AIU Blueprint
AIUBlueprint
In design

The AI requirements studio: bring a fuzzy idea, answer numbered decisions it mostly predicts for you, and ratify a plan that writes straight into RoadWork.

Your studio

Welcome back, Mara. One blueprint in progress.

A blueprint turns a fuzzy idea into a decided, buildable plan. Pick up where you left off, or start a new one.

Your studio/First run

No blueprints yet

Nothing here is a placeholder — this studio is genuinely empty until you bring an idea. Start with a rough one; you don't need to know how to build it.

Why this mattersAn empty studio says it is empty. No fake sample projects, no invented counts — the honest zero (design-standards U2).

Your studio/Bring the idea
Step 1 · Bring the idea

Tell it your idea. Rough is fine.

Write it the way you'd say it out loud — messy, uncertain, half-formed. Blueprint reads it and starts asking. You never have to know the right words or how to build it.

This is Dr. Mara Okafor's real, rough idea. Edit it or write your own — then start.

Write a line or two first — even a rough idea is enough to start the interview.

What happens next: Blueprint reads your idea and asks a handful of plain questions, one at a time. Then it turns everything into numbered decisions — most already answered with a prediction, so you mostly just confirm.
Back to studio
Bring the idea/Interview
Step 2 · Interview

A few questions. One at a time.

Blueprint asks the way a sharp product lead would — never a form-dump. Answer in a sentence; it listens and moves on.

Question 1 of 4
Who is this for first — your front-desk team, the vets, or the pet owners?

Knowing whose screen this is decides almost everything else, so it goes first.

The front desk owns the list day to day. The vets need to see it. Owners get the reassuring part — but that's second.
What does "done well" look like, in one sentence?

The outcome in the owner's own words becomes the measure of success later.

No recovering patient goes more than a few days without a check-in, and owners know what's normal and when to worry.
What exists today that this replaces or plugs into?

Blueprint would rather plug into what you already have than make you enter things twice.

Our practice-management system has the visit records. The follow-up part is sticky notes and memory.
What's the one thing that would make you trust it on day one?

The non-negotiable becomes a decision the plan is built to protect.

It never invents a medical claim. It only says "time for a check-in." I decide the medicine.

No-JavaScript noteWith scripting off, all four questions and Mara's answers render stacked and readable — the "one at a time" pacing is a progressive enhancement, never a way to hide content.

Interview/Drafting

Reading your answers, drafting each decision

Blueprint is turning the interview into numbered decisions — and predicting an answer for each from what you actually said. This is real work, not a spinner for show; when a decision has no honest prediction it will be flagged as an open question, not guessed.

Your studio/Something went wrong

Couldn't reach the studio

Your idea and every answer are saved — nothing is lost. This is a connection problem on our side, not yours. Try again, or come back to it later.

Interview/Decisions
Step 3 · Decisions

Every open question, numbered and mostly answered.

Blueprint drafted a decision for each thing your plan needs settled — and predicted an answer for most from your own words. Confirm the ones that look right; change any you want; answer by number. One (D7) it won't guess.

3 of 9 confirmed Confirm the rest to unlock Ratify.
What do we call it?Confirmed: Recovery Rounds Confirmed
Predicted answer

Recovery Rounds

Why we predict thisIt reads as your clinic's own program — a thing your team and owners would say — not software jargon. Familiar words earn trust on day one.
Also considered
  • Alt AAftercare Tracker
  • Alt BCheck-In List

You confirmed this. Reopen any time before you ratify.

Who owns the daily list?Confirmed: the front-desk team; vets can see it Confirmed
Predicted answer

The front-desk team owns the daily list; every vet can see it.

Why we predict thisYou said the front desk owns it day to day and the vets need visibility — so the list is one shared view, owned by the desk, readable by the clinicians.
Also considered
  • Alt AEach vet owns their own patients' list

You confirmed this. Reopen any time before you ratify.

What starts a follow-up?Confirmed: a visit tagged surgery or serious diagnosis Confirmed
Predicted answer

A visit tagged "surgery" or "serious diagnosis" automatically starts a recovery track — no one has to remember to add it.

Why we predict thisThe whole pain is that people fall through the cracks. Auto-starting from the visit removes the human step that gets forgotten.
Also considered
  • Alt AStaff add each patient manually
  • Alt BRules per procedure type

You confirmed this. Reopen any time before you ratify.

The check-in schedule?Predicted: day-3 call + day-10 recheck nudge, adjustable per case Predicted — your call
Predicted answer

A day-3 check-in call and a day-10 recheck nudge, adjustable per case.

Why we predict thisYou named "around day three" yourself, and a recheck nudge closes the loop you said you drop. Per-case adjustment keeps the vet in control.
Also considered
  • Alt AOne fixed schedule for everyone
  • Alt BThe vet sets the schedule per patient
Change it

Decided.

What does the owner see?Predicted: a plain "what's normal / when to worry / your recheck date" note Predicted — your call
Predicted answer

A plain note: what's normal in recovery, when to worry, and your recheck date. No diagnosis, no medical advice.

Why we predict thisYou called the owner part "reassurance," and you drew a hard line at medical claims (D6). This gives owners comfort without ever crossing that line.
Also considered
  • Alt AOwners see nothing in v1 (staff-only)
Change it

Decided.

