Code While You Take a Breath

A few months ago, the author began feeling like they were falling behind in AI despite having developed with AI for nearly four years. They had been using AI to learn technologies, solve problems, write code, review solutions, organize ideas, and speed up various aspects of their work long before the latest wave of AI hype. However, an increasing number of new concepts and tools, such as RAG, embeddings, vector search, agents, tool calling, MCP, LangChain, LangGraph, and various open-source models and coding tools, seemed to appear daily.

The author also noticed a surge in local-model alternatives that claimed to eliminate the need for subscriptions. Amidst this flurry of new developments, software development experienced a few "dies" where developers felt disheartened, but always seemed to recover. The author decided to take a break from the AI hype after about ten months of feeling overwhelmed.

Instead of ignoring the developments, they sought to understand the reality of the situation. They found that looking at job postings mentioning AI, writing down frequently occurring terms, and asking AI to help organize them allowed them to gain a higher-level view of the technology. This approach helped them identify what was genuinely new, what was an application of existing concepts, what was likely to become integral to their work, and what was receiving undue attention due to marketing efforts.

Once they started grasping these concepts, previously intimidating terms like "embeddings" became more familiar. While acknowledging the potential importance of AI, the author realized that the hype was less critical to their understanding and concerns as a developer.