← Back to projects

PAM / Personal Asset & Life Manager

One life.
Many connected parts.

A private, self-hosted home for the information I use every day—organized around me.

My role
Product direction & design
Development
Built with AI coding assistance
Status
Working app · Still evolving

The starting point

Life doesn’t fit neatly
into separate apps.

A reminder can be about an object I own. A place can be connected to a trip, a photo, or a cost. I wanted those relationships to remain visible instead of being scattered across separate lists.

PAM is my attempt to put them in one place. It is designed for one person: me. The measure of success is whether it becomes useful enough that I turn to it every day.

Inside PAM

Three views of everyday life.

The home view brings the day together. Dedicated views make it easier to work through finances or retrace a trip.

01 / Home

Keep the day in view.

Calendar, notes, tasks and a financial summary live together in a layout I can rearrange. On my phone, these views become a swipeable deck.

The body scan is a navigation choice: selecting an area opens related belongings or categories. It is one way into the information, alongside the everyday tools.

For me: fewer separate places to check before getting on with the day.

How I’m building it

I define the system.
AI helps implement it.

I built the personal server that hosts PAM, and I’m learning how to turn a personal need into a working product. I choose what belongs in PAM, define how the pieces should relate, and refine the experience as I use it. Claude handles the coding from that direction.

That distinction matters to me: this project represents my product decisions and ongoing learning, with AI-assisted implementation.

A closer look at the setup

PAM is the app’s name. INDI is the repository name and the hostname of the tower that runs it.

The current implementation uses Next.js and React, with Prisma over SQLite. It runs as a service on the Ubuntu server I built and is accessible remotely through Tailscale. The personal dashboard is not exposed as a public website.

Its progressive web app supports offline note and task edits. A scheduled server process handles reminders. The next stage introduces local inference through llama.cpp.

What comes next

An assistant with context.

The local assistant is the next stage, not a finished feature. The scaffolding is in place; a GPU upgrade will support running the model on my own tower.

I want the assistant to help me understand and navigate my life using the information already in PAM. Some of its most useful jobs will only become clear once I’m living with it.

This project overview reflects the current build description. PAM itself stays private, so there is no public dashboard or live personal-data demo here.