The safety line?Predicted: never states a diagnosis or outcome — only "due for a check-in" Predicted — your call
Predicted answer

The tool never states a diagnosis, a prognosis, or a medical outcome. It only surfaces "due for a check-in" and shows the vet's own notes. The clinician decides the medicine.

Why we predict thisThis is your day-one non-negotiable, in your words. The plan is built to protect it: the tool reports only what a person can verify — a date is due — and never makes a claim it can't stand behind.
Also considered
  • Alt ATool suggests likely issues to watch
Change it

Decided.

Where does the visit data come from?Open question — a fact we won't guess Open question
We won't invent this

You want to read visits from your practice-management system instead of typing them twice. Whether that system lets another tool read its data — and how — is a real fact only you or your vendor can confirm. Guessing it would put a made-up capability in your plan. So Blueprint leaves it open, on purpose.

Until you answerThe plan carries D7 as a marked open question. RoadWork gets a "confirm the data source" task instead of a fake integration — so nothing false ships downstream.

Noted — carried as an open question.

What's in the first version?Predicted: staff list + auto-triggered tracks + the schedule; owner note is v2 Predicted — your call
Predicted answer

Version one: the staff list, auto-triggered recovery tracks (D3), and the day-3 / day-10 schedule (D4). The owner-facing note (D5) follows in v2.

Why we predict thisThe staff list is what fixes the dropped-patient pain fastest. Owner reassurance is real but lower-risk to defer — so v1 ships the thing that hurts most today.
Also considered
  • Alt AEverything at once, including the owner note
Change it

Decided.

How will you know it works?Predicted: % of recovering patients checked in on time Predicted — your call
Predicted answer

One number the clinic can see: the share of recovering patients who got their check-in on time.

Why we predict thisIt's your "done well" (Q2) turned into something the system can actually count — not a feeling, not a claim, a measurable rate the tool can show you.
Also considered
  • Alt AStaff satisfaction survey
  • Alt BOwner reply rate
Change it

Decided.

The magic, plainlyYou don't fill in a blank spec. Blueprint predicts each answer from what you actually said, so you mostly confirm — and where it honestly can't predict (D7), it flags an open question instead of guessing. That's the difference from every "requirements template."

Decisions/Ratify
Step 4 · Ratify

One look at the whole plan, then you ratify.

This is everything you settled, in one place. Ratifying locks it and writes it into RoadWork for the build. You can still reopen any decision until you do.

Recovery Rounds · the settled plan
D1What we call itRecovery Rounds
D2Who owns the daily listFront-desk team; vets can see it
D3What starts a follow-upA visit tagged surgery / serious diagnosis
D4The check-in scheduleDay-3 call + day-10 recheck, adjustable
D5What the owner seesPlain "normal / worry / recheck" note
D6The safety lineNever a medical claim — "due for a check-in" only
D7Where the visit data comes fromOpen question — confirm with the vendor
D8What's in version oneStaff list + auto tracks + schedule; owner note v2
D9How you'll know it works% checked in on time

6 decisions still need you before you can ratify — D4, D5, D6, D8, D9 to confirm, and D7 is an open question you can carry as-is. Back to the decisions.

Honest fenceYou can't ratify an unfinished plan — the button stays off until every decision is settled. Forcing this screen early shows the "still needs you" fence with a way back, never a half-written handoff.

Ratify/To RoadWork
Step 5 · Handed off

Ratified. Your plan is now tracked work in RoadWork.

Blueprint's job ends at "decided." The moment you ratified, the plan was written into RoadWork as a work-tree the agents can build — and from here RoadWork owns it.

Blueprint owns everything up to decided RoadWork owns tracking + priority + the build
What was written — the work-tree (Idea → Task)
Idea
Recovery Rounds — never drop a recovering patient
From your intake + D2 outcome
Initiative
Follow-up program for post-surgery / serious-diagnosis patients
From D3 (what starts a follow-up)
Epic
Staff recovery list — version one
From D8 (v1 scope)
Feature · with tasks
Auto-triggered tracks + day-3 / day-10 schedule
P1Read tagged visits, open a recovery track automaticallyD3
P1Day-3 check-in + day-10 recheck, adjustable per caseD4
P1Guardrail: surface "due for a check-in" only — no medical claimD6
P2Owner note: normal / worry / recheck dateD5
P2Show the on-time-check-in rateD9
Epic · open question
Confirm the visit-data source with your practice-management vendor
From D7 — carried as an open question, not a fake integration

Every task traces backEach task cites the decision it came from (the small D-tag on the right). Nothing in the plan is here by accident — and the one thing you didn't know (D7) is a task to confirm, never an invented capability.

To RoadWork/Store of record
Step 6 · Store of record

Your plan's source of truth, kept for good.

The Design Hub holds what Recovery Rounds should look like and what each part is supposed to do. If the build and this ever disagree, this wins. Come back any time — it's yours.

One truth, not twoThe store of record is the truth for UI + requirements. The build reads from it; if they ever drift, the store wins and the build gets corrected. That's what keeps a plan from quietly rotting after it's decided.

Blueprint — functional prototype. Customer world: Dr. Mara Okafor, Riverside Animal Care · "Recovery Rounds." Ground: AGENT-ENGINE-DASHBOARD-PROPOSAL §10 · AIU-STYLE-GUIDE · DESIGN-STANDARDS-CHECKLIST. CVD-safe by glyph + word + lightness. --c-bp PROPOSED-PENDING-CVD (Verifier locks via cvd-sim.mjs).