Search100

Production-grade search engine. PostgreSQL-native. No Elasticsearch required.

Built by H33.ai — the team behind the only post-quantum FHE + ZK + Dilithium authentication pipeline running at internet scale.

What is Search100?

Search100 is a full-featured search engine that runs on PostgreSQL. No Elasticsearch. No Solr. No separate search infrastructure. Just your existing Postgres database with full-text search, fuzzy matching, natural language parsing, faceted filtering, real-time WebSocket updates, and vector embeddings.

Features

• Full-text search — PostgreSQL tsvector with ranking and highlighting
• Fuzzy matching — Trigram similarity (pg_trgm) for typo-tolerant search
• Natural language parsing — "show me urgent cases from last month" → structured query
• Faceted filtering — Dynamic facets with counts, multi-select, range filters
• Search relevance tuning — Configurable field weights, boost factors, decay functions
• Query suggestions — Autocomplete with popularity-weighted suggestions
• Search history — Per-user search history with analytics
• Saved searches — Save and share search configurations
• Real-time search — WebSocket-powered live search results
• Vector embeddings — Semantic search with cosine similarity (pgvector)
• Search analytics — Click-through rates, popular queries, zero-result tracking
• Circuit breaker — Production-grade fault tolerance
• Smart caching — Redis-backed with intelligent invalidation

Data Source Connectors

Pull data from external sources into your search index:

Connector	Status
Slack	Production
Gmail	Production
Google Drive	Production
Notion	Production
Confluence	Production
SharePoint	Production

Quick Start

npm install @h33/search100

import { SearchService } from '@h33/search100';

const search = new SearchService({
  database: process.env.DATABASE_URL,
  redis: process.env.REDIS_URL // optional, for caching
});

// Full-text search with fuzzy matching
const results = await search.searchCases('patient smith', {
  organizationId: 'org_123',
  limit: 25,
  filters: { status: 'open' }
});

// Natural language search
const nlResults = await search.naturalLanguageSearch(
  'show me urgent cases from last week assigned to Dr. Jones'
);

// Faceted search
const faceted = await search.facetedSearch('insurance claim', {
  facets: ['status', 'priority', 'assignee', 'dateRange']
});

Requirements

• Node.js >= 18
• PostgreSQL >= 14 (with pg_trgm and optionally pgvector extensions)
• Redis (optional, for caching and real-time features)

PostgreSQL Extensions

-- Required for fuzzy search
CREATE EXTENSION IF NOT EXISTS pg_trgm;

-- Optional for vector/semantic search
CREATE EXTENSION IF NOT EXISTS vector;

Architecture

┌──────────────────────────────────────────────────┐
│                   Search100                       │
├──────────────┬───────────────┬───────────────────┤
│  NL Parser   │  Facet Engine │  Relevance Scorer │
├──────────────┼───────────────┼───────────────────┤
│  Full-Text   │  Fuzzy Match  │  Vector Search    │
├──────────────┴───────────────┴───────────────────┤
│              PostgreSQL (pg_trgm + pgvector)      │
├──────────────────────────────────────────────────┤
│              Redis (cache + real-time)             │
└──────────────────────────────────────────────────┘

Need Encrypted Search?

Search100 searches plaintext data. If you need to search encrypted data — where the database, the search engine, and the infrastructure never see your plaintext — check out H33 Encrypted Search.

H33 Encrypted Search uses Fully Homomorphic Encryption (FHE) to perform pattern matching on ciphertext. Your data stays encrypted during storage, transit, and search. One API call.

License

MIT — use it however you want.

About H33

H33.ai builds post-quantum encryption infrastructure. FHE, ZK-STARKs, Dilithium signatures, encrypted biometrics — one API call, 2.17M auth/sec sustained. 114 patent claims pending.

---

Search100 is extracted from H33's internal search platform. Battle-tested in production.

---

H33 Products: H33-74 · Auth1 · Chat101 · Cachee · Z101 · RevMine · BotShield

Introducing H33-74. 74 bytes. Any computation. Post-quantum attested. Forever.

---

H33 Product Suite

Product	Description
H33.ai	Post-quantum security infrastructure
V100.ai	AI Video API — 20 Rust microservices, post-quantum encrypted
Auth1.ai	Multi-tenant auth without Auth0
Chat101.ai	AI chat widget with Rust gateway sidecar
Cachee.ai	Sub-microsecond PQ-attested cache
Z101.ai	20+ SaaS products on one backend
RevMine.ai	Revenue intelligence platform
BotShield	Free CAPTCHA replacement

Introducing H33-74. 74 bytes. Any computation. Post-quantum attested. Forever.