Privacy Policy

Last updated: June 8, 2026

This policy describes how Switchy handles your data. It complies with the EU General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA).

What changed (June 2026): Switchy now connects to Slack and Microsoft Teams and can ingest message content from channels you explicitly link to a Project; an AI agent can fetch from and (only after a human approves any write) act on your connected tools; admins can import historical chat from an export archive; and your team memory can be exposed to external AI clients (Claude, ChatGPT, Cursor) through a per-organization MCP server. This policy now describes each of those flows, expands the sub-processor list, and corrects our data-retention windows.

1. Controller and Contact Information

The data controller responsible for processing your personal data is:

Switchy

Email: contact@switchy.build

Website: https://switchy.build

If you have questions about how we handle your data or wish to exercise your rights, please contact us at contact@switchy.build.

2. Data We Collect

We collect and process the following categories of personal data:

2.1 Account Data

When you register, we collect your name, email address, and profile picture (if provided via OAuth). If you sign in via Google or GitHub, we receive your public profile information from those providers.

2.2 Chat and Conversation Data

We process the content of messages you send and receive through AI models. This includes text prompts, model responses, and any files or images you upload during conversations.

2.3 Memory Data

Switchy stores conversation memories (facts, preferences, context) to provide persistent memory across sessions and models. You have full control over these memories and can view, pin, or permanently delete them at any time.

Every memory carries a visibility tag - Private (only the author can ever retrieve it), Project (members of one chat room), or Org (everyone in the organization). The default is Private. Visibility is enforced at the database query layer, not in application code, so private memories never leak across users even if a teammate searches for the same content.

For memory created from a connected surface, integration, or imported archive, the visibility is computed from the source - you do not pick it manually, and the system fails closed to the most private option when a source can't be mapped. The subsections below explain each source.

2.3.1 Organization-scoped data

When you join or create an organization, your account becomes a member of that org. Org admins can see your org-wide presence, role, and aggregate usage but cannot read your private memories or your messages in Projects you have not added them to. Billing and credit-pool data is org-scoped, not personal.

2.3.2 Slack and Microsoft Teams message content

When an admin links a Slack or Teams channel to a Project and turns on auto-memory, Switchy's bot reads the messages posted in that channel and a heuristic extractor decides what (if anything) is worth keeping as a memory. The bot answers only when it is @mentioned, but it ingests every message on a linked channel (very short messages, under five characters, are dropped). On Teams, the bot can only read channels it has been added to, which Microsoft enforces.

Visibility is computed from where the message came from: a public channel maps to Org (or to the linked Project, if the channel is mapped to one); a private channel maps to that Project, and if it can't be mapped it fails closed to Private. A direct message to the bot (a Slack DM or a Teams 1:1) becomes a Private memory owned by you in your personal organization. If the person messaging the bot is not a linked Switchy user, nothing is written.

2.3.3 Connected-tool data during agent tasks

When you ask the AI to perform a task, it can fetch data from your connected accounts (for example GitHub, Notion, or Gmail) through our integration provider, Composio. Any write or otherwise sensitive action requires explicit human approval before it runs (see our Terms and Security pages). The credential for a connected account is never placed in tool output, message content, logs, or the model's context - it is injected only at the moment of the outbound call. We do store analytics about each tool call (the tool name, the surface, whether it succeeded, and the organization/Project/user identifiers). Where tool output is saved as memory, it fails closed to Private, owned by the person who ran the task.

2.3.4 Imported export archives

An organization admin on a Pro or higher plan can upload a Slack or Teams Export archive to backfill historical chat into memory. Those historical messages run through the same pipeline and visibility rules as live ingestion (channel messages map to Org or the relevant Project; direct messages map to Private per sender; messages from senders who are not matched to a Switchy user are skipped). Personal data detected in the content is redacted before anything is persisted. Every memory created by an import is tagged with the import's identifier, and an admin can undo an entire import in one action, which deletes exactly that import's memories within the organization.

An export archive can contain the historical messages of third parties (for example former teammates or external participants). The admin who uploads it is responsible for having a lawful basis to import and process that content.

