Subprocessors

Last updated: 2026-10-11

A subprocessor is a third party that Unshift LLC ("Unshift") engages to process personal data on behalf of our customers in the course of delivering the Services (the Unshift Studio, hosted websites, and the deployments Unshift runs for partners). This page lists our current subprocessors and is incorporated by reference into our Privacy Policy and our Data Processing Addendum.

If you are a customer who requires advance notice of new subprocessors, subscribe to this page's changes at subprocessors@unshift.ai and we will email you at least 30 days before adding or replacing a subprocessor that processes Customer Personal Data.

Current subprocessors

Customer-initiated integrations (for example, connecting a customer's own AI provider over the Model Context Protocol, sending form submissions to a webhook or to a tool the customer connects, such as Zapier, Make, Mailchimp, Brevo, ConvertKit or Klaviyo, importing content from a source the customer connects, such as WordPress or Contentful, or configuring a custom analytics provider in their site) add additional data flows that are outside the scope of Unshift's subprocessor obligations because the customer is controlling them directly. See "Your own AI assistant is not an Unshift subprocessor" below.

Form submissions from published sites are not stored in any database. Each one passes through our servers only to be delivered: by email through Amazon SES, and to a webhook or a connected tool where the customer has set one up.

Partner deployments

A software platform can offer Unshift's website builder to its own customers under the platform's brand (a "partner deployment"). On a partner deployment:

• The database that holds the accounts and site content of the partner's customers is a Supabase project dedicated to that partner. Either Unshift operates it for the partner, in Unshift's own Supabase organization and in the AWS region agreed with the partner, or it is in the partner's own Supabase account, in which case Supabase is engaged by the partner directly for it. Either way, that data is not stored in Unshift's own database.
• Unshift runs the partner's builder and site API on Vercel, and hosts the published sites of the partner's customers on Cloudflare. These deployments are dedicated to the partner or, on shared hosting, serve several partners, each with its own database as above.
• Some services are shared with Unshift's own: the build worker on Fly.io, image storage on Amazon S3 in Unshift's AWS account (in a folder of the partner's own), and email sent through Amazon SES in Unshift's AWS account, including form submission emails.
• Form submissions are not stored in any database, as on Unshift's own service.
• Commerce, where the partner offers it, runs on a Stripe Connect platform set up for that partner, not on Unshift's.
• The Partner Portal, where a partner's team manages its customers' sites, is an Unshift application hosted on Vercel. It keeps no content or contact details of the partner's customers: it reads them from the partner's builder when a page is opened. The accounts of the partner's team (name, email address, role and sign-in times) are in Unshift's own central database on Supabase, for which Unshift is the controller (see our Privacy Policy).

The partner's written agreement with Unshift, including its data processing terms, governs this processing. The subprocessors listed above apply to the parts of a partner deployment that Unshift operates.

How we select subprocessors

We require every subprocessor to:

1. Execute a Data Processing Agreement (DPA) that includes the EU Standard Contractual Clauses (Module Three or Module Four as applicable) and the UK International Data Transfer Addendum where relevant.
2. Maintain security certifications appropriate to the data they handle (SOC 2 Type II, ISO 27001, or equivalent).
3. Notify us of security incidents affecting our tenant within a defined timeframe (24 to 72 hours depending on the vendor).
4. Be contractually prohibited from using Customer Personal Data to train their own models, except as strictly necessary to provide the service we purchased.

Your own AI assistant is not an Unshift subprocessor

You can connect your own AI assistant (for example, Claude or ChatGPT) to the Unshift Studio over the Model Context Protocol (MCP), authorized through OAuth. That reasoning runs inside your own AI application on your own provider account, under your own agreement with that provider.

As a result:

• Your conversation with your own AI assistant stays between you and the provider you chose. Unshift does not receive your model prompts from it.
• Unshift's MCP server only performs deterministic, non-AI work that your assistant asks it to do (for example, reading a URL you provide, extracting a theme, assembling or editing blocks and pages, and deploying a site). It receives structured tool calls, not your model prompts.

Because the AI provider relationship is yours and you control it directly, your own AI provider is not a subprocessor of Unshift, in the same way that a tool you connect to receive your form submissions is not. On a partner deployment, the partner decides whether this connection is available to its customers. Your use of that provider is governed by the provider's own terms, privacy policy, and data-handling and training practices, which you should review. If you want to understand how prompts are retained or whether they are used for training, consult your provider's documentation and account settings.

This is separate from the AI features that Unshift runs itself, such as AI features inside the Studio and the site importer. Those use Anthropic under Unshift's own account, and Anthropic is listed as a subprocessor in the table above. That processing is held to the requirements in "How we select subprocessors" above, like every other subprocessor's, and does not use an AI provider account of yours.

How we notify you of changes

We maintain this page as the single source of truth. When we add, remove, or replace a subprocessor that processes Customer Personal Data, we:

1. Update this page.
2. Email customers on the notification list (subscribe at subprocessors@unshift.ai) at least 30 days in advance.
3. If a customer objects to a new subprocessor on reasonable grounds related to data protection, we work with the customer to find an alternative. If no alternative is possible, the customer may terminate the affected part of the Services with a pro-rata refund.

In emergency cases (for example, a subprocessor suffers a material breach and we must replace them faster than 30 days allows), we notify as far in advance as practical and document the reason.

Historical changes

Future additions, removals, and replacements will be appended here.

Contact

• Subprocessor questions or objections: privacy@unshift.ai
• Subscribe to subprocessor change notifications: subprocessors@unshift.ai
• Postal address: Unshift LLC, 1309 Coffeen Avenue STE 1200, Sheridan, Wyoming 82801, United States