dukesky/PolarBear
dukesky/PolarBear has 201 GitHub stars and 33 forks. GitHot measured 0 stars in 24 hours. Last checked 2026-11-18 11:04 UTC. Last push 2026-08-22.
What you can do
- π Hybrid Search: seamlessly blends keyword matching (BM25) with AI-powered semantic search (Embeddings) for superior result relevance.
- β‘ No-Code Ingestion: Upload your data via CSV, Excel, or Google Sheets. No coding required.
- π§ AI-Ready: Built-in vectorization pipeline using state-of-the-art embedding models.
- π Insights Dashboard: Track user behavior, top queries, zero-result searches, and conversion metrics.
Measured star growth
- in 24h
- 0
- in 7 days
- Collecting history
- in 15 days
- Collecting history
- in 1 month
- Collecting history
Net star changes over overlapping periods, not daily totals. Missing measurements are left blank. Stars measure attention, not software quality.
About
Fully Open Source Search Engine for SME (Small Medium Size Enterprise)
Developer ToolsREADME.md
Original
PolarBear π»ββοΈ
The Open-Source Hybrid Search Engine for SMEs
Features β’ Getting Started β’ Documentation β’ Contributing
π Introduction
PolarBear is a powerful, no-code, AI-enhanced search engine designed specifically for Small and Medium-sized Enterprises (SMEs). It democratizes access to advanced search technology, allowing business owners to create a professional search experience for their products, services, or inventory in minutesβcompletely free and open source.
Unlike complex enterprise solutions, PolarBear focuses on simplicity without compromising on power. It combines Keyword Search (Meilisearch) and Semantic Vector Search (FAISS) to deliver results that are both accurate and contextually relevant.
PolarBear is open-core: this repository is the complete product under the MIT license β clone it, self-host it, study it, extend it. The same codebase can also run as a hosted multi-tenant service where store owners simply register, upload their catalog, and paste two lines of code into their website.
ποΈ Add Search to Your Store β in 3 Steps
Store owner with no technical background? This is for you.
- Upload your products β a CSV with your product names, descriptions, and prices.
- Copy your embed code β the Settings page generates it, with platform-specific guides for Shopify, WordPress, and Wix.
- Paste it into your site β into your site builder's "custom code" box. Done.
<div id="polarbear-search"></div>
<script src="https://your-polarbear-server/widget.js" data-key="pb_search_..." async></script>
Your shoppers get a search-as-you-type overlay with AI semantic matching ("cozy gift" finds your scented candles); you get a dashboard of what they searched and clicked.
π Full integration guide β docs/INTEGRATION.md β step-by-step per platform, CSV format, widget options, API reference, self-hosting, and troubleshooting.
See it working in 2 minutes
The repo ships a demo storefront β a small hardware store with 18 products β so you can experience both sides without writing anything:
cd demo-store && python3 -m http.server 4173
Upload demo-store/products.csv to your PolarBear account, paste your search key into the page, then search βmy faucet dripsβ. You get the faucet washer assortment β a product whose title shares no word with your query. That's the difference between keyword search and PolarBear.
π¦ Demo storefront β demo-store/
π Features
- π Hybrid Search: seamlessly blends keyword matching (BM25) with AI-powered semantic search (Embeddings) for superior result relevance.
- β‘ No-Code Ingestion: Upload your data via CSV, Excel, or Google Sheets. No coding required.
- π§ AI-Ready: Built-in vectorization pipeline using state-of-the-art embedding models.
- π Insights Dashboard: Track user behavior, top queries, zero-result searches, and conversion metrics.
- ποΈ Product Management: Built-in catalog management to edit products and upload images directly.
- π Multi-Tenant: Register an account, get isolated indexes and API keys β one deployment serves many businesses.
- π§© Embeddable Widget: paste two lines into any website (Shopify, WordPress, Wix, custom) and get an AI search box for your catalog.
- π± Works in apps too: the search API is a plain HTTP call β integrate iOS, Android, or any backend in a few dozen lines. See the integration guide.
- βοΈ Cloud-Native: Dockerized for easy deployment on Google Cloud Run, AWS, or your own server.
- π Open Source: 100% free to use, modify, and distribute.
π οΈ Tech Stack
| Component | Technology | Description |
|---|---|---|
| Frontend | Next.js (React) | Modern, responsive admin and search UI. |
| Backend | FastAPI (Python) | High-performance API for ingestion and search. |
| Search | Meilisearch | Lightning-fast keyword search engine. |
| Vector DB | FAISS | Efficient similarity search for embeddings. |
| Infrastructure | Docker | Containerized for consistent deployment. |
π Getting Started
Prerequisites
- Node.js 18+
- Python 3.11+
- Docker & Docker Compose
Quick Start
Clone the Repository
git clone https://github.com/dukesky/PolarBear.git cd PolarBearStart everything
cd infrastructure docker compose up -dThat's the whole stack β Meilisearch, Postgres, the API and the dashboard. Open http://localhost:3000/login and register.
Set
SECRET_KEYbefore exposing it to anyone:SECRET_KEY=$(openssl rand -hex 32) docker compose up -dPrefer to run the services yourself while developing? The three-terminal setup is below.
Start Infrastructure only (development)
cd infrastructure docker compose up -d meilisearch postgresStart Backend (new terminal, from the repository root)
cd backend poetry install poetry run uvicorn app.main:app --reload --port 8000Start Frontend (new terminal, from the repository root)
cd frontend npm install npm run devExperience PolarBear
- Register: Go to
http://localhost:3000/loginand create an account. - Upload Data: Go to
http://localhost:3000/uploadand upload a CSV (e.g.,sample_products.csv). - Search: Visit
http://localhost:3000/searchto try the hybrid search. - API Keys: Visit
http://localhost:3000/settingsto copy your public search key. - Embed: In Settings β "Add search to your website", copy the snippet into any web page β the widget talks to your backend.
- Insights: Check
http://localhost:3000/insightsfor analytics and product management.
- Register: Go to
Upgrading from a pre-multi-tenant version? Indexes are now per-account. After upgrading, register an account and re-upload your CSV β old data in the shared
productsindex is not migrated automatically.
π Documentation
For store owners & integrators:
- Integration Guide β add PolarBear search to any website (Shopify, WordPress, Wix, custom) in 3 steps; CSV format, widget options, search API, self-hosting, troubleshooting.
Development history β detailed walkthroughs for each phase:
- Phase 1: Setup & MVP Core - Infrastructure and basic search.
- Phase 2: Ingestion & Indexing - CSV parsing and hybrid indexing pipeline.
- Phase 3: Search Interface - Frontend UI and search logic.
- Phase 4: Analytics - Tracking user queries and dashboard.
- Phase 5: UX & Product Mgmt - Image support, catalog editing, and advanced UX.
π€ Contributing
We welcome contributions from the community! Whether it's fixing bugs, improving documentation, or suggesting new features, your help is appreciated.
- Fork the Project
- Create your Feature Branch (
git checkout -b feature/AmazingFeature) - Commit your Changes (
git commit -m 'Add some AmazingFeature') - Push to the Branch (
git push origin feature/AmazingFeature) - Open a Pull Request
π License
Distributed under the MIT License β see LICENSE for the full text.
π¬ Contact
Project Link: https://github.com/dukesky/PolarBear