2.3.5 Memory exposed to external AI clients (MCP)

Your team memory can be exposed to external AI clients - such as Claude, ChatGPT, or Cursor - through a Switchy MCP server, so the same context shows up inside the tools your team already uses. Access is authenticated by a per-organization API key (stored only as a hash and revocable at any time). An organization-scoped key with no associated user sees only Org-visibility memory - never Project or Private memory. A key that carries a user identity and the “act on behalf” scope applies that user's normal visibility rules, always clamped to the key's organization. A request for a Project that doesn't exist (or belongs to another organization) returns “not found,” so the existence of other organizations' data is never revealed.

2.4 Usage and Analytics Data

We collect data about how you use the platform, including: models used, number of requests, token counts, API calls, feature usage patterns, and billing-related metrics. This data is used to provide usage dashboards, enforce plan limits, and improve the service.

2.5 Technical Data

We automatically collect technical information including IP addresses, browser type and version, operating system, device identifiers, referral URLs, and session duration. We use cookies and similar technologies as described in Section 8.

2.6 Payment Data

Payment processing is handled by Stripe. We do not store your full credit card number, CVV, or bank account details. We retain only the last four digits of your card, card type, and billing address for record-keeping purposes.

2.7 Communication Data

If you contact us via email or the contact form, we process your name, email address, message content, and any attachments you provide.

3. Legal Basis for Processing (GDPR Art. 6)

We process your personal data on the following legal bases:

  • Contract performance (Art. 6(1)(b) GDPR): Processing necessary to provide the Switchy service, manage your account, process payments, and deliver AI model responses.
  • Legitimate interests (Art. 6(1)(f) GDPR): Service improvement, fraud prevention, platform security, analytics, and marketing our services to existing customers. Our legitimate interest is balanced against your privacy rights.
  • Consent (Art. 6(1)(a) GDPR): Where we rely on consent (e.g., for optional cookies or marketing communications), you may withdraw consent at any time without affecting the lawfulness of prior processing.
  • Legal obligation (Art. 6(1)(c) GDPR): To comply with tax, accounting, and regulatory obligations under applicable law.

4. How We Use Your Data

We use your personal data for the following purposes:

  • Providing and operating the Switchy platform, including AI chat, memory storage, and API access
  • Processing your prompts through third-party AI model providers via OpenRouter
  • Storing and retrieving conversation memories to enable cross-model, cross-session context
  • Managing your account, authentication, and subscription billing
  • Providing usage analytics, dashboards, and billing reports
  • Communicating with you about your account, service updates, and support requests
  • Detecting and preventing fraud, abuse, and security incidents
  • Improving and developing new features based on aggregated, anonymised usage patterns
  • Complying with legal obligations, including tax and accounting requirements

4.1 No training on your content

Switchy never trains any model on your content, and never sells your data. Inference is performed by the model providers via OpenRouter under their API terms; the major providers (Anthropic, OpenAI, Google) do not train on data submitted through their APIs by default.

5. Data Sharing and Third-Party Processors

We share your data with the following categories of recipients, each bound by data processing agreements (DPAs) compliant with GDPR Art. 28:

ProcessorOpenRouter
PurposeModel inference routing
DataChat, image, and video prompts and responses
LocationUnited States
ProcessorModel providers via OpenRouter (Anthropic, OpenAI, Google, xAI, Meta, Mistral)
PurposeModel inference
DataThe prompts and responses routed for inference
LocationUnited States / provider regions
ProcessorGoogle Cloud Platform
PurposeHosting, database, Secret Manager (credential storage). At-rest encryption is Google-managed (AES-256).
DataAll platform data
LocationEU (europe-west1)
ProcessorAbly
PurposeRealtime message delivery
DataMessage envelopes + typing indicators
LocationEU + global edge
ProcessorStripe
PurposeBilling and payments
DataBilling and payment information (no chat content)
LocationUnited States
ProcessorComposio
PurposeOutbound tool execution. Holds per-user OAuth tokens for connected accounts; Switchy stores only opaque connection handles.
DataPer-user OAuth tokens (held by Composio) + the inputs and outputs of each tool call the agent runs
LocationUnited States
ProcessorResend
PurposeTransactional email delivery
DataEmail addresses and message content
LocationUnited States / EU
ProcessorSlack
PurposeSurface integration; ingests linked-channel and DM message content
DataMessage content from channels you link and DMs to the bot
LocationUnited States
ProcessorMicrosoft (Teams)
PurposeSurface integration; ingests added-channel and 1:1 message content
DataMessage content from channels the bot is added to and 1:1 messages
LocationUnited States / global
ProcessorPostHog
PurposeSession replay and product analytics (masks all inputs; excludes auth/billing pages; 30-day retention)
DataMasked session recordings + product usage events
LocationEU
ProcessorGoogle Analytics
PurposeAggregate web analytics (cookieless, storage: 'none' mode)
DataAggregate, non-cookie page-view data
LocationUnited States
ProcessorGoogle Ads
PurposeAd conversion tracking
DataConversion events
LocationUnited States
ProcessorAmplitude
PurposeProduct analytics
DataUsage events
LocationUnited States
ProcessorNextAuth / Auth.js
PurposeAuthentication
DataOAuth tokens and session data
LocationSelf-hosted (EU)

