# Waiting Room Readiness Scorecard

## PURPOSE
Liam Ottley's video argues that most businesses should stop buying AI systems for at
least a year because nobody — not the labs, not the startups — has shipped a complete
"AI native" platform. This project is a FastAPI web app that scores your own business
readiness AND any vendor pitch against the exact 7-piece platform rubric from the
transcript: context layer, data layer, agents, workspace, ledger, admin panel, and
custom apps. You answer the video's 3 self-questions and 4 vendor-questions and receive
a go / wait / go-all-in verdict instead of a gut feeling.

## WHY THIS MATTERS
The core lesson is "don't just dip a toe… a half migration is the worst of both." Most
buyers are currently paying for prototypes the labs will soon ship for ~$20 a seat.
Turning the transcript's advice into a numeric scoreboard means you are "ready before
anyone comes to try to sell to you" — the video's explicit recommendation.

## HOW TO BUILD
FastAPI plus Pydantic for models, and a pure scoring function so it is unit-testable.
Define the 7 components as a list, let the vendor payload set a delivery level for each,
and weight the answers to the 4 vendor questions (install service, real monthly cost,
model-agnosticism, 3-year survival). A key snippet:

```python
COMPONENTS = ["context", "data", "agents", "workspace",
              "ledger", "admin", "custom_apps"]

def coverage_score(delivered: dict[str, float]) -> float:
    return sum(delivered.get(c, 0.0) for c in COMPONENTS) / len(COMPONENTS)
```

Expose `POST /score` returning coverage, vendor-question score, and a verdict string.
Add `GET /components` and `GET /health`. Keep scoring deterministic so results are
comparable between vendors.

## WHEN TO USE IN PRODUCTION
Use it inside an agency that installs clients onto platforms (the video says the winning
agencies "pick a platform and go very deep on it"), in procurement to compare quotes, or
as an internal pre-sales scoreboard. Do NOT use it for real-time agent orchestration, and
do NOT treat the score as legal advice — compliance paperwork (for example a clinic
needing SOC 2 and a BAA like Lindy advertises) must still be checked separately.

## HOW IT HELPS STUDENTS
You practise FastAPI routing, Pydantic request/response modelling, extracting business
logic into a pure function, and returning structured JSON. It teaches production habits:
health checks, typed payloads, and separating "the market's opinion" (transcript scores)
from "your business's inputs".

## STARTER CODE
See `starter/main.py` — a runnable FastAPI skeleton with the component list, a
deterministic `coverage_score`, and `# TODO` markers for persistence and weights.
