Personal systems
A personal assistant for founders and managers
Weekly reflection, goals, and a patterns engine that watches my calendar, tasks, and CRM for the behaviors I keep repeating, then nudges me at the moment they are about to happen again.
I built this demo as one self-contained page: no video, no backend. It shows one part of the app, the patterns engine. Names, tasks and numbers are fictional. The “someone is waiting on it” split and the one-click message are a concept, not built yet; the rest is the real flow.
Best viewed on a laptop: Open the full demo
Context
Sixty-four weeks of manual weekly tracking in a spreadsheet produced data but no insight. Reading your own journal and hoping you notice a pattern does not work. I wanted something that reads it for me and speaks up at the right time.
What I built
- A local-first Hebrew web app on top of a private Google Sheet
- Weekly and monthly reflection modes, an AI coach with persistent memory, a four-layer goal map
- A patterns engine that mines real telemetry: tasks postponed, meetings moved, CRM follow-ups missed
- A pattern only proposed with three or more pieces of evidence, one active at a time
- Every proposal needs my approval; rejected ones are never re-proposed
How it runs
Two GitHub Actions run even when my laptop is off: a daily job that reads my calendar and DMs me a Slack cue 15 minutes before a trigger-related meeting, and a weekly encrypted backup. Every connector degrades gracefully: a missing token disables one source and the app keeps running.
What it taught me
The same discipline that makes a sales machine work makes a personal one work: instrument the behavior, wait for evidence, act at the trigger moment instead of at review time.
Stack
- Python / Flask
- Google Sheets as DB
- Google Calendar
- Todoist
- HubSpot
- Slack
- OpenAI
- GitHub Actions
Want the walkthrough?
The code is private. The walkthrough isn't.
Book 30 minoranov.con@gmail.com