Building Apps with the Gemini API

What's inside

1. Level 1: Google AI Studio and Your First API KeyFree
   Learn what Google AI Studio and the Gemini developer API are, why this course is independent of Google, and how to create an API key and keep it out of your code.
2. Level 2: First Calls with the Gen AI SDKsFree
   Install Google's Gen AI SDK for Python or JavaScript and make a first text-generation call, reading the response object and handling common errors.
3. Level 3: Choosing a Gemini Model: Quality, Speed, Cost
   Compare the current Gemini model families by quality, latency and price, tell stable models from preview ones, and set generation parameters to suit the task.
4. Level 4: System Instructions, Multi-Turn Chat and Streaming
   Build a conversational assistant that uses a system instruction for persona and rules, keeps chat history across turns, and streams tokens to the user as they are generated.
5. Level 5: Images and PDFs as Input
   Send images and PDF documents to Gemini as inline data, and learn the size limits and prompt patterns that give reliable visual and document understanding.
6. Level 6: Audio, Video and the Files API
   Upload larger media such as audio and video through the Files API, reference the uploaded files in prompts, and manage their lifecycle.
7. Level 7: Structured Output with JSON Schemas
   Constrain Gemini's responses to a JSON schema so your application can parse and validate them reliably, using Pydantic or Zod-style type definitions.
8. Level 8: Function Calling: Connecting Your Own Code
   Declare your own functions as tools so Gemini can ask your app to call them, then send the results back to finish the answer.
9. Level 9: Grounding with Google Search and URL Context
   Ground answers in fresh web information with the Google Search tool and in specific pages with the URL context tool, and show citations to users.
10. Level 10: Built-in Code Execution
    Let Gemini write and run Python in Google's sandbox to calculate, analyse data and check its own reasoning, and know where the sandbox's limits are.
11. Level 11: Long Context and Context Caching
    Work with very large inputs in Gemini's long context window, and cut cost and latency on repeated prompts with implicit and explicit context caching.
12. Level 12: Embeddings and a Small Retrieval Example
    Generate embeddings with Gemini's embedding model and build a small retrieval-augmented question-answering example over your own documents.
13. Level 13: Safety Settings and Handling Blocked Output
    Configure Gemini's adjustable safety filters, detect blocked prompts and responses, and design graceful fallbacks for your users.
14. Level 14: Tokens, Pricing, Free Tier and Rate Limits
    Count tokens before sending, estimate cost against current per-model pricing, and design around free-tier and paid rate limits.
15. Level 15: The Live API for Real-Time Voice (Outline)
    Get an outline of how the Live API streams audio over a bidirectional session for low-latency voice agents, which models and features are preview, and when to use it.
16. Level 16: Gemini Image Generation (Outline)
    Get an outline of generating and editing images with Gemini's native image models, including preview status, SynthID watermarking and cost.
17. Level 17: When to Move to Gemini Enterprise Agent Platform
    Decide when to move from the Gemini developer API to Google Cloud's Gemini Enterprise Agent Platform (the successor to Vertex AI), and how the same SDK targets both.
18. Level 18: Deploying a Small Gemini-Backed Service
    Wrap a Gemini feature in a small HTTP service, keep the key secret on the server, and deploy it to a managed platform such as Cloud Run.
19. Level 19: Final Challenge: Ship a Grounded Multimodal Assistant Exam
    Bring the whole course together by building, costing and deploying an assistant that combines multimodal input, structured output, tools, grounding, retrieval and safety.

Access

The first 2 levels are free with a free account. Every level, the podcast edition and the AI tutor come with All Access at £4.99/month or any Creator plan — see pricing.

Start free →