Privacy

Data we use

last30days.dev processes account, billing, research, and delivery information needed to provide the service. If you connect Google Calendar, we request read-only event access and store only the event fields needed to decide whether to prepare and deliver your own pre-meeting brief.

Google Calendar data

Calendar data is used only to provide meeting briefs to the user who connected that calendar. We do not use it for advertising, sell it, or use it to train general-purpose AI models. Event descriptions and locations are not sent to the meeting-identity providers described below.

Our use and transfer of information received from Google APIs follows the Google API Services User Data Policy, including its Limited Use requirements.

Meeting-brief subprocessors

We use Parallel for bounded public-web lookup when an external invitee cannot be identified from the invite alone. Parallel receives only the unresolved external invitee evidence needed for that lookup: their email address, available name, company domain, and one short attendee-specific meeting-title segment. Owner, internal, and resource attendees and unrelated title segments are excluded.

We use Anthropic to evaluate minimized invite evidence and bounded public-search excerpts. Anthropic receives a bounded meeting title and owner aliases for exclusion, plus external attendee slots, available names and domains, organizer flags, and unresolved address evidence. Provider results are temporary and are not stored with the meeting. Only a locally validated person name and safe provenance may become part of the research target.

Retention and control

Synced event rows are removed on a rolling basis seven days after an event ends. Disconnecting Google Calendar removes the stored connection secret and synced event rows and stops pending briefs; completed research follows the service's existing usage and retention rules. You can disconnect from Integrations.

Security

Google refresh tokens are encrypted in managed secret storage. Meeting-target fingerprints use a dedicated keyed digest. Raw invitee addresses, public-search excerpts, URLs, and provider responses are excluded from meeting-target columns, owner-visible failure events, and operational counters.