DocsIntegrationsActivity Analysis

Activity Analysis

Nareli captures and analyzes your desktop activity to automatically suggest time entries, helping you account for every productive minute of your day.

Overview

Activity analysis is Nareli's desktop-aware intelligence layer. It observes which applications you use and generates time entry suggestions to fill gaps in your manual tracking.

Captures application names and window titles at regular intervals
AI groups related activities into coherent work sessions
Suggests time entries only for untracked periods
Runs entirely on your local machine via Ollama

Activity tracking requires the Pro plan (€7/mo) or higher. Advanced AI learnings and fingerprints require the Business plan. See Subscription & Billing for plan details.

What Gets Tracked

The capture process records lightweight metadata about your active application - no screenshots, keystrokes, or screen content.

Application name (e.g., VS Code, Chrome, Figma, Terminal)
Window title (e.g., file path, page title, tab name)
Timestamp of each activity record
Focus duration per window

Activity capture requires accessibility permissions on macOS. Nareli prompts you to grant these on first launch.

Activity Analysis Pipeline

The processor runs every 10 minutes (configurable), collecting unprocessed records and sending them to the local Ollama model.

Capture - Records application metadata continuously while Nareli is open

Queue - Accumulates records until the next analysis interval

Classify - AI groups activities into work sessions using your project/task context

Suggest - Generates timeentry_create suggestions for untracked periods

Review - Suggestions appear in the day view and Suggestions page for accept/decline

The processor skips runs when there are no new records and backs off during high system load.

Manual vs AI-Suggested Entries

Activity suggestions complement your manual tracking rather than replacing it.

	Manual Entry	AI-Suggested Entry
Source	Created by you	Generated from desktop activity data
Accuracy	Exact - you define the details	Inferred - review before accepting
Coverage	Depends on your tracking discipline	Catches gaps you may have missed
Description	Written by you	Auto-generated from app/window context
Overlap	N/A	Overlap detection flags conflicts with existing entries
Workflow	Create directly in day view	Accept/decline from suggestion block

Smart Classification

The AI understands work context and groups related activities intelligently.

Sequential activities in the same project are merged into a single session
Brief interruptions (e.g., 2 min of Slack mid-coding) don't split sessions
Descriptive task and project names improve matching accuracy
Short activity bursts (email + browsing) are filtered out as non-substantive
Suggestions are conservative - durations are rounded to reasonable increments

Configuration

Enable, disable, or tune activity analysis independently of other Nareli features in Settings > Services.

Setting	Default	Description
Analysis interval	10 minutes	How often the AI processes accumulated records
Activity capture	Enabled	Toggle desktop activity recording on/off
App exclusions	None	Prevent specific apps from being captured (for sensitive work)

The processor status is visible in the system status indicator - see whether it is running, idle, or queued.

Privacy & Local Data

All activity data is stored and processed exclusively on your machine.

Raw records, classifications, and suggestions stored in your local database
AI runs through local Ollama - no cloud AI calls
View, delete, or bulk-clear activity records anytime from the dashboard
No screenshots, keystrokes, clipboard, or mouse movements are captured
Uninstalling Nareli removes all activity data