Cogno Subprocessor List

Version: v1.0 Effective Date: August 25, 2026 Last Revised: August 25, 2026

This page is the Subprocessor list referenced in Section 10 of the Cogno Terms of Service and Sections 10 and 11 of the Cogno Privacy Policy.

A Subprocessor is a third party engaged by Genaxis, Inc. ("Company") to process Customer Data on the Company's behalf in the course of providing Cogno (the "Service"). The Company remains responsible to Customers for the acts and omissions of its Subprocessors.

Every Subprocessor below is bound by a written agreement imposing confidentiality and data protection obligations no less stringent than the Terms and the Privacy Policy, including standard contractual clauses or an equivalent transfer mechanism where personal information leaves Japan.


1. Infrastructure and Hosting

ProviderService ProvidedData HandledProcessing Location
Google LLC (Google Cloud)Application runtime, primary database, object storage, secret management, container registryAll Customer Data: tasks, comments, documents, uploaded files, meeting recordings, account information, application logsasia-northeast1 (Tokyo, Japan)
Vercel, Inc.Hosting and CDN for the marketing site (cogno.studio) and the web application (app.cogno.studio)Access logs, request metadata, IP addresses, cookie identifiersServer-side rendering in the United States (iad1, Washington, D.C.); static delivery from edge locations worldwide, including Tokyo
Supabase Pte. Ltd.Authentication (sign-in, session issuance, JWT) and the managed database backing the Company's self-hosted integration gatewayEmail addresses, user identifiers, authentication tokens, integration OAuth credentialsap-northeast-1 (Tokyo, Japan)

The Company's collaborative editing server (Hocuspocus) and integration gateway (Nango) run as self-hosted images on Google Cloud Run in Tokyo. They are the Company's own infrastructure, not third-party services, and no Customer Data is transmitted to their upstream vendors.


2. AI Processing

ProviderService ProvidedData HandledProcessing Location
Amazon Web Services, Inc. (Amazon Bedrock)Large language model inference (Anthropic Claude Sonnet 4.5)Content submitted as prompts: task descriptions and comments, documents, meeting transcripts, and repository context supplied by the Customerap-northeast-1 (Tokyo), using the Japan cross-Region inference profile jp.anthropic.claude-sonnet-4-5

Data residency. The Service calls Claude exclusively through the Japan cross-Region inference profile. Requests are processed entirely within Japanese AWS Regions (Tokyo and Osaka) and routed over the AWS internal network without traversing the public internet.

Model provider access. Because inference runs inside Amazon Bedrock, prompt content is not disclosed to the model provider (Anthropic, PBC) and is not used to train any model, in accordance with the Amazon Bedrock service terms.

Sole inference path. All large language model inference in staging and production runs through Amazon Bedrock. The Google Generative AI SDK present in the codebase serves a fallback that activates only in local development environments where Bedrock is unconfigured, and is not invoked in production.


3. Meeting Recording (Cogno Notetaker)

ProviderService ProvidedData HandledProcessing Location
Hyperdoc Inc. (d/b/a Recall.ai)Meeting bot that joins calls, captures audio and video, and produces transcripts; calendar sync that schedules bots for upcoming meetingsMeeting audio and video, transcript text, participant display names, meeting metadata; for Customers who connect a calendar, event titles, times, and meeting URLsap-northeast-1 (Recall.ai Asia region, Tokyo)

Retention. Video recording is enabled on the Service, so recordings are deleted from Recall.ai after 7 days. Recordings the Customer retains are transferred to the Company's Google Cloud Storage bucket in Tokyo and deleted after 90 days, with a bucket-level lifecycle rule deleting any remaining object at 120 days.

Notice to participants. The bot joins under a visible display name and, by default, posts a notice in the meeting chat stating that the meeting is being recorded.

Recall.ai's own subprocessors. Recall.ai engages its own subprocessors, including for transcription. Its current list is published at security.recall.ai/subprocessors.


4. Integration Infrastructure

ProviderService ProvidedData HandledProcessing Location
Sampark Inc. (d/b/a Composio)OAuth brokering and tool catalog for the Customer's own Sentry connection, the only integration routed through Composio todayOAuth credentials for that connection and the issue data retrieved through itUnited States

5. Payments

ProviderService ProvidedData HandledProcessing Location
Stripe, Inc.Subscription billing, checkout, and the customer billing portalBilling contact details, email address, subscription and invoice records, payment method identifiersUnited States

Card numbers and other payment credentials are collected by Stripe directly and never pass through or are stored on the Company's systems.


6. Monitoring and Analytics

ProviderService ProvidedData HandledProcessing Location
Functional Software, Inc. (d/b/a Sentry)Error tracking and performance monitoringStack traces, error messages, request metadata, user and workspace identifiersEuropean Union (Sentry EU data region, Germany)
PostHog, Inc.Product usage analyticsProduct events, user and workspace identifiers, device and browser information, IP addressesUnited States (PostHog US Cloud)

7. Email

ProviderService ProvidedData HandledProcessing Location
Plus Five Five, Inc. (d/b/a Resend)Transactional email delivery — workspace invitations, notifications, and contact form messagesRecipient email address, display name, message contentUnited States

8. Documentation Site

ProviderService ProvidedData HandledProcessing Location
Mintlify, Inc.Hosting for the developer documentation served at cogno.studio/docsDocumentation page requests and their metadataUnited States

The documentation site contains no Customer Data.


9. Not Subprocessors

The following receive data but are not Subprocessors of the Company, because the Customer directs the transfer and holds its own relationship with the provider.

Customer-connected services. Slack, Discord, Microsoft Teams, GitHub, Google Workspace (Calendar and Gmail), and Chatwork. The Customer authorizes each connection and controls its scope; the Company transmits data to these services only as instructed through the Service. Each provider's own terms and privacy policy govern the data once it arrives.

Customer-supplied coding agent credentials. Where a Customer supplies its own Anthropic API key, Claude Code subscription token, or OpenAI API key, the resulting coding agent runs on the Customer's own host under the Customer's own agreement with that provider. The Company does not hold a processing relationship with Anthropic, PBC or OpenAI for this traffic.


10. Changes to This List

10.1 The Company will update this page before, or as soon as reasonably practicable after, engaging a new Subprocessor.

10.2 Material additions will be announced on the Company's website and, for Customers on a written agreement, by email to the registered administrative contact.

10.3 A Customer with a reasonable data protection objection to a new Subprocessor may raise it in writing to privacy@cogno.studio within thirty (30) days of notice. The Company will work in good faith to offer a commercially reasonable alternative; if none is available, the Customer may terminate the affected Service in accordance with the Terms.


11. Requests and Contact

For a copy of this list as of a particular date, a data processing agreement, or a security questionnaire, contact:

Genaxis, Inc. Delaware, United States of America Japan contact office: MIEUX Shibuya Building 8F, 5-3 Maruyamacho, Shibuya-ku, Tokyo 150-0044, Japan Contact: privacy@cogno.studio

Subprocessors | Cogno