AI Platforms for Developers

When you outgrow direct API keys — or your org standardizes on a cloud — you integrate through a platform layer.

Platform Comparison

For coding assistants in the IDE, you typically use Copilot, Cursor, or Claude Code — not these platforms directly. Platforms matter for product features you build.

Direct API vs Platform

flowchart TD
    start[Need LLM in product] --> cloud{Existing cloud contract?}
    cloud -->|No / startup| direct[Direct OpenAI or Anthropic API]
    cloud -->|AWS| bedrock[AWS Bedrock]
    cloud -->|GCP| vertex[Vertex AI]
    cloud -->|Azure + AD| azure[Azure OpenAI]
    direct --> sdk[Vercel AI SDK route handler]
    bedrock --> sdk
    vertex --> sdk
    azure --> sdk

Official comparisons: AWS Bedrock, Vertex AI, Azure OpenAI Service.

Integration Pattern

Regardless of vendor, the shape is the same:

1. Authenticate with IAM / service account (not long-lived keys in code)
2. Call unified or provider-specific SDK
3. Log usage for cost allocation
4. Apply VPC/private endpoints if data residency requires

Prefer higher-level SDKs (Vercel AI SDK) when they support your provider.

AWS Bedrock

import { BedrockRuntimeClient, InvokeModelCommand } from "@aws-sdk/client-bedrock-runtime";

const client = new BedrockRuntimeClient({ region: "us-east-1" });

const response = await client.send(
  new InvokeModelCommand({
    modelId: "anthropic.claude-3-5-sonnet-20241022-v2:0",
    contentType: "application/json",
    accept: "application/json",
    body: JSON.stringify({
      anthropic_version: "bedrock-2023-05-31",
      max_tokens: 1024,
      messages: [{ role: "user", content: "Summarize this ticket…" }],
    }),
  })
);

Docs: Bedrock InvokeModel API.

Google Vertex AI (Gemini)

import { VertexAI } from "@google-cloud/vertexai";

const vertex = new VertexAI({ project: process.env.GCP_PROJECT!, location: "us-central1" });
const model = vertex.getGenerativeModel({ model: "gemini-2.0-flash" });

const result = await model.generateContent({
  contents: [{ role: "user", parts: [{ text: "Summarize this ticket…" }] }],
});

Docs: Vertex AI Node.js SDK.

Azure OpenAI

import { AzureOpenAI } from "openai";

const client = new AzureOpenAI({
  endpoint: process.env.AZURE_OPENAI_ENDPOINT!,
  apiKey: process.env.AZURE_OPENAI_API_KEY!,
  apiVersion: "2024-08-01-preview",
});

const response = await client.chat.completions.create({
  model: "gpt-4o", // deployment name in Azure
  messages: [{ role: "user", content: "Summarize this ticket…" }],
});

Docs: Azure OpenAI quickstart.

Local and Self-Hosted

For air-gapped or privacy-sensitive dev:

• Ollama, llama.cpp, vLLM — run open weights locally
• Trade quality and speed for data control
• Useful for experimentation; rarely matches cloud frontier models for hard coding tasks

Docs: Ollama, vLLM.

For Teams

Route procurement questions through Choosing a Tool security questionnaire.

Related

• LLM APIs and Tool Calling
• Streaming Chat Tutorial
• AI in Web Apps
• Cost, Latency, and Tokens
• LLM API Route Handler cheat sheet