The Personal Assistant We'll Build › How This Book Teaches

Every chapter follows a similar rhythm:

1. We introduce a problem or limitation with your current assistant
2. We explain the concept that will solve it, using analogies and examples
3. We implement the solution with working code you can run
4. We test it to see the improvement
5. We discuss trade-offs so you understand when to use this approach

The code examples are deliberately simple. We're not trying to build production-grade software in every snippet. We're trying to make concepts crystal clear. Once you understand the principles, you can apply them to more sophisticated implementations.

For intermediate readers: While we keep examples simple for clarity, we'll discuss the trade-offs of different approaches throughout. When there are multiple ways to solve a problem, we'll explain why we chose one over another, what the alternatives are, and when you might prefer a different approach. The goal is to give you both working code and the judgment to adapt it to your specific needs.

You'll also notice we use different AI providers for different examples. We'll use OpenAI (ChatGPT) for basic prompting and text generation, Claude Agent SDK for agent loops and multi-step workflows with tools, and Gemini for multimodal examples and certain integrations. This isn't about picking favorites. It's about showing you the landscape and helping you understand which tools work best for which jobs.

Each provider has strengths. OpenAI excels at general text generation and has the most mature ecosystem. Claude's SDK makes building complex agent workflows easier with built-in tool orchestration. Gemini shines with images and video while offering tight integration with Google's services.

The choice of provider often involves trade-offs. OpenAI's models are widely adopted and well-documented, but can be more expensive for high-volume applications. Claude's agent-focused SDK reduces boilerplate code for complex workflows, but locks you into their ecosystem. Gemini offers competitive pricing and multimodal capabilities, but has a smaller community and fewer third-party integrations. In production systems, you might even use multiple providers (OpenAI for user-facing chat, Claude for complex agent tasks, and Gemini for image processing), choosing the best tool for each job rather than committing to a single vendor.