Automations running on real products

Automations that do the work while you sleep.

I build AI automations on top of tools you already use: GitHub, Claude and your own machine. Each one started as something I needed for my own products, and each one runs every day.

See the products Tell me what to automate

~/automations — last night

02:00:00startissue agent · 1 issue ready
02:11:37claude#142 implemented · 6 files
02:12:58gatenpm run build · passed
02:13:06prdraft PR opened → dev
06:30:01startdaily briefing
06:30:14fetch40 sources · 187 new items
06:33:55checkedition drafted · schema ok
06:35:18publishpushed · site deploying
07:00:00you wake up · 1 PR to review · 1 edition published

01 · Products

Ready-made automations

Set up on your accounts and adjusted to your project. You own everything that runs.

All products

Available

Daily Briefing Autopilot

A news site about your field that writes itself every morning.

Trigger: Every day · 06:30
Result: A published edition

See how it works

Available

Issue-to-PR Agent

Label a GitHub issue before bed. Wake up to a draft pull request that builds.

Trigger: Label · agent-ready
Result: A draft pull request

See how it works

Something else eating your week?

Most repetitive work with a clear input and a clear result can be automated. Describe it and I will tell you honestly whether it is worth it.

Describe your task

02 · Process

From inquiry to running automation

01
You describe the task
A short message: what repeats, which tools you use, and what a good result looks like.
02
I set it up on your stack
Installed on your accounts and machine, tuned to your project, handed over with a short walkthrough.
03
It runs, you review
Results arrive as drafts, pull requests or published pages. Nothing important happens without a human gate.

03 · Principles

How these automations are built

Your accounts, your code
No platform to subscribe to. It runs on your GitHub, your Claude plan and your machine.
A human at the gate
Drafts and pull requests, never silent pushes to production.
Checked in code
Schema validation and build gates decide what passes, not the model’s confidence.
Used on my products first
Every automation here already runs on something I ship and maintain.

04 · Built and shipped

Products I build and run

All work

Live product · 2026

Dnevnik trudnoće

A Serbian pregnancy tracker: week-by-week guide, due-date calculator, a private diary and a directory of 60 maternity hospitals. Private data is protected in the database itself, not just hidden in the UI.