Your people, in one list.

1. 1

   Your people live in three places: a spreadsheet someone started years ago, the WhatsApp groups, and the pastor's memory. None of them agree.

2. 2

   The same person is written down two or three ways, so nobody can say how many people the church has.

3. 3

   Someone drifting is noticed late, usually when a friend says they have not been seen in a while.

4. 4

   The few who always say yes end up on three rosters. Nobody sees it until someone is tired.

Claude reads the list by number, never by name: P-23, not a person. You keep the names and put them back. That follows the rule in my note on member data and AI.

• It finds patterns across both lists at once: came often, then stopped; serves on three teams; comes every week but has never been asked.
• It drafts: a short note to someone you have missed, a new roster. It sends nothing.
• A person decides. The pastor reads the lists, makes the calls and moves the slots.
• Your church agrees in writing before any member information goes in, even as numbers.

A paragraph about rosters gets skipped. A button gets pressed. So this is a drawing you work, not a page you read.

1. 1

   The numbers came first: one small model of an example church, written before anything was drawn. Every number on the page comes from it, so the list, the rings, the rosters and the counts always agree.

2. 2

   One button for each step, only the next one live. Each step plays on the drawing: names fly into one list, rings fill a week at a time, Claude reads the list.

3. 3

   Auto plays the whole thing on a loop, with a pointer pressing each step, for anyone who would rather watch.

4. 4

   A close-up of the part doing the work. Detail A follows one person through both lists.

5. 5

   Before it went up, every step was checked as a still, and every move was scanned frame by frame for jumps.

Claude wrote the model, the drawing and every step with me in one sitting. The site runs without JavaScript, so all of it is plain HTML and CSS and works on any phone.

The model

Rule or count	Value	Where it comes from
Names on file, before	81 entries	52 spreadsheet rows, 22 in the WhatsApp group, 7 only in the pastor's memory
People, after one list	64	17 entries were the same person written down twice
Teams	6 teams, 32 slots each month	Welcome 8, Kids 8, Media 4, Tea 4, Prayer 4, Cleaning 4
How slots are shared	Equally within a team	An assumption. Real rosters are lumpier
Average Sunday	42 people	The mean of the eight Sundays
Drifting	6 people	Came at least 4 of the first 5 Sundays, none of the last 3
Carrying too much	4 people, 38% of slots	On 3 or more teams
Every week, never asked	9 people	At least 7 of 8 Sundays, on no team
After the pastor's change	3 people, 28% of slots	Kids slots moved from P-23 to P-16 and P-27: 25 serving, 7 never asked

These are example numbers for a made-up church of 64. On your list, the numbers come from your list.

• Your own list. Export the spreadsheet, list the WhatsApp groups and sit with the pastor for a morning. That is enough to start.
• In your church's accounts. The list lives in your church's own Google or Microsoft account, and Claude is your church's subscription, not mine.
• The same two questions, every month. Who are we missing, and who is carrying too much? Claude does the reading; you do the calling.
• The next one, built in your church. With this one as the pattern, a volunteer with Claude can build the next: the Sunday setup, children's check-in, the backup power.