When you connect a third-party MCP integration (e.g. GitHub or Notion via the one-click connectors, or any custom HTTPS MCP), Switchy acts as an intermediary that forwards your queries to that third party and returns their responses to your chat. The third-party processor terms in those services govern that data flow:

ProcessorGitHub (when connected)
PurposeRepo / PR / issue queries on your behalf
DataThe query body + your GitHub OAuth token (held by us in Google Secret Manager, never in plain Postgres)
LocationUnited States
ProcessorNotion (when connected)
PurposeWorkspace search + page reads on your behalf
DataThe query body + your Notion OAuth token (same handling)
LocationUnited States
ProcessorCustom MCP servers
PurposeWhatever the org admin registered them for
DataThe query body + the credential the admin supplied
LocationWherever the admin pointed us

You can disconnect any integration from Settings → MCP integrations; the credential is purged from Google Secret Manager when you do.

For transfers to processors outside the EU/EEA, we rely on EU Standard Contractual Clauses (SCCs) or the EU-US Data Privacy Framework, as applicable. We do not sell your personal data to any third party.

6. International Data Transfers

Our primary infrastructure is hosted on Google Cloud Platform in the EU (region: europe-west1, Belgium). However, some sub-processors (OpenRouter, Stripe) are based in the United States. For such transfers, we ensure appropriate safeguards are in place:

  • EU Standard Contractual Clauses (SCCs) pursuant to Commission Decision 2021/914
  • EU-US Data Privacy Framework certification where applicable
  • Supplementary technical measures including encryption in transit and at rest

7. Data Retention

We retain your data for the following periods:

  • Account data: For the duration of your account.
  • Organization deletion: Deleting an organization soft-deletes it and starts a 7-day recovery window during which it can be restored. After the window, a purge job erases the organization's surface-vault secrets, memory, messages, and chat sessions, then the organization record itself - permanently and irreversibly.
  • Chat and conversation data: For the duration of your account. Deleted when you delete a conversation or your account.
  • Memory data: Until you delete individual memories or your account. Deletion is immediate and permanent. An admin can also undo a single import - this deletes exactly that import's memories within the organization.
  • Member removal: When a member is removed from a team organization, their Private memories in that organization are hard-deleted, their API keys and MCP connection tokens are revoked, and their Slack/Teams identity is unlinked - atomically. Memory they authored at Project or Org visibility remains with the team.
  • Usage and analytics data: Aggregated and anonymised after 12 months. Raw logs retained for up to 90 days.
  • Billing records: Retained for up to 10 years as required by applicable commercial and tax law.
  • Backups and technical logs: Backups containing personal data are deleted in the ordinary course of business, no later than approximately 90 days. Technical logs are automatically purged after 90 days.

8. Cookies and Tracking

We use the following types of cookies:

  • Strictly necessary cookies: Required for authentication, session management, and security. These do not require consent under GDPR.
  • Functional cookies: Store your preferences such as theme selection and language settings.
  • Analytics cookies: Help us understand how you use the platform. Only set with your consent.

We do not use third-party advertising cookies or cross-site tracking.

