Fallowmere, the first 90 days.

A note from Chris to Anita, written the week the build started. This is the plan a real client gets: short, in order, and honest about what has to happen before the clever part works.

Anita,

I tell people that are bold enough to finally explore AI in their organization that the only way AI works is with documented data. Right now Fallowmere's answers live in a handbook PDF nobody can search, a bus board in a group text and a spreadsheet of admissions families. A new system built on that is a recipe for revolt. Your office will stop trusting it the first time it is wrong, and it will be wrong on day one. So here is the order, and why.

The stakes are small and constant. Beth answers the same 300 questions. A late bus becomes a dozen phone calls. A family that toured in September goes quiet in October and nobody notices. None of that is a crisis. All of it adds up to a school that feels slower than it is.

The payoff is a school that answers the same day. Families hear about a late bus from you first. Tom sees every inquiring family on one page. Your teachers walk into conference week having already had the hard conversation once.

Days 1 to 30: family requests, the bus board, and the handbook

• Family requests. Overnight emails are routed to the nurse, the bursar or the right division office, each with a drafted reply in the school's voice. Beth approves them. Nothing goes out until she says so.
• The bus board. All 12 routes on one screen, with a late-bus text to families and a printed roll-out sheet for the drivers. Marcus tells me which routes are wrong on day one, and I fix the board before anyone trusts it.
• The handbook, scanned in. Every page, with its page number. If a rule is not in the handbook, the answer will say so. It will not guess.
• What I need from you: one hour with Beth and one with Marcus, and a decision on who owns the handbook when a rule changes.

Days 31 to 60: the first read, and the handbook desk

• RadRoot, first read. "Family pulse," confidential and grouped by division. Eleven of twelve parents answered in the sandbox version. Expect a different number in real life and expect one answer that changes how you run a division.
• Handbook desk, for the office. Beth and her team use it first. Every answer cites its page. When it is wrong, we fix the handbook, not the answer.
• Handbook desk, for families. Only after the office has trusted it for two weeks.
• What I need from you: let the division heads see the read. The summary is for them, not for a drawer.

Days 61 to 90: admissions, and rehearsal before conferences

• The admissions pipeline. Live before open house season. A page for every inquiring family, the RSVP, and a drafted follow-up for the ones who went quiet. Tom approves every message.
• RadSim for every teacher. The week before conferences, ten minutes each. The upset parent after a grade and the tuition-assistance conversation come first, because those are the ones that go badly.
• Decide the next build from the reads. I will bring you the three things the data says, ranked. You say go to one.

What this costs and what it does not

The Office app is a RadWorks build: a monthly subscription, no setup fee, and the first month's scope is written before you pay. RadRoot and RadSim are each their own subscription and you can switch either off. My time is the part that scales down, not up: by day 90 you should need me for a monthly note and a quarterly build decision, and not for the Tuesday.

Got a Monday-morning inbox problem? Got a "where do I even get started" problem? Happy to chat.

Chris

Fallowmere is a fictional company. The plan is real in shape: it is how Rad sequences a build for any organization with a binder, a group text and a spreadsheet that holds more than it should.