Last Revised: October 8, 2026
This list describes third-party services that process personal data for Timeglass, including providers available for AI fallback routing. The data a provider receives depends on the features enabled and the route used. Marketing and advertising services may have different data protection roles under their applicable terms.
Sub-Processor List
| Service Provider | Purpose | Data Types | Legal Basis |
|---|---|---|---|
| Customer.io | Email notifications, activity digests and customer engagement | Email, user identifiers, usage events and activity-summary content | Contract (Legitimate Interest) |
| Google Analytics | Website analytics and usage tracking | Browsing behavior, device info, IP address | Legitimate Interest |
| Google Ads | Website ad conversion measurement and campaign attribution | Browsing behavior, device info, IP address, campaign identifiers | Consent |
| PostHog | Product analytics and session recording | Product usage events, session recordings, user and workspace identifiers, email, device info, IP address | Legitimate Interest |
| Sentry (Functional Software, Inc.) | Desktop application error monitoring and crash diagnostics | Error events, stack traces, application and device metadata, and diagnostic context that may contain personal data | Legitimate Interest |
| Meta Pixel | Website ad conversion measurement and campaign attribution | Browsing behavior, device info, IP address, campaign identifiers | Consent |
| LinkedIn Insight Tag | Website ad conversion measurement and campaign attribution | Browsing behavior, device info, IP address, campaign identifiers | Consent |
| Reddit Pixel | Website ad conversion measurement and campaign attribution | Browsing behavior, device info, IP address, campaign identifiers | Consent |
| beehiiv Ad Network Pixel | Newsletter ad conversion measurement and campaign attribution | Browsing behavior, device info, IP address, campaign identifiers | Consent |
| HubSpot | Sales and customer CRM, support and email communication records, and consent-controlled website tracking | Company and contact info, user/workspace identifiers, trial and subscription metadata, aggregate usage totals, support messages, email content, browsing activity, cookie identifiers, IP address | Legitimate Interest / Consent |
| Clerk | Authentication and user identity management | Email, user credentials, authentication data | Contract |
| Supabase | Database and backend infrastructure | All application data | Contract |
| RB2B | B2B visitor identification and lead generation | Business contact information, company data | Legitimate Interest (Opt-in for EU) |
| Apollo | Sales intelligence, prospecting and email-based profile enrichment | Business contact information, company data | Legitimate Interest (Opt-in for EU) |
| Resend | Transactional email delivery and inbound assistant email processing | Sender and recipient details, notification content, inbound messages and assistant context | Contract |
| Google Cloud Platform | Cloud hosting, storage, queues and backend processing | Screenshots, activity data, meeting audio and derived reports | Contract |
| Google Gemini Developer API | AI text and audio processing, report questions, avatar validation and embeddings | Activity and report context, meeting content, avatar images and embedding inputs | Contract |
| OpenRouter | AI inference routing and TypeSafe decision-model access | Screenshot images, activity text, meeting audio and transcripts | Contract |
| Microsoft Azure | AI inference host via OpenRouter: meeting transcription with speaker separation, identifying meeting speakers, and meeting reports | Meeting audio, meeting transcripts, participant names and work context | Contract |
| Morph (AutoInfra, Inc.) | Legacy AI inference host via OpenRouter; excluded from current host pools | Screenshot images, activity text | Contract |
| Parasail | AI inference host via OpenRouter | Screenshot images, activity text, meeting transcripts | Contract |
| Fireworks AI | AI inference host via OpenRouter; fallback for text analysis and meeting reports | Activity text, meeting transcripts | Contract |
| DeepInfra | AI inference host via OpenRouter; fallback for text analysis, identifying meeting speakers, and meeting transcription | Activity text, meeting transcripts, meeting audio when the primary transcription model rejects it | Contract |
| Venice | AI inference host via OpenRouter; fallback for text analysis and identifying meeting speakers | Activity text, meeting transcripts | Contract |
| CoreWeave / Weights & Biases | AI inference host via OpenRouter | Screenshot images, activity text | Contract |
| OpenAI | Optional workspace AI assistant | Chat messages, workspace context | Contract |
| Groq | Legacy AI text inference; not in current active backend use | Query text | Contract |
| Stripe | Payment processing | Payment information, transaction data | Contract |
| Plain | Customer support chat and conversation management | Customer identity, support messages and attachments | Contract |