Analytics specifics. We load Google Analytics in storage: ‘none’ mode, which means GA does not write cookies and instead generates a per-page-view client identifier that is discarded when the tab closes. We can see aggregate page views and referrers; we cannot link visits back to a specific browser. Because no cookies are written for analytics, no consent banner is required under GDPR for this flow. Browser-level Do Not Track or any standard ad-blocker disables it cleanly.

Session replay. We use PostHog session replay to understand how the product is used and find usability problems. Recordings mask all form inputs (passwords, emails, payment info, free-text fields) and exclude all auth, login, signup, and billing pages entirely. Recordings are retained for 30 days, then deleted. Hosted in the EU.

You can manage all cookie preferences at any time through your browser settings.

9. Your Rights Under GDPR

As a data subject under the GDPR, you have the following rights:

  • Right of access (Art. 15): Request a copy of your personal data and information about how it is processed.
  • Right to rectification (Art. 16): Request correction of inaccurate or incomplete data.
  • Right to erasure (Art. 17): Request deletion of your personal data (“right to be forgotten”).
  • Right to restriction (Art. 18): Request that we restrict processing of your data in certain circumstances.
  • Right to data portability (Art. 20): Receive your data in a structured, machine-readable JSON format - available in-app and, for organization data, via the admin GDPR export route. An organization export excludes other members' Private memory. You can also delete your own Private data.
  • Right to object (Art. 21): Object to processing based on legitimate interests, including profiling.
  • Right to withdraw consent (Art. 7(3)): Withdraw consent at any time where processing is based on consent.
  • Right to lodge a complaint: File a complaint with a supervisory authority. The competent authority is the relevant data protection supervisory authority in your jurisdiction.

To exercise any of these rights, contact us at contact@switchy.build. We will respond within 30 days as required by GDPR.

10. Your Rights Under CCPA (California Residents)

If you are a California resident, you have additional rights under the CCPA:

  • Right to know: Request disclosure of the categories and specific pieces of personal information we have collected about you.
  • Right to delete: Request deletion of your personal information, subject to certain exceptions.
  • Right to opt-out of sale: We do not sell personal information. No opt-out is necessary.
  • Right to non-discrimination: We will not discriminate against you for exercising your CCPA rights.

In the preceding 12 months, we have collected the following categories of personal information: identifiers, commercial information (billing), internet activity information, and inferences drawn from the above. We have not sold any personal information.

11. Data Security

We implement appropriate technical and organisational measures to protect your data, in accordance with GDPR Art. 32:

  • Encryption in transit (TLS 1.2+) and at rest (AES-256, managed by Google Cloud across Secret Manager and disk)
  • Connector credentials and OAuth tokens stored in Google Secret Manager - the database holds only an opaque reference, each reference is bound to its owning organization, and the credential is injected only at the moment of the outbound call, never in the database, logs, or model context
  • Side-effecting actions on connected tools require explicit human approval before they run
  • Secure authentication with OAuth 2.0 and session tokens
  • API keys stored using irreversible cryptographic hashing
  • Memory, connections, secrets, and messages scoped per organization at the SQL query layer; a cross-user / cross-tenant leak test suite runs as a required CI merge gate
  • Infrastructure hosted on Google Cloud Platform (which holds ISO 27001 and SOC 2 Type II certification)
  • Role-based access control and principle of least privilege for internal access
  • Regular security reviews and dependency audits

12. Automated Decision-Making

We do not use your personal data for automated decision-making or profiling that produces legal or similarly significant effects on you within the meaning of GDPR Art. 22. AI model responses are generated based on your prompts and are not used to make decisions about your access to services or pricing.

13. Children's Privacy

Switchy is not directed at children under the age of 16 (or the applicable age of digital consent in your jurisdiction). We do not knowingly collect personal data from children. If you believe a child has provided us with personal data, please contact us at contact@switchy.build and we will promptly delete it.

14. Changes to This Policy

We may update this privacy policy to reflect changes in our practices, technology, legal requirements, or other factors. We will notify you of material changes by email or by posting a prominent notice on our website at least 30 days before the changes take effect. The “Last updated” date at the top indicates the most recent revision.

15. Contact and Complaints

For any privacy-related questions, data subject requests, or complaints:

Email: contact@switchy.build

General inquiries: contact@switchy.build

If you are not satisfied with our response, you have the right to lodge a complaint with your local data protection supervisory authority.