Thinking in AI
Learn to Think About AI, Not Just Use It
Artificial intelligence is evolving faster than most organizations can absorb. New models, frameworks, libraries, architectures, agentic patterns, retrieval techniques, and AI platforms appear continuously. Yet the difficult part is rarely learning what a technology does.
The difficult part is understanding how to think about it.
What problem does it actually solve? Where does it belong in an AI system? What assumptions does it introduce? What trade-offs does it create? When should it be used, and when should it not? How does it interact with knowledge, context, reasoning, tools, agents, applications, security, governance, reliability, and economics?
Thinking in AI is a content series dedicated to answering these questions.
It is designed for AI engineers, architects, technology leaders, and practitioners who want to move beyond feature-level understanding and develop a deeper mental model of how AI technologies and systems work.
From Technology Knowledge to AI Thinking
Knowing a library is not the same as understanding the problem it solves.
Knowing how to implement RAG is not the same as understanding when retrieval is architecturally necessary.
Knowing how to build an agent is not the same as understanding when autonomous execution adds value.
Knowing the capabilities of a foundation model is not the same as understanding how model selection affects architecture, latency, quality, risk, and economics.
Thinking in AI focuses on the reasoning behind these decisions.
Each topic is examined through its underlying concepts, architectural role, implementation considerations, trade-offs, limitations, and relationship to the broader AI system.
The objective is not to memorize technologies. It is to develop the ability to reason about them.
What You Will Find Here
The series explores AI across multiple levels of abstraction, from foundational concepts to enterprise architecture.
AI Concepts
Understand the fundamental ideas that shape modern AI systems, including intelligence, reasoning, context, knowledge, inference, embeddings, retrieval, memory, tool use, agents, and multimodal interaction.
AI Technologies
Examine models, platforms, libraries, frameworks, runtimes, databases, and infrastructure through the problems they solve and the architectural decisions they enable.
AI Architectures
Explore patterns for designing AI systems, including RAG architectures, agentic systems, multi-agent ecosystems, model-routing architectures, AI platforms, and AI-native applications.
AI Frameworks and Libraries
Go beyond API usage to understand the architectural abstractions, execution models, responsibilities, boundaries, and trade-offs introduced by frameworks and libraries.
AI Engineering
Examine the engineering disciplines required to turn AI capabilities into reliable production systems, including evaluation, observability, security, governance, reliability, performance, and cost management.
AI Architecture and Strategy
Connect technical decisions to enterprise architecture, business capabilities, operating models, platform strategy, governance, and long-term evolution.
A Consistent Way of Thinking
Every topic can be approached from multiple perspectives.
What is it?
Establish the concept and its boundaries.
Why does it exist?
Understand the problem or limitation that led to it.
How does it work?
Build a conceptual and architectural mental model.
Where does it fit?
Understand its role within the larger AI system.
What does it replace or complement?
Identify its relationship with existing approaches.
What are the trade-offs?
Examine quality, latency, scalability, complexity, security, reliability, and economics.
When should you use it?
Identify the conditions under which the approach makes architectural sense.
When should you avoid it?
Understand its limitations and failure boundaries.
How does it evolve?
Consider how the concept changes as models, workloads, data, infrastructure, and business requirements evolve.
This approach turns isolated technology knowledge into reusable architectural reasoning.
The Goal
The goal of Thinking in AI is not to make you an expert in every AI technology.
That is neither practical nor necessary.
The goal is to help you develop a mental model that allows you to approach unfamiliar AI technologies with confidence.
When a new model, framework, protocol, database, agent architecture, or AI capability emerges, you should be able to ask the right questions, identify its architectural significance, understand its trade-offs, and determine where it belongs in the system.
That is the difference between knowing an AI technology and thinking in AI.
Explore the Series
The topics covered in this series will continue to evolve with the AI landscape.
Some articles will explain foundational concepts. Others will examine emerging technologies, architectural patterns, engineering practices, or enterprise AI decisions.
The common thread is simple:
Understand the idea. Understand the system. Understand the trade-offs. Then make the architectural decision.
Welcome to Thinking in AI.