Subprocessors

Last updated: October 6, 2026

A subprocessor is a third party we use to deliver the Service that may handle customer data in the process. This page lists ours, what each one is for, and what reaches it. It is the companion to our Privacy Policy, which governs what we do with your data ourselves.

Layro is a production pipeline, so the longest groups below are the AI providers and the places we search: the work you ask for is by definition carried out by models we do not host, over material we do not own. Nothing in either group receives your account identity — the first receives the material to work on, the second a search query.

1. Infrastructure

Hetzner Online GmbH (EU)

Hosting of the application, database, and file storage

Data: All data you store in Layro, at rest in the European Union

Cloudflare

DNS and the network edge — every request to the Service passes through it, and TLS terminates there

Data: Connection metadata for every request: your IP address, the address requested, timing, and the request itself in transit

Amazon Web Services (S3)

Object storage for durability copies of project media

Data: Media files stored in your projects — files you upload, stock and other material you add, and generated media

2. Identity and payments

Supabase

Authentication, account records, two-factor enrolment

Data: Email address, credentials (we never see or store your password), session tokens, MFA factors

Google

Sign-in, where you choose to use it instead of an email address and password

Data: The email address and name the provider returns to us when you approve the sign-in

Paddle

Payments and subscriptions, acting as merchant of record

Data: Billing name and address, payment method (held by Paddle, never by us), tax location, invoice history

Resend

Transactional email — account and Picker Seat invitations

Data: Email address and the contents of the message sent to it

3. AI processing

These providers receive the material you ask us to work on: your script text, the audio and images in your project, and the instructions you give. For the voiceover provider you can connect a key of your own under Settings → Integrations, and those calls then run under your agreement with the provider instead of ours.

Anthropic

Language models — script writing, scene analysis, direction

Data: Script text, project brief, and prompts derived from them

Groq

Speech recognition for the subtitles workflow

Data: Audio you submit for transcription

ElevenLabs

Text-to-speech voiceover

Data: Script text to be spoken, and voice settings

Replicate

Image and video generation

Data: Generation prompts and any reference images you supply

Jina AI

Embeddings used to match footage to scenes

Data: Scene descriptions and candidate footage metadata

4. Research and content discovery

These services receive search queries generated from your script, or the identifier of an item you linked — not the script itself, and not your account details.

Perplexity

Web research for script material

Data: Research questions derived from your topic

Tavily

Web search for script material

Data: Search queries derived from your topic

Serper and Google Programmable Search

Web and image search

Data: Search queries derived from your scenes

YouTube (Google)

YouTube API Services — the title, channel, length and thumbnail of the YouTube videos you link, and the embedded player that plays them. YouTube is not searched on your behalf, and no video is downloaded from it

Data: The identifiers of the YouTube videos you link. The embedded player loads from YouTube in your browser, under the Google Privacy Policy

Pexels, Pixabay, Openverse

Licensed stock libraries

Data: Search queries derived from your scenes

Internet Archive

Openly licensed footage from an archive.org item you linked yourself: we read the item’s licence and file details and add the file to your project. The archive is not searched on your behalf

Data: The identifier of the archive.org item in the link you added

5. Operations

Sentry

Error monitoring

Data: Diagnostic data about failures — request paths and stack traces. Request bodies on authentication and conversation endpoints are scrubbed before transmission.

Changes to this list

We update this page when we add or remove a subprocessor. If you have a contract with us that requires advance notice of changes, we honour the notice period in that contract. To be told when this list changes, write to [email protected] and ask to be added to the notification list.

Data processing agreements

If you need a data processing agreement, or a copy of the safeguards we rely on for transfers outside the European Economic Area, write to [email protected] and we will send you our current terms.