Archived — This repository is archived as a hackathon submission (Tokyo AI Festival 2025). No further development is planned.

## Hackathon submission (Tokyo AI Festival 2025)

- Status: submitted on **2025-09-10**; attending **Demo Day (Sep 13)** as observer.
- 4-min demo video:[![Demo Video](https://img.youtube.com/vi/2EWPqylrXTA/hqdefault.jpg)](https://www.youtube.com/watch?v=2EWPqylrXTA&t=15s)
- Slide deck (PDF): ./pitch/AIPMO_PoC_slides_v010.pdf
- Release tag: **v0.1.0**

---

## What this PoC does (1-minute read)

**AI as a calm “pilot” (PMO)** for first-time real-estate sellers.  
It quietly shows **where you are** and **the next single action**, without alarm fatigue.

- **Action → Pack:** Journey compass + checklist + risks + email draft + Slack snippet + **.ics** (calendar memo includes key points)
- **KPI (Calm):** green-heavy / minimal red to reduce anxiety; alerts only when truly risky
- **Optional “Evidence”:** brief references (RAG), shown only when needed

---

## Story & How-to (EN)

- Long-form note (EN): **[Selling a Condo with AI as PMO — Story & Playbook](https://note.com/usekbota/n/n580a95f811f3)**
  > Why “calm mode” matters, how the journey compass guides decisions, and a step-by-step walk-through.
- (JP preview available separately)

---

## Quick start

```bash
python3 -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
python app/app.py  # http://127.0.0.1:7860

# AI Real Estate PMO (PoC)

Bilingual Gradio app that guides first‑time home sellers through a stepwise journey (Prep → Listing → Viewing → Offer → Finance → Close).
Calm KPI cards (green‑heavy thresholds), pack generation (Checklist + Email + Slack + ICS), and optional Evidence (RAG).

## Features
- 🇯🇵/🇬🇧 toggle switches **UI and outputs**.
- **Pack generator**: checklist, risks, email draft, chat snippet, and `.ics` calendar.
- **Calm KPI**: response/viewing/offer rates with gentle thresholds.
- **Evidence (optional)**: show note snippets via `data/rag_chunks.jsonl`.

## Project layout

AIPMO_RealEstate_PoC/ ├── app/ │ └── app.py # main Gradio app ├── data/ │ ├── events_sample.csv # required: event timeline │ ├── contacts.csv # optional: actor → to/cc/attachments (ignored by git) │ ├── kpi.csv # optional: KPI timeseries (ignored by git) │ └── rag_chunks.jsonl # optional: RAG evidence (ignored by git) └── README.md

## Requirements
- Python 3.10+
- macOS/Windows/Linux

## Quick start
```bash
cd AIPMO_RealEstate_PoC
python3 -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install --upgrade pip
pip install gradio pandas pyyaml
python app/app.py
# Open http://127.0.0.1:7860

CSV schema (events_sample.csv)

Columns (header row required):

event_id,date,actor,category,description,expected_action,success_criteria,risk_level
E101,2025-07-01,Seller,Prep,売却検討を開始（要件整理）,要件メモ作成・家族合意,掲載準備OK,Low
...

• date accepts YYYY-MM-DD or YYYY/MM/DD.

KPI (data/kpi.csv)

Columns: date,pv,inquiries,viewings,offers (any case).

Evidence (data/rag_chunks.jsonl)

One JSON per line:

{"text":"Use a standard term sheet to avoid verbal ambiguity","source":"MyNote","tag":"offer"}

English translations

• Exact phrases → JP2EN dictionary
• Substrings → GLOSSARY Add entries in app/app.py and restart.

License

MIT (or choose your own before publishing).

Release v0.1.0

• Initial PoC demo (tagged).