Automatic capture
Build a continuous Mac timeline without relying on start-and-stop timer memory.
drifty automatically captures your Mac activity and turns it into useful focus and drift patterns. The complete timeline stays on this Mac by default, while Local AI, remote AI, and Device Sync remain separate choices.
A privacy policy can explain how uploaded data is handled. The more practical question is whether the source record needed to leave your Mac at all. drifty keeps automatic capture local first, then makes every remote path visible and optional.
“Cloud” is not one switch in drifty. Capture, classification, sync, analytics, and public sharing have separate boundaries.
| Surface | Default location | Leaves your Mac when | Your control |
|---|---|---|---|
| Complete activity timeline | Local SQLite database on this Mac | You explicitly enable the paid Device Sync Preview | Enable per Mac; pause or purge the private remote copy in the app |
| Local AI | Inference runs on this Mac | Never for the classification request | Choose Local AI as the classification route |
| drifty Cloud | Local timeline remains the source | You choose Cloud classification; selected, minimized fields go to the hosted provider route | Change route and field-level AI privacy settings |
| OpenRouter (your key) | Local timeline remains the source | You choose your OpenRouter account for classification | Your key, model, quota, and route choice |
| Custom endpoint | Local timeline remains the source | You configure and choose that endpoint | You select the endpoint and its operator |
| Product analytics | On by default; raw timeline is excluded | Limited named events and bounded aggregates are sent while analytics is enabled | Disable future product analytics in Settings → Privacy |
| Profile and leaderboard | Private until you opt in | You publish selected profile fields and aggregate focus totals | Choose what to publish or keep the surfaces private |
These products optimize for different jobs. This compares capture, interpretation, and the documented default data boundary—not vague “privacy-first” badges.
| Tool | Capture and interpretation | Documented data boundary | Best fit |
|---|---|---|---|
| drifty | Automatic Mac capture with context-aware AI classification | Complete timeline local by default; Local AI sends no classification request; remote AI and sync are explicit choices | Smart personal review with a selectable processing boundary |
| Timing | Automatic app, document, and domain capture with project review and optional AI summaries | Local database by default; Timing Sync merges tracking data into its sync account, and optional AI summaries process some Mac-usage data with OpenAI | Mature Mac time accounting and billing workflows |
| ActivityWatch | Automatic local capture with user-authored regular-expression category rules | Its privacy documentation says usage data is stored locally and is not transmitted to external servers | Open-source local history and self-managed categorization |
| Rize | Automatic capture with AI client, project, and task tagging | Its policy defines tracked activity as User Data and retains activity data while subscribed; optional Screen Text can send an image to Vertex AI in Cloud mode, while Rize says it does not store that image | Cloud-backed project attribution and team reporting |
Sources checked September 13, 2026: drifty data boundaries; Timing FAQ, Sync, and AI summaries; ActivityWatch privacy and categorization; Rize privacy policy and automatic tracking.
Local-first does not have to mean raw logs with no interpretation.
Build a continuous Mac timeline without relying on start-and-stop timer memory.
Turn activity into focus and drift patterns instead of stopping at an unstructured event archive.
Keep inference local or deliberately choose the remote route whose convenience and tradeoffs you accept.
Not by default. The complete app, site, window, and session timeline stays in the local Mac database. If you explicitly enable the paid Device Sync Preview, raw activity segments are copied into private, account-scoped sync storage so your own Macs can merge them.
With Local AI, no classification request leaves your Mac. The model download and memory use depend on the model, while the complete activity timeline remains the local source of truth.
The chosen provider receives selected, minimized fields for that classification request, such as app name, domain, duration, and optional privacy-controlled context. drifty excludes a replayable raw timeline, document or message bodies, screenshots, passwords, secrets, and provider keys.
No internet-connected software should promise absolute security. drifty makes a narrower, verifiable promise: the complete timeline stays local by default, Local AI sends no classification request, and remote AI or device sync require an explicit choice.
No. drifty analyzes activity information needed for time review, but it is not a continuous screen recording app.
Public leaderboard and profile surfaces are optional and use aggregate focus totals, not raw app, site, window, or session history.
Get automatic capture and meaningful AI classification without making the cloud your only option.