Deploying an AI App to Vercel

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-progx-deploy-vercel-ai-creators

1. A developer is preparing to deploy their AI chatbot to Vercel. They notice their API key variable starts with NEXT_PUBLIC_. What should they do before deploying?

   1. Deploy anyway since Vercel automatically secures all environment variables
   2. Rename the variable to remove the NEXT_PUBLIC_ prefix so the key stays server-side
   3. Move the API key to the client-side JavaScript code
   4. Add more NEXT_PUBLIC_ prefixes to make the key more secure

2. What does Vercel's Fluid Compute feature primarily improve for AI applications?

   1. It keeps functions warm and handles long streaming responses better than classic serverless
   2. It provides free GPU instances for model training
   3. It encrypts all data at rest automatically
   4. It automatically generates the streaming API route for you

3. In the AI SDK architecture, which function is responsible for handling the server-side streaming logic?

   1. useChat
   2. getStaticProps
   3. useState
   4. streamText

4. A developer wants to deploy an AI chatbot to Vercel. How many environment variables are typically required to get streaming working?

   1. Zero - Vercel provides all necessary AI credentials automatically
   2. Three - one for the AI provider, one for the model, and one for the domain
   3. Three - typically the API key, model name, and a maxDuration setting
   4. One - just the API key

5. What is the purpose of the useChat hook in a Next.js AI application?

   1. To configure the Vercel deployment settings
   2. To encrypt the API keys before sending requests
   3. To define the streaming response logic on the server
   4. To manage the client-side state and display streaming messages in real-time

6. What does the maxDuration environment variable control in a Vercel AI deployment?

   1. The number of concurrent users allowed
   2. The maximum size of the AI model files
   3. The maximum number of messages a user can send
   4. The longest time a streaming response can run before timing out

7. Which three components work together to create a streaming AI chatbot, according to the deployment path described?

   1. TypeScript, Docker, and Kubernetes
   2. React, Vercel, and OpenAI
   3. Python, Flask, and AWS
   4. Next.js, AI SDK, and Vercel

8. In the big idea described, where does the AI SDK processing primarily occur?

   1. In a content delivery network
   2. On the Vercel server (or serverless function)
   3. On a separate GPU cluster you must provision
   4. In the browser using WebGPU

9. What happens if you deploy an AI app with a NEXT_PUBLIC_ prefixed API key?

   1. The app works normally and Vercel automatically hides the key
   2. Vercel will automatically rotate the key for security
   3. The key becomes visible in browser developer tools, creating a security vulnerability
   4. The app will run faster because the key is cached

10. What advantage does Fluid Compute provide over classic serverless functions?

    1. It offers built-in rate limiting
    2. It maintains warm instances and handles extended execution times better
    3. It provides cheaper storage for model weights
    4. It automatically scales down to zero to save costs

11. What are the 'three files' mentioned in the lesson that are needed to put an AI app online?

    1. A streaming route file, a UI component, and a configuration file
    2. package.json, tsconfig.json, and next.config.js
    3. README.md, .gitignore, and license.md
    4. index.html, style.css, and app.js

12. Which client-side component from the AI SDK manages the chat interface and streaming display?

    1. streamText
    2. useAI
    3. useChat
    4. ServerAction

13. What is the primary role of Vercel in the three-way architecture described?

    1. It stores the chat history in a database
    2. It generates the AI responses itself
    3. It provides the AI models
    4. It acts as the hosting platform that connects the client to the server-side AI SDK code

14. What does it mean for a function to be 'warm' in the context of serverless computing?

    1. The function has been optimized to use less memory
    2. The function only runs during business hours
    3. The function returns thermal data about the server
    4. The function instance is already running and ready to respond immediately

15. When deploying to Vercel, where should the code that calls the AI API (using streamText) reside?

    1. In API routes or server actions that run on the server
    2. In client-side components that load in the browser
    3. In environment variable files
    4. In a separate Lambda function on AWS