Building AI Agents with the Vercel AI SDK

Appearance

Getting Started ​

Welcome to Building AI Agents with the Vercel AI SDK. This guide gets the repo running on your machine so you can follow along with every lesson by running the real code.

1. Clone the repo ​

bash
git clone https://github.com/blendsdk/building-agents-vercel-ai-sdk.git
cd building-agents-vercel-ai-sdk

2. Install dependencies ​

This project uses Yarn 1.x:

bash
yarn install

3. Add your OpenAI API key ​

The tutorials call OpenAI through the Vercel AI SDK. Create a .env file in the project root:

bash
echo "OPENAI_API_KEY=sk-..." > .env

The .env file is git-ignored — never commit your key.

4. Run a tutorial ​

Each lesson maps to a numbered source file under src/agents/. Run one by its source file number with yarn dev:

bash
yarn dev 1     # runs src/agents/01-oneshot.ts
yarn dev 5     # runs src/agents/05-rag.ts

5. Run the tests ​

The deterministic logic (calculator parser, RAG ranking, eval scorer) is covered by a fast, network-free test suite:

bash
yarn test

How this course works ​

  • Read a lesson on this site — every code sample is imported directly from the real source files, so what you read is exactly what runs.
  • Run the matching file with yarn dev <n> and watch the agent work.
  • Experiment — tweak prompts, tools, and schemas, then re-run.

When you're ready, the course begins with Part I — Foundations (added as the curriculum lands).

Next: make sure you meet the Prerequisites.