Rize's loop is small enough for one sitting: sample the frontmost app, window title and browser URL every few seconds, merge the samples into blocks, sort them with rules and show a day and a week. An agent can build that as a menu-bar app with a local report page and no outside service. It takes a few macOS permission prompts and, for browser URLs, one AppleScript permission per browser. What you give up is the Windows app, AI that suggests clients and projects for you, client reports and integrations, and team dashboards. ActivityWatch already covers the tracking part for free if you'd rather not build.
Before you buildRize has no free plan, only a 7-day trial. ActivityWatch is a free, open-source tracker that logs apps and sites; try it first to see whether automatic tracking suits you.
Build me a macOS time tracker that records my work automatically, to replace Rize. Requirements: - Python 3.12 with `rumps` for the menu-bar icon, `pyobjc` (AppKit, Quartz, ApplicationServices) for window data and FastAPI for a report page. `python tracker.py` starts both; the page is at http://127.0.0.1:4950 and listens on localhost only. - Sampling: every 5 seconds record the frontmost app (NSWorkspace), its window title (Accessibility API) and, for Safari, Chrome, Arc and Brave, the current tab URL through AppleScript. 3 minutes without input (CGEventSourceSecondsSinceLastEventType), sleep or screen lock closes the open block. - Storage: sqlite3 at ~/TimeTracker/time.db; consecutive identical samples merge into blocks (start, end, app, title, url, category). Store nothing else: no screen contents and no keystrokes. - Categories: rules in ~/TimeTracker/rules.yaml (app, title regex or domain > category, client, project), first match wins. Unmatched blocks go to a "Needs a category" inbox with a button that turns a block into a new rule. Optional LLM suggestions for that inbox through an OpenAI-compatible endpoint in .env, sending titles and domains only, off by default. - Calendar: an optional .ics path in .env, so meeting time counts as meetings even when another window is in front. - Report page: Today and Week views with a timeline, hours by category, client and project, the longest unbroken work stretch, and a table where I can split, merge and relabel blocks; CSV export for a date range. - Menu bar: current category, time tracked today, pause and resume, and "I was away" to relabel the last idle stretch. - Single user. No accounts, no telemetry, local-first. `python tracker.py --install` sets up a LaunchAgent. - Out of scope: Windows and Linux, team dashboards, screenshots, invoicing, a mobile app. - README: granting Accessibility and Automation permissions in System Settings, why URL capture asks once per browser, and a reminder that the database holds window titles, so keep it out of cloud-synced folders.
Opens your agent with the prompt filled in, you press enter. Or copy it and paste it into an empty folder.
The first prompt builds the core. Run these one at a time in the same folder to grow it into your own Rize.
Add a daily summary to the time tracker in this folder. Requirements: - At 18:00 (configurable), build a markdown note for the day: hours by category and client, the three longest blocks, and a count of context switches per hour. - Optionally send only block titles and domains to the OpenAI-compatible endpoint in .env for a three-sentence recap; off by default, and the note says whether an LLM was used. - Write it to NOTES_DIR/YYYY-MM-DD.md, appending under a "Time" heading if the file exists. - Add a test with a fixed day of sample blocks. - Update the README.
Add focus nudges to the time tracker in this folder. Requirements: - In rules.yaml, mark categories as "distracting" and set a limit in minutes per hour. - When the running total in the last 60 minutes passes the limit, show one macOS notification (osascript display notification), then stay quiet for 30 minutes. - A "Focus session" menu item blocks nudges and writes the session to the database with a label. - Show nudges sent and focus sessions on the Today view. - Update the README.
Add invoice export to the time tracker in this folder. Requirements: - Define rates per client in rules.yaml (hourly rate, rounding to 1, 6 or 15 minutes, minimum block length). - An Invoice page picks a client and a month and lists billable blocks by day with rounded hours, plus a total. - Let me exclude blocks, and keep exclusions in SQLite so a re-export matches. - Export to CSV and a plain HTML page that prints cleanly. - Update the README.
Rize does the boring part without being asked: you never start a timer, and a week later the hours are already grouped by client. People pay to stop remembering, and agencies pay for the team and profitability views on top.
Estimate · Free: it runs on your Mac. The optional LLM suggestions cost a few cents of API usage a month if you switch them on. The build is token usage on your agent plan.
| Plan | Monthly | Annual, per month | What you get |
|---|---|---|---|
| Basic | $12.99 | $9.99 | One user: automatic app and website tracking, AI categorisation, personal reports, data exports and 500 AI credits a month. |
| Pro | $28.99 | $23.99 | Everything in Basic plus client and scheduled reports (PDF, CSV), AI automations and agents, MCP server access, API, webhooks and Zapier, and 1,000 AI credits a month. |
| Max | $49.99 | $39.99 | Everything in Pro with 3,000 AI credits a month and priority support. |
| Team | $39.99 | $29.99 | Two seats minimum; shared dashboards, admin settings, budgeting, client and project profitability and 2,000 AI credits per seat a month. |
| Enterprise | custom | — | SSO and SCIM, a mobile app, custom contracts and SLA, a dedicated account manager. |
No free plan. A 7-day trial gives full access without a card.
Yes. An AI coding agent such as Claude Code, Codex or Cursor can build a usable personal version in one session with the prompt on this page, running on your own machine.
About $12.99 a month on the Basic plan, $155.88 a year (checked 2026-10-19).
A Windows app and a mobile app (the mobile app is Enterprise only); AI that suggests clients and projects for each block, plus AI reports and agents; Client and scheduled PDF reports, and links to QuickBooks, Asana, ClickUp, Linear and Slack; Team dashboards and project profitability; Focus music and smart breaks. If any of those matter to you, keep paying.
Yes: ActivityWatch (open source), Kimai (open source), solidtime (open source).