Back to docs

Walkthrough

Every screenshot and clip below is the real app, running against a seeded demo session — not redrawn mockups. This is exactly what plan → track → analyze → ask looks like end to end, including one real, live exchange with the chat.

01

Plan

It starts on the planner’s timeline — drag across the days you mean to work on something.

Drag to create

A drag across a few days opens a quick-create popover — title, category, done.

The plan, opened
An open plan item panel showing title, category, dates, and a High priority set

Click it to open the full panel — category, dates, and notes, all editable in place. Set a priority and it shows up as a small flag right on the timeline bar.

02

Track

When the day comes, start it right from the plan — no re-typing what you already decided.

Started, linked

The timer runs right inside the plan panel, linked back to it from the moment it starts — a real, live timer, pause and resume whenever.

03

Close the loop

The same timer shows up on the Recorder dashboard too — it’s one piece of state, not two apps pretending to sync.

log(ger) — Spring 2026
The dashboard showing the same active timer, plus the current streak

Same timer, same second, on the dashboard — note the streak, already at 52 days from the seeded history.

Stop Timer
The stop dialog, showing it is fulfilling a plan, with a Mark plan complete checkbox

Stopping it shows exactly which plan it’s closing out — completing the plan is still your call, not automatic.

log(ger) — Spring 2026
The dashboard after stopping, with today's total, entry, and a longer streak

Today’s total, the entry, and the streak all update together — one write, everywhere it matters.

04

Analyze

Eight weeks of seeded history, rolled up into the real dashboard — the same charts documented in Analyzing.

Analytics
The analytics page showing a stacked area chart, category breakdown, and a real activity heatmap

80h 37m across 53 days — the stacked chart, the category split, and a real activity heatmap, all from data that actually got logged.

05

Ask

One real exchange with the chat — not a staged answer. Every tool call it makes against your data is visible, and nothing runs before you approve it.

Send this to Claude?
The approval gate before Claude is allowed to query the database

Asking “How much time have I spent on Research this month?” stops here first — eight read-only tools, nothing runs until you approve.

Claude Sonnet 5
Claude's real answer, grounded in the actual seeded data, with the tool calls it made shown below

The real answer — grounded in the actual numbers, with every query it ran shown right below.

That’s the whole loop. Run it locally to try it on your own time.