Agentic systems
Hermes Agent, MCP, memory, tool use and the question: how do you make an agent genuinely helpful rather than merely impressive?
Personal wiki · Aberdeen, Scotland
I translate technology for humans, build things to understand them, and keep notes so the next person starts further ahead.
I spent ten years at Apple Retail in Madrid, moving through Specialist, Technical Specialist, Genius and finally In-Store Experience Lead. My work lived at the boundary: diagnose the problem, understand the person, explain the trade-off, and help the system work better next time.
Since moving to Aberdeen with Marina in 2024, I have been retraining in Artificial Intelligence and Full Stack development. I am not pretending the pivot is finished. The unfinished part is the point: I learn by building, documenting, breaking things and trying again.
Hermes Agent, MCP, memory, tool use and the question: how do you make an agent genuinely helpful rather than merely impressive?
Small tools, written explanations and honest experiments. The goal is a body of work that shows how I think, not just a list of claims.
MDIB is forming as a practical agency for deploying useful AI workflows, with scope and support made explicit rather than hand-waved.
Tools for personal knowledge management, durable notes and change tracking around agentic work.
A structured collection of models for thinking, decisions and clearer work.
A small, tested desktop utility — evidence that useful software can remain simple.
Import, enrich and inspect video knowledge without losing the source trail.