No hype. No empty buzzwords. Just the map that matters.
An educational, interactive, visually strong website that teaches what you actually need to understand to work in engineering in a world of agents, using accessible, human language.
The market already treats AI as a normal part of software work. Adoption is massive, perceived productivity has increased, but trust has not kept up. The gap is not between using AI and not using it; it is between using it poorly and using it well.
This site exists to close that gap: to organize the chaos, translate the terminology, and show that AI-native engineering is about context, judgment, validation, and flow, not about "asking for code."
- QA, Product Managers, Product Designers, Tech Recruiters, founders, and technology leaders
- Junior and mid-level developers who still mix up tools, models, agents, CLIs, and workflows
- Any curious person who wants a clear mental map of the topic
The home page starts with a compact foundations primer, then the site continues through 3 progressive sessions:
| # | Session | Route | What It Teaches |
|---|---|---|---|
| 1 | Glossary | /sessions/glossario/ |
LLM, model, token, prompt, agent, harness, MCP, and more |
| 2 | Tools and Models | /sessions/ferramentas/ |
IDEs, CLIs, cloud agents, product vs model, task-fit choices |
| 3 | SDD and Harness Design | /sessions/maturidade/ |
Spec-driven contracts, agent harnesses, validation, governance |
Each session follows a fixed template: 30-second summary, main explanation, why it matters, real example when useful, where it breaks, takeaway, and references. Interactive blocks are included only when they clearly improve learning.
- Teach without idolizing tools, showing categories and patterns instead
- Simple explanation first, technical depth second
- Examples from QA, PM, and product, not only hardcore development
- Every interaction must explain better, organize better, or improve retention
- Acronyms are explained before any deep dive
Contributions are welcome — whether it's suggesting new content, fixing inaccuracies, or improving clarity.
- Report issues or suggest content: open an issue
- Start a conversation: join the discussions
- Submit changes: fork the repo, make your edits, and open a pull request
- Keep PRs reviewable: code and config changes should stay under 400 changed lines per pull request (insertions + deletions), measured against the PR base branch and excluding
src/content/. Editorial-only PRs undersrc/content/are exempt. See AGENTS.md → PR discipline for the full policy.