retention for Google Cloud and Azure AI and speech services; location corrections (PostHog, ContactOut, payment providers).
- Disclosed both Google Cloud and Microsoft Azure; reconciled with the Trust Center.
The exact text of every earlier version is kept on record and available on request at [email protected].
Get started
Hire Finn and scale with confidence.
Move from idea to live voice automation — securely, reliably, and without operational risk.
Fort-nightly Launches
We move quickly and get you what you need
Powerful Tools
Pre-built dashboards, reports, automations, more