frontend-angular-ai

name	frontend-angular-ai
description	AI integration patterns for Angular applications using Genkit, Firebase AI Logic, or Gemini API. Includes streaming responses, tool calling, and RAG patterns. Use when building AI-powered features in Angular apps.
kb-sources	["wiki/software-engineering/angular-ai-tools"]
updated	"2026-05-21T00:00:00.000Z"

Angular AI Integration

Patterns for integrating AI capabilities into Angular 21+ applications.

Integration Approaches

Approach	Where	Best For
Genkit	Server-side	Multi-model, complex flows, tool calling
Firebase AI Logic	Client-side	Firebase apps, quota management
Gemini API	Direct	Maximum flexibility, multimodal

Quick Decision

Need server-side processing? → Genkit
Using Firebase already? → Firebase AI Logic
Need image/audio analysis? → Gemini API (multimodal)
Building chatbot with history? → Genkit (server-side state)
Simple text completion? → Firebase AI Logic

Genkit Overview (Recommended)

Google's open-source toolkit for production AI applications.

// Server: Express + Genkit
import { genkit } from 'genkit';
import { googleAI, gemini20Flash } from '@genkit-ai/googleai';

const ai = genkit({ plugins: [googleAI()] });

app.post('/api/chat', async (req, res) => {
  const { message, history } = req.body;

  const response = await ai.generate({
    model: gemini20Flash,
    messages: [...history, { role: 'user', content: message }]
  });

  res.json({ reply: response.text });
});

Key Patterns

Streaming Responses

// Use Resource API for signal-based streaming
streamingResponse = resource({
  stream: async () => {
    const data = signal<{ value: string }>({ value: '' });
    // Stream handling logic
    return data;
  }
});

Tool Calling

// Define tools for LLM to call
const tools = [
  {
    name: 'searchProducts',
    description: 'Search product catalog',
    parameters: { query: 'string' }
  }
];

Temperature Control

Temperature	Use Case
0.0 - 0.3	Factual, deterministic
0.4 - 0.7	Balanced creativity
0.8 - 1.0	Creative, varied

Security Patterns

Pattern	Description
API Key Protection	Keys stored server-side only, never exposed to client
Output Sanitization	AI responses sanitized before DOM rendering
Rate Limiting	Request throttling to prevent quota exhaustion
Server Proxy	All LLM API calls routed through backend proxy

See reference.md for detailed patterns and examples.md for implementations.