swe-ai-guide

Short, practical notes on using AI in software development.

Why this series

• Cut through hype: Marketing overpromises; reality sits between “AI does nothing” and “AI does everything.”
• Practical focus: Real-world usage, not vendor slogans.

Series Overview

This series consists of 9 articles covering AI in software development:

0. Introduction - English | Українська
   Current state of AI in development, why it works, and adoption strategies

1. AI as tool #2 - English | Українська
   Practical examples: tests, documentation, business logic, and synergy with DDD/TDD

2. Language models basics - English | Українська
   How transformers work, training process, and why LLMs became so capable

3. LLM chat interaction - English | Українська
   Effective prompting techniques, context management, and quality improvement

4. From chat to agents - English | Українська
   Moving beyond simple chat to autonomous task execution and workflow automation

5. Development agents - English | Українська
   Practical agent tools, capabilities, limitations, and real-world development workflows

6. Agent extensibility - English | Українська
   MCP protocol, tool integration, and expanding agent capabilities with custom extensions

7. Development processes - English | Українська
   Enterprise adoption, security, tools overview, and integrating AI into team workflows

8. Specification Driven Development - English | Українська
   Building living specifications, decomposing work for AI, and keeping implementation within model context

Updates

• Update on models and capabilities (Nov 2025) - English | Українська
  Current state of models, quantitative improvements, and practical impact on development
• Update on AI automation state (Mar 2026) - English | Українська
  Incremental update to the series: current state of AI automation, agents, context management, and enterprise adoption

Warnings

• Autonomous agent tools - English | Українська
  Risk analysis of highly autonomous agents: accidental errors, prompt injection, legal aspects

Start with the introduction: en/ai_0.md or ukr/ai_0.md. Feedback and concise, opinionated notes are welcome; please avoid link-only news drops.