brittany wilson-valencia Work with me
← AI operating system series · 02 of 04

Program Work Hub

A private work hub that turns about ten recurring programs into dated, trackable work, all from a few anchor dates.

My role

Product owner, designer, and tester. I directed an AI assistant to build it.

Timeline

Late October to mid-November 2026. 18 releases, each previewed first.

Scope

About ten recurring programs: all hands, newsletters, recognition, mentorship, internships, and more.

Status

Live, private, and single user.

90 rules

Playbook rules that generate every dated action from a few anchor dates

81 actions

Imported from my existing tracker, with no dates silently shifted

18

Production releases, every one previewed and approved by me first

The problem

Communications programs run on dates that depend on other dates. A newsletter call for content is due three weeks before the send. Recognition needs nominations, voting, winners, and payout messages. Multiply that by ten programs and tracking took up most of my day-to-day operations.

The real cost wasn't effort, it was exposure. When an event moved or got cancelled, nothing told me. Deadlines were easy to remember the week they mattered, and easy to miss the week before.

My approach

  1. 1. Start from what I trust

    I built from my existing 90-rule playbook, 81-action list, and a rough mockup, not a blank page.

  2. 2. Set boundaries first

    Private by default, and read-only toward my calendar and mail. The hub never sends anything.

  3. 3. Preview, then publish

    Every change was shown to me first, and I approved each release.

  4. 4. Ask "is this the best way?"

    One follow-up question turned a manual workaround into an automatic calendar check.

Program Work Hub overview with sample data
Program Work Hub decisions needed view with sample data

Screens show sample data.

What I learned

The most valuable moments were questions, not instructions.

Naming the boundaries out loud (read-only, never sends, never decides alone) made every later feature easier to approve. Previews, undo, and honest "not tested" labels built more trust than any promise of accuracy.

Next · 03 of 04Work OS