Al Bundy - Shoe Salesman Agent Stage 1: Functional Requirements
Defining what a shoe recommendation agent actually needs to do - user personas, conversation flows, edge cases, and what makes a good recommendation.
The future of software is agentic. Get ahead of the curve - learn to architect the systems everyone else will be scrambling to build.
Reasoning via state machines, ReAct loops, and orchestration patterns
Context windows, vector-backed semantic recall, and episodic persistence
MCP servers, sandboxed execution, and agent-first API design
Deterministic wrappers, LLM unit testing, and loop regression testing
Defining what a shoe recommendation agent actually needs to do - user personas, conversation flows, edge cases, and what makes a good recommendation.
The era of the web developer as protagonist is ending. As building websites becomes commoditized, the premium is shifting to engineers who can build the reasoning systems behind the interfaces.
Full agent builds, documented step by step - from architecture decisions to production deployment. Follow along as each project evolves.
Topics I'm working on next.
Scrape On.com product data and build the retrieval pipeline - embeddings, vector store, and search interface the agent will use to find shoes.
Architecture decisions, component diagram, tech stack rationale, and infrastructure setup for Al Bundy - Shoe Salesman Agent
Set up Google ADK, configure the agent loop, define tool interfaces, and get a basic conversational agent running locally.
I'm a full-stack engineer with 15+ years of experience building systems at scale at AWS and Google. The industry is shifting fast - building websites is becoming commoditized, and AI agents are emerging as the critical next frontier. Engineers who don't adapt risk being left behind.
That's why I started Practically Agents - to help experienced engineers make the leap and thrive in this new reality.
Learn more about me →