Earlier this month I offered Universidad San Sebastián a workshop series — nine sessions on agentic engineering, ad honorem, for final-year students, professors, and engineers who'd already used AI but hadn't gone past "Claude wrote me an app." Given how fast the AI landscape is moving,
universities — and even companies — are struggling to keep up, and I wanted to help.

Then I sat down to actually plan it.

The artifact

I used Claude Design — a new tool from Anthropic Labs — to draft the session map. I kept the default Claude-themed palette on purpose: the workshop is about Claude Code, so the aesthetic landed on-brand by accident.

Nine sessions on a real legacy Java codebase (rulyone/dotachile), ending in a Capstone where a real feature ships end-to-end in sixty minutes — skill + agent + hook + MCP, integrated.

The pivot

Mid-planning I read my own audience description back to myself:

Engineers who already use AI and want real workflows & best practices — not another introductory tutorial.

That's not a syllabus filter. That's a market description.

Companies adopt Claude Code in a day. The capabilities that actually create leverage — hooks, skills, agents, RAG, MCP, custom team workflows — take weeks or months to configure and implement.

And the people best positioned to learn it deeply are also the people delivering this quarter's roadmap.

I can help close that gap and keep it closed as the platform evolves — introducing capabilities to your strongest engineers gradually, so they stay focused on the roadmap and grow into full ownership.