| Daytona | Hosted sandboxes for the optional workspace AI assistant | Chat messages, workspace context and authenticated service access | Contract |
| MailerLite | Newsletter, subscription and referral email workflows | Subscriber contact details, mailing-list membership and engagement data | Contract |
| Calendly | Demo and onboarding meeting scheduling | Contact details, booking responses and scheduling metadata | Contract |
| TypeSafe AI | Activity segmentation and project, client and category decisions via OpenRouter | Activity evidence, workspace classification context and decision results | Contract |
| Novita AI | Screenshot and activity inference hosting via OpenRouter | Screenshot images, activity text and workspace context | Contract |
| OpenInference | Permitted screenshot and activity inference host via OpenRouter | Screenshot images, activity text and workspace context | Contract |
| io.net | Permitted text inference host via OpenRouter | Activity text, meeting transcripts and workspace context | Contract |
Application Delivery and Website Icons
The desktop application uses GitHub to check for updates, retrieve release notes and download installers. These requests expose the device’s IP address and technical request metadata, such as the requested release, platform and application client information, to GitHub and its delivery infrastructure. This delivery flow does not upload screenshots, meeting recordings or workspace content. GitHub’s handling of its service-access data is described in its privacy statement.
To display icons for websites saved in desktop capture filters, the application requests icons from those websites and may follow redirects or load icons from hosts specified by those sites. The receiving hosts see the device’s IP address and technical request metadata, including the Timeglass client name. Paths and query strings from saved filter rules are omitted. Icon results are cached on the device for seven days. These websites are not Timeglass service providers; their own privacy practices apply to those requests.
AI Routing and Retention
Requests routed through OpenRouter are processed by OpenRouter and the inference provider selected for that request. Current chat-model and meeting-transcription routes request zero-data-retention endpoints and prohibit data collection. These controls apply to the configured endpoints; they do not mean that account or billing metadata is never retained. TypeSafe decision-model requests use a separate API whose retention terms must be considered independently. Direct Google and OpenAI services follow their applicable service terms and account settings.
Some AI providers retain submitted content after processing for abuse monitoring, security or legal obligations under their applicable terms. Google documents 55-day retention of prompts, contextual information and responses in the paid Gemini Developer API for abuse monitoring. This retention is separate from optional developer logging, and disabling that logging does not make the service zero retention. See Google’s abuse-monitoring policy. Timeglass’s zero-day screenshot deletion setting applies to its own storage; it does not determine retention by providers that receive content.
Voice Signatures
To tell meeting speakers apart, Timeglass computes a numerical voice signature for each speaker in each minute of recorded meeting audio, and, from the microphone of each Timeglass user recording the meeting, a voice signature of that user. Voice signatures are computed on Timeglass's own infrastructure, are not sent to any sub-processor, are used only to link the same speaker across the minutes of that meeting and to recognise which speakers are the Timeglass users recording it, and are deleted as soon as the meeting has been processed. They are never used to identify people across meetings.
Data Transfers
Most of our sub-processors are located in the United States. For processors outside the EU/EEA, Timeglass relies on Standard Contractual Clauses (SCCs), adequacy decisions, or explicit opt-in consent where applicable to ensure adequate safeguards for personal data transfers.
Changes to Sub-Processors
Timeglass may add or change sub-processors from time to time. We will provide notice of material changes to this list at least 30 days in advance when possible. You may object to the use of new sub-processors by contacting us at privacy@timeglass.ai.
Sub-Processor Agreements
Timeglass requires applicable processor agreements to address the following obligations:
- Process personal data only on our instructions
- Implement appropriate technical and organizational security measures
- Assist with data subject rights requests
- Assist with compliance obligations
- Delete or return personal data upon request
- Permit audits and inspections
Contact & Inquiries
For questions about our sub-processors, data transfers, or to request changes to this list, please contact our Privacy Team at privacy@timeglass.ai.