Drop-in replacement for your existing RAG
Same API shape. Better results. No migration headaches.
import { Dexrag } from 'dexrag'
const dex = new Dexrag({
apiKey: process.env.DEXRAG_API_KEY
})
// Monte Carlo search with adaptive learning
const results = await dex.search({
query: "What are the key terms in section 4.2?",
documents: legalDocs,
options: {
exploration: 'monte-carlo',
adaptiveLearning: true
}
})
console.log(results.confidence) // 0.89Everything you need to ship
→ const dex = new Dexrag({ apiKey: 'YOUR_KEY' });
→ const collection = await dex.create({
→ name: 'docs', mode: 'adaptive'
→ });
Quick Start
Full code example with 5-minute integration. Upload documents, search naturally, get smarter automatically.
Zero Cold Start
No index building or embeddings to generate. Start searching immediately.
GET
POST
PUT
REST API
Use any language, no special SDKs required. Simple HTTP endpoints.
→ GET /api/search
→ POST /api/collections
→ PUT /api/docs/:id
API Documentation
Navigate complex technical docs with context awareness. Adapts to developer queries.
Benchmarks don't lie
18%
GPT-4 Embeddings
67%
Dexrag (Day 1)
89%
Dexrag (Day 30)
Measured on Google DeepMind LIMIT benchmark with 50K documents. Run it yourself →