The Persistent Context Hypothesis

Building a Session-Based Agent Workflow System for Complex Software Development How we transformed chaotic multi-conversation development into a systematic, persistent workflow using specialised AI agents. The Problem: Development Context Loss Every developer knows this frustration: you are deep into a complex feature implementation across multiple conversations with AI assistants, and suddenly the thread is lost. Previous decisions are forgotten, architectural choices are second‑guessed, and work goes in circles. We faced this exact challenge while building a Laravel application with Vue....

August 15, 2025 · 8 min · Marc Hampson

The Multi-Agent Conundrum

The Multi-Agent Conundrum Multi-agent systems are growing in popularity as a way to improve AI task performance by dividing responsibilities between specialised roles (agents). In this tutorial, you’ll build a web-based assistant using CrewAI and OpenAI, wrapped in a Streamlit interface. You’ll compare single-agent vs multi-agent outputs, measure token usage and cost, and generate markdown reports for easy export. This post is designed to be practical and modular—ideal for anyone experimenting with LLM-powered workflows....

April 21, 2025 · 8 min · Marc Hampson