Reflections
Digital solution design, AI and software architecture — case studies.
When a Bug Resists AI Agents
A firsthand account of an integration issue where more context did not help AI agents converge on the right solution.
Recent posts
Why does using AI still feel uncomfortable?
A friend spotted phrasing typical of generative AI in one of my texts. My reaction, for a brief moment, says a lot about how we still view people who use these tools.
From "snail mail" to "snail brain"
What if the words we use today already reveal the place AI will take tomorrow in how we think?
The day my way of working with AI changed
The tools didn't change. How I assembled them did. The story of a shift that transformed how I collaborate with AI agents.
All posts
When a Bug Resists AI Agents
A firsthand account of an integration issue where more context did not help AI agents converge on the right solution.
Why does using AI still feel uncomfortable?
A friend spotted phrasing typical of generative AI in one of my texts. My reaction, for a brief moment, says a lot about how we still view people who use these tools.
From "snail mail" to "snail brain"
What if the words we use today already reveal the place AI will take tomorrow in how we think?
The day my way of working with AI changed
The tools didn't change. How I assembled them did. The story of a shift that transformed how I collaborate with AI agents.
ChatGPT Codex Connector: unable to add a GitHub organization (postmortem)
Postmortem on an issue hit during the transition to ChatGPT Codex's GitHub connector, and the workaround that got me unblocked.
AI assistants are no longer chatbots: what MCP tunnels made me realize
A field report from an architect's perspective: discovering the tunnels OpenAI introduced to connect ChatGPT to self-hosted MCP servers, and the shift in thinking it triggered about where AI platforms are headed.
The day my AI account's security became a selection criterion
A field report on digital trust: how a long-standing habit of securing my accounts with security keys led me, one Sunday in July 2026, to look at AI assistant security differently.
Developers barely type the code anymore — but they still own the outcome
AI has erased the technical friction. The difficulty no longer lives in the code but in the clarity of intent — and the responsibility has remained entirely yours.
GitHub Copilot Cut My Momentum: When Quotas Kill Innovation
Critical analysis of GitHub Copilot's quota policy and how it blocks founders and small teams. Why local LLMs are becoming relevant again.
Multi-layer Architecture with Docker: Isolation over Technology Debates
Learn how to isolate your application layers with Docker to avoid technology lock-in and maintain your projects long-term.
MCP + Sentry & GlitchTip: giving context to your AI agents
A hands-on account of integrating Sentry and GlitchTip MCP servers into an AI workflow — why, how, and what you learn along the way.
I'm Not a Person, I'm a Constellation of AI
Why I stopped looking for the best LLM and started building a workspace made of several AIs instead.
The LLM, a Mirror of Our Identity
Switching LLMs made me realize I wasn't just changing tools. That experience led me to reflect on how artificial intelligence gradually becomes an extension of the way we think.
Why GPG became a reflex in my digital daily life
A field report on five years of using GPG to sign, encrypt, and preserve the integrity of personal documents. A reflection on digital hygiene rather than a technical tutorial.