Version 1 — published July 27, 2026
Subprocessor List
Version: July 25, 2026. This page is referenced by contract: keep it literally accurate, keep the version date current, and change it only through the process at the bottom.
Honeybee Logic uses the following third-party providers ("Subprocessors") to deliver the Services.
How to read the tiers:
- Tier 1 — core processing. May process Customer Data, including Personal Information. Held to the Data Privacy Agreement's full bar: no training on Customer Data (we never grant permission), any retention limited to short-term service-integrity/abuse-prevention storage (currently ≤30 days), Canadian region wherever available.
- Tier 2 — limited scope. Receives only the content submitted to specific media features (for example, text-to-speech or image generation). Those features are documented and signposted in-product as not intended for personal information. Tier 2 providers never receive student records, account databases, or bulk Customer Data.
Tier 1 — core processing
| Provider | Purpose | Processing location | Data handling |
|---|---|---|---|
| Amazon Web Services (AWS) | Application hosting, database, file storage, backups, user authentication (Amazon Cognito) | Canada (AWS ca-central-1, Montréal) | Customer Data at rest stays in Canada |
| AWS — Amazon Bedrock | AI model inference for product features | Canada (in-region models). Any cross-region route that processes in the US will be marked here before use | No training on customer content; retention governed by the configured Bedrock data-retention mode; model providers do not receive data on Bedrock's default modes |
| Stripe | Subscription payments, billing, and invoicing (individual accounts) | United States (global infrastructure) | Card data is held by Stripe (PCI-DSS Level 1), never by us; Stripe receives only what checkout requires (name, email, payment details, transaction records) — no Customer Data or student-related content |
| AWS — Simple Email Service (SES) | Application-generated email (invitations, account and service notifications) | Canada (ca-central-1) | Receives recipient addresses and message content only for delivery; no bulk Customer Data |
| Google Workspace | Company email and business communications (including support requests customers send us) | United States / global | Receives whatever a customer chooses to email us; not connected to product systems and never used to store or process Customer Data from the Services |
Tier 2 — limited scope (media features only; not for personal information)
| Provider | Purpose | Processing location | Data handling |
|---|---|---|---|
| ElevenLabs | Text-to-speech for designated features | United States | No training on customer content (including via its upstream LLM providers, which its terms bar from training); provider-standard retention; feature signposted in-product as not for personal information |
| fal.ai | Image/media generation for designated features | United States | Output expiration configured per request; generated files we keep are copied to our Canadian storage and expired at the provider; feature signposted in-product as not for personal information |
Pre-vetted alternates (not currently in use)
These providers are vetted and listed for early transparency, but do not currently receive Customer Data. If we decide to activate one, we will follow the DPA §6.3 process — at least 30 days' written notice to organizational customers (or the urgent-replacement path, where needed) — and move it into the tables above.
| Provider | Would be used for | Conditions we impose on ourselves |
|---|---|---|
| OpenRouter | AI model routing (fallback / model optionality) | Zero-data-retention-only routing enforced at the account level; downstream endpoints restricted to no-training providers; no OpenRouter plugins/tools (e.g., web search) on Customer Data paths |
| Moonshot AI Kimi models — via Amazon Bedrock only | AI model inference | Bedrock-hosted exclusively (AWS terms apply; no data flows to Moonshot). The direct Moonshot API is never used for Customer Data |
How this list changes
For organizational customers, we give at least 30 days' written notice before adding or replacing a Tier 1 Subprocessor (DPA §6.3). If a replacement is urgently needed to protect the security or availability of the Services, we may make it immediately and notify without undue delay afterward, with the same objection right. This page always shows the current list and its version date.