No-DB Mini Bot

Tiny corpus, no infrastructure

A vectorless (PageIndex/BM25) chatbot in Streamlit. No embeddings, no GPU, no database — the index is a single file on disk. Perfect for small/static corpora, demos, and offline use.

Download ZIP

Live previewinteractive mock — no backend

localhost:8501Streamlit

Offline · BM25

Local recall.

Try one — live mock, no backend.

Stack

Frontend

Streamlit

Architecture

Vectorless (BM25)

Framework

raw python

Vector DB

None (BM25)

Corpus

small

Complexity

Low

How it works

Browser→

Streamlit app→

FastAPI /query→

Vectorless (BM25)

The frontend talks only to the backend; your API keys + pipeline URL stay server-side.

Get started

Option A · Docker Compose (one command)

cp .env.example .env        # add your API key(s)
docker compose up --build

Option B · Backend (manual)

cd backend
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env        # add your API key(s)
python pipeline.py ingest ./corpus
uvicorn serve:app --reload  # http://localhost:8000

Option B · Frontend (Streamlit)

cd frontend
pip install -r requirements.txt
cp .env.example .env
streamlit run app.py        # http://localhost:8501

The downloaded README.md has the full guide — vector DB setup, API keys, and deployment to Render/Railway + Streamlit Cloud.

Project structure21 backend · 6 frontend · 31 files total

backend/

.github/

workflows/

deploy.yml

eval/

__init__.py

README.md

run_eval.sh

synthetic_qa.py

test_answer_quality.py

test_retrieval.py

ingestion/

__init__.py

chunker.py

loader.py

observability/

__init__.py

tracing.py

.env.example

config.yaml

docker-compose.yml

Dockerfile

pipeline.py

README.md

render.yaml

requirements.txt

serve.py

frontend/

.streamlit/

config.toml

.env.example

.gitignore

app.py

Dockerfile

requirements.txt

.env.example

.gitignore

docker-compose.yml

README.md

Ready to build?

Download the full monorepo and follow the README.

Download ZIP