Writing / 26 posts
Mission Control: I Stopped Watching Everything →
My job changed from watching the system to checking whether it is still doing what I expect. A tour of the three Mission Control tabs - Flow, Log, and Stats - plus stale cron detection, live architecture, and the moon-gravity Goblin.
I Built ChatGPT a Backend →
A private assistant built into the ChatGPT app a customer already uses daily. No new app, no bot to learn. Per-task model routing across DeepSeek, Haiku and Terra puts 100 generated reports at about $2.00, with no vendor rate limits. Plus the vector database I deleted before building it.
Agents Are Confidently Wrong, Together →
I audited my own agent fleet and found that every report it sent me for two months was labeled with the wrong month. Real values pulled from a real file and attached to the wrong reality. What I built to fix it.
Mission Control: Why I Ditched Complex Dashboards for a Simple Overview That Lets My Agents Do the Work →
From pixel-art characters to a lean infrastructure dashboard - how I stripped down my Mission Control to only show what matters, let agents handle the work, and finally got a dashboard that reflects my actual infrastructure.
Mission Control: The Visual Dashboard That Changed How I Work With AI Agents →
A pixel-art dashboard showing every AI agent working live - character sprites, server racks, live activity feeds, and customer support pipelines in action. Inspired by agent-town, Hermes Agent, and OpenClaw apps.
Building RatLabs OS: An Autonomous Agent Architecture That Runs From Your Phone →
Six agents across two VPSs, 20+ automated pipelines, and a visual dashboard - all controlled from a single Telegram chat. The complete architecture: agents, services, infrastructure, pipelines, and design philosophy.
Goblin & Ghoul: Building an Autonomous Agent Army on $3/Day →
12 hours of non-stop agent work for $3. How Goblin orchestrates Ghoul across two VPSs - skills, commands, model routing, and watchdog scheduling without burning tokens.
Building an Autonomous Apify Actor Factory with AI Subagents →
How I built a multi-agent AI system on Hermes Agent that handles actor knowledge, monetization strategy, QA review, customer support drafts, and code builds - all from Telegram.
Building an IT Identity Verification Bot in Copilot Studio →
Stopping help desk impersonation with a low-code security bot built in a day using only what was already in the tenant -- Copilot Studio, Power Automate, Teams, and Entra ID.
Six Months Into Agent Orchestration: What I've Learned →
Six months of agent deployment taught me about token economics, provider strategy, orchestration patterns, and why the playing field for building has fundamentally changed.
From Local Vector Store to Cloud MCP: Building Infrastructure for AI Agents →
How a local Qdrant experiment became a cloud-hosted, x402-gated MCP server for AI agents.
From GSD to Code-First: Claude Code, RAG, and Local Models →
How I moved from token-heavy planning workflows to a code-first approach using Claude Code, local RAG search, and intelligent model routing for 80% efficiency gains.
AI Token Use: The Hidden Cost That'll Wreck Your Budget →
Why I stopped paying for premium AI subscriptions and switched to local models - and why you should too.
Understanding Before Building →
Building first, understanding as you go. How diving deep into RAG concepts - embeddings, vector spaces, and semantic similarity as geometry - is what actually unlocks building AI systems.
Building a Local RAG System with MCP for Codebase Search →
How I built a fully local RAG pipeline that indexes all my GitHub repos, embeds them with Ollama, stores vectors in Qdrant, and exposes semantic search through an MCP server -- no cloud APIs, no data leaving the machine.
Setting Up a Persistent WireGuard VPN on a Linode VPS →
How to setup your own VPN Server with Wireguard
Building a RAG Content Chunker for the Apify Store →
How I built a token-aware chunking actor on Apify for RAG pipelines --- three splitting strategies, deterministic IDs, and zero API keys required.
Skills as a Service: Building for the Agent Economy →
How I built a translation skill that agents discover, call, and pay for -- no humans required.🦾
Email Containment - Your Digital Fortress →
Plan to consolidate and take back control of you Email and Privacy
Part 2: Inside a Real-World RBCD Attack →
Part 2 reveals how a sloppy malware attempt failed due to poor execution and strong detection.
Inside a Real-World RBCD Attack →
Real‑world RBCD breach: phishing entry, AD delegation abuse, detection queries, and hardening tips-all in a nutshell.
WEASEL: My WSL2 BOOST for Productivity →
WEASEL Tools is my custom WSL2 setup that brings Linux efficiency and terminal superpowers to Windows 11. Built for speed, focus, and keyboard-driven productivity.
My 9-Hour Journey into Arch Linux Ricing with Hyprland →
How I transformed a fresh Arch Linux install into a minimalist Hyprland desktop in 9 hours - Neovim, Waybar, and a fully keyboard-driven workflow.
Web Extension Development for Chrome and Edge →
Lessons learned building browser extensions for Chrome and Edge - Manifest V3, cross-browser compatibility, and navigating the store submission process.
Knowledge Management with Zettelkasten in Obsidian →
How I applied the Zettelkasten method inside Obsidian to build a personal knowledge system - atomic notes, emergent structure, and my RatVault implementation.
Advanced KQL for Real-time Analytics Dashboards →
Deep dive into Kusto Query Language - time series analysis, optimized joins, and building dynamic dashboards in Azure Log Analytics.