JavaScript & TypeScript: AI in the Web Stack

For web developers using Node.js, Next.js, or React, Ollama provides a native JavaScript library that follows the same clean patterns as the Python version.

1. Installation

npm install ollama

---

2. Server-Side Usage (Node.js)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'llama3',
  messages: [{ role: 'user', content: 'Why is the sky blue?' }],
})

console.log(response.message.content)

---

3. Streaming in Browser/Node

The JS library uses "Async Generators," which are perfect for real-time UI updates.

import ollama from 'ollama'

const message = { role: 'user', content: 'Write a poem about space.' }
const response = await ollama.chat({ model: 'llama3', messages: [message], stream: true })

for await (const part of response) {
  process.stdout.write(part.message.content)
}

---

4. Frontend Integration (React/Next.js)

When calling Ollama from a Frontend app:

1. CORS: Ensure OLLAMA_ORIGINS is set (see Lesson 1).
2. Service: It’s often better to call Ollama from a "Server Action" or an "API Route" in Next.js. This protects your users and allows you to easily swap Ollama for a cloud provider later if needed.

---

5. TypeScript Support

The library is written in TypeScript, so you get full autocompletion for model names, roles (User, Assistant, System), and response types.

---

Key Takeaways

• The ollama-js library works in both Node.js and the Browser.
• It uses Async Generators for clean streaming logic.
• TypeScript definitions ensure type safety in your AI applications.
• Be careful with CORS if calling Ollama directly from a client-side website.