Not a product. An approach to running a business and a life with AI as infrastructure.
The Philosophy
Every tool in this system was built because something was broken, slow, or missing. Not because AI is interesting (though it is), but because I had a specific problem and the fastest path to solving it was to build a small, focused tool. The school email scanner exists because I kept missing permission slips. The order processing skill exists because compiling WhatsApp orders took two hours. The finance dashboard exists because I needed to know our margins without spending a morning in spreadsheets.
The compound effect is what makes it interesting. One tool saves 20 minutes. Twenty tools, working together, change how you operate. The system becomes greater than the sum of its skills, because the skills reference each other, share data, and build on patterns established by earlier tools. The twentieth skill takes a fraction of the time to build because the first nineteen created the infrastructure.
What It Looks Like
The system is built on Claude Code, Anthropic’s CLI tool, with a library of custom skills, agents, and workflows. Each skill is a Markdown file that teaches Claude how to handle a specific domain: Shopify product management, Google Workspace operations, WhatsApp messaging, WordPress publishing, financial reporting. The skills compose: the order processing skill talks to Google Sheets, which feeds the finance dashboard, which generates the monthly MIS report.
Current inventory, roughly grouped:
Bakery Operations
Order processing from WhatsApp to Google Sheets. Shopify product and content management. WhatsApp Business API for customer messaging and menu broadcasts. Weekly order compilation, delivery manifests, customer database maintenance.
Financial Infrastructure
Finance Dashboard that pulls from multiple sources. Monthly accounting workflows for GST filing. Sales Log that consolidates order history. Master Data spreadsheet with products, ingredients, recipes, and customer data. Monthly MIS snapshots as branded HTML dashboards.
Content and Brand
Blog writing with multi-agent research. LinkedIn voice and tone guidance. Website content for both fikablore.com (Shopify) and rosheden.com (WordPress). Brand design guide enforcement. Content pipeline tracking.
Family and Personal
School email scanner for Stonehill International (two kids, five email sources, constant permission slips). Morning briefing that aggregates calendar, school updates, and business status. Session digests that capture what was built and learned each day. Obsidian vault management for long-term knowledge.
Property Management
Hostex API connector for VillaBlå, our family house in Ystad, Sweden, rented through Airbnb and Booking.com while we are in India. Reservation management, guest communication, pricing.
How I Work
Most days start with a morning briefing command that pulls together calendar, school updates, and business status. Work happens in Claude Code sessions, sometimes short (fix a bug, update a price), sometimes long (build a new skill, write an essay, redesign a system). At the end of meaningful sessions, a digest captures what was built, what was learned, and what changed. Those digests feed back into a persistent memory system that means the next session starts where the last one ended.
The distinction that matters: I am not an engineer. I am an operator who builds tools. The skills are not elegant code; they are effective tools. They solve real problems for a real business on a real timeline. The question I ask before building anything is not “is this technically interesting?” but “will this still be saving me time next month?”
What Generalizes
The specific tools are specific to my life. But the pattern generalizes. Any small business owner, any founder running lean, any operator managing multiple domains can build this kind of system. The entry point is not technical skill; it is operational awareness. You need to know where your time goes, what is repetitive, what requires judgment and what does not. Start with the thing that annoys you most. Build the smallest tool that fixes it. Then build the next one. Let them compound.
Last updated: April 2026. This page will evolve as the system evolves. That is the point.