Vector Search & AI

NaijaBase supports pgvector — store and search vector embeddings from any AI model, with data hosted in Lagos, Nigeria.

Enable vector search

pgvector is enabled automatically on every NaijaBase project. No setup required.

Create a vector table

CREATE TABLE documents (
  id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
  content TEXT NOT NULL,
  embedding vector(1536),
  metadata JSONB,
  created_at TIMESTAMPTZ DEFAULT NOW()
);

-- Index for fast cosine similarity search
CREATE INDEX ON documents
  USING ivfflat (embedding vector_cosine_ops)
  WITH (lists = 100);

Store embeddings

import { createClient } from '@naijabase/js'
import OpenAI from 'openai'

const naijabase = createClient(URL, KEY)
const openai = new OpenAI()

// Generate embedding
const { data } = await openai.embeddings.create({
  model: 'text-embedding-3-small',
  input: 'Your text here'
})

// Store in NaijaBase (Lagos)
await naijabase.from('documents').insert({
  content: 'Your text here',
  embedding: data[0].embedding
})

Semantic similarity search

// Generate query embedding
const queryEmbedding = await openai.embeddings.create({
  model: 'text-embedding-3-small',
  input: 'Nigerian fintech regulations'
})

// Find similar documents via RPC
const { data } = await naijabase.rpc('match_documents', {
  query_embedding: queryEmbedding.data[0].embedding,
  match_threshold: 0.8,
  match_count: 10
})

console.log(data) // Top 10 similar documents

Or use the SDK shorthand:

const { data } = await naijabase.vectorSearch(
  'embedding',
  queryEmbedding.data[0].embedding,
  { threshold: 0.8, limit: 10 }
)

Use cases

Semantic search — find similar documents, articles, products
AI chatbot memory — store conversation history as embeddings
Fraud detection — find similar transaction patterns
Recommendation engine — suggest similar products or content
Document similarity — match contracts, CVs, legal documents

Supported embedding dimensions

Model	Dimensions
OpenAI `text-embedding-3-small`	1536
OpenAI `text-embedding-3-large`	3072
OpenAI `text-embedding-ada-002`	1536
Cohere `embed-multilingual-v3.0`	1024
Custom models	Any

Why store embeddings in Lagos?

Nigerian AI teams working with sensitive data (medical records, financial data, legal documents) must keep that data in Nigeria under NDPA 2023. NaijaBase + pgvector means your entire AI pipeline — from raw data to embeddings — never leaves Nigerian soil.

Enable vector search​

Create a vector table​

Store embeddings​

Semantic similarity search​

Use cases​

Supported embedding dimensions​

Why store embeddings in Lagos?​