built for control.
secured carefully.

Local-first source model

the dripos mac app can read local Mac data such as iMessage only after you grant the required macOS permissions. local source data is indexed on your machine so the app can search, understand context, and prepare messages without treating our cloud as the primary copy of your conversation history. App keys, provider keys, and local bridge secrets are stored in macOS Keychain where applicable.

cloud services store or process feature-dependent data needed to run enabled features, such as:

• account identity, workspace name, member roles, and sign-in state
• plan, entitlement, usage, and billing records handled with Stripe
• OAuth tokens and connected-account metadata for hosted integrations, encrypted at rest
• hosted integration records such as Slack, Gmail, Calendar, Stripe, or lead-source data when enabled
• selected context sent to hosted dripos AI, if you use the hosted AI path
• provider, model, operation, and token counts for BYOK or local LLM usage accounting
• support data, error reports, and diagnostic telemetry needed to operate and secure the product

Source and automation boundaries

• iMessage. iMessage access is local to the Mac app and depends on macOS permissions. Message sending uses the permissions and automation available on that Mac, and supported iMessage group chats can be read and replied to through the Mac app.
• LinkedIn. LinkedIn messaging requires a connected user account and configured limits, people, templates, and rules before automation runs.
• Group chats. dripos can work with supported group conversations such as iMessage groups, LinkedIn group threads, and Slack channels or threads when the source is connected and the rules allow it.
• Gmail, Slack, Calendar, Stripe, and meeting tools. these sources can provide context, scheduling data, billing signals, and account signals. The exact data processed depends on the integrations and features you enable.
• Booking. dripos can send scheduling links or create calendar bookings when the connected account, calendar permissions, and automation rules allow it.

What may leave your machine

raw local files such as chat.db are not uploaded as source files. Selected context may leave your machine only when the feature path you choose needs it, such as a hosted AI request, BYOK cloud provider request, cloud automation step, connected-source sync, or support request.

• Hosted dripos AI. selected messages, instructions, and context are sent to the dripos backend and the configured hosted AI provider for that request.
• BYOK cloud providers. selected context is sent from the Mac app to the Anthropic or OpenAI provider configured with your key. dripos records provider, model, operation, and token counts for entitlement and usage accounting.
• Local LLM / Ollama. selected context is sent to the local Ollama base URL. dripos checks entitlement and may record provider, model, operation, and token counts, but not prompt text, for local LLM usage accounting.
• Cloud automation and integrations. account metadata, OAuth tokens, scheduling context, hosted integration records, and support logs may be processed when those cloud-backed features are enabled.

User controls

• choose which tools are connected
• choose which people, channels, limits, templates, and rules automation can use
• choose hosted dripos AI, BYOK cloud providers, or local LLM where your plan allows it
• pause automation at any time
• revoke OAuth access from the connected provider or from dripos settings
• remove the local index by deleting the app data and local database from the Mac

Encryption

In transit

requests between the Mac app, the dashboard, and our backend use TLS 1.2 or higher.

At rest

managed databases and cloud storage use encryption at rest. Sensitive tokens and secrets are additionally protected with application-level encryption where configured.

Access controls and monitoring

• production access is limited to authorized personnel
• access is granted by role and reviewed as part of our security program
• administrative and production actions are logged where the platform supports it
• security controls are monitored through Vanta as part of our compliance program

No shared model training

dripos does not train a shared AI model on your messages. When an AI feature is used, selected context is sent only as needed to the configured provider, which may be hosted dripos AI, a customer-managed provider key, or a local Ollama model depending on your plan and settings.

Compliance and trust center

• SOC 2. SOC 2 is in progress through Vanta. dripos is not SOC 2 certified until the audit report is complete.
• Vanta Trust Center. trust center access can be shared during security review. Email [email protected] to request access.
• GDPR and CCPA. privacy access, export, correction, and deletion requests can be sent to [email protected].
• dripos is not intended to process protected health information and is not HIPAA covered.

Incident response

we monitor service health and security signals. Confirmed incidents are triaged through our incident process, and impacted customers are notified as required by law and contract.

Responsible disclosure

if you find a vulnerability, send a detailed report to [email protected]. include reproduction steps, affected endpoints, and your contact info. we commit to:

• acknowledging your report within 2 business days
• providing a timeline for triage within 5 business days
• not pursuing legal action against good-faith research that respects user privacy
• publicly crediting you, with your permission, once a fix ships

do not access customer data, run automated scans against production, or test denial-of-service scenarios. we will work with you on safer reproduction paths.

Contact

security and disclosure: [email protected]. privacy requests: [email protected].