← Writing

Building OneAction: Week 1

February 3, 2026

I'm building OneAction in public — a tool that converts saved links into actionable tasks.

The Problem

I have 217 saved articles in Pocket. Read maybe 12 of them. Not because I don't want to. Because reading feels like a black hole. No outcome. Just consumption.

What if every saved link came with one action? Not a summary. Not highlights. An actual todo item.

That's OneAction.

Week 1 Progress

Shipped:

  • Auth (Clerk)
  • URL extraction API
  • Waitlist landing page
  • Supabase schema

Learning:

  • The gap between "saved" and "done" is real
  • Watching alpha users is teaching me more than any research
  • Distribution validation > feature building

Next:

  • Onboarding flow
  • Todoist integration
  • Beta user recruitment

Technical Decisions

Stack: Next.js 16, Tailwind, Supabase, Clerk

Why these? Speed. I need to ship fast and iterate.

AI extraction: Using structured output to pull actionable items. Not summaries — verbs. Tasks. Things you can check off.

The Real Challenge

Not the tech. The positioning.

"Read-it-later killer" is catchy but negative. "Action extraction" is accurate but jargon.

Still figuring out the language that resonates.

Follow Along

Building in public on Twitter/X.

If you have 50+ saved articles you never read, join the waitlist.

Week 1 of building OneAction. More updates coming.