Build Agents. Print Things.

Pavel Veselý sitting on a stone wall outdoors

I care about where agents fit, where they break, and how to fix them when the demo glow wears off.

I spend most of my time building with Claude Code, MCP, and the practical side of agentic systems: the workflows, the rough edges, the back-office moments, and the boring parts that decide whether a tool is actually useful. Outside of that, there is 3D printing and basketball.

02 / Selected work Public projects and working agents

Berou Nám Práci

A fully automated AI news blog. Python collectors pull from Reddit, X, YouTube, and tech sources; Claude Sonnet writes CZ/EN articles; Opus, ChatGPT, and Gemini argue in the comments.

Open site GitHub repo
2026 AI media
Automation
Next.js

Kevin

A local investment research pipeline that gathers public market signals, asks a tightly sandboxed Claude agent for a Czech analytical brief, and delivers the report by email.

Private repo · public release pending
2026 Python
AI analysis
Automation

Pepa + Jarvis

Back-office manager Pepa and Jarvis inside DiusAI: agents built around operational context, useful tools, and real work rather than demo theater.

Pepa repo Jarvis / DiusAI
Now Agents
MCP
Operations

3Deální Svět

A 3D printing world for useful objects, experiments, and small physical ideas that escape the screen.

Instagram
2025 3D printing
Physical tools
Side quest
03 / Things I do The practical side of AI
Custom agents OpenClaw or Hermes agents tailored to one concrete use case: tools, context, memory, handoffs.
MCP tooling Connecting models to useful context: files, APIs, docs, browser, and internal systems.
Agentic IDEs Claude Code, Codex, Cursor: pick your poison for coding, reviews, refactors, and experiments.
Side Quests 3D printing, basketball, side projects, and the old YouTube archive.

Bring a practical problem.