AfferLab is a local-first AI workspace designed to make LLM systems more controllable and programmable. Instead of treating AI as a simple prompt-to-response interface, AfferLab introduces a strategy-driven workflow where users can shape how context, memory, and behaviour are organised during interaction.

The project explores a more structured approach to working with AI systems: not only chatting with models, but also experimenting with reusable strategies, controllable context assembly, and extensible runtime behaviour. It is built as a desktop application with a strong focus on local-first architecture, modular design, and future expansion toward more advanced agent workflows.

I built AfferLab as an independent long-term project to explore the intersection of AI product design, system architecture, and developer tooling.

• Electron
• React
• TypeScript
• SQLite
• Local-first AI architecture