The engagement follows four steps. The client describes their machine and what the assistant should do. I design the topology: which channels it lives in (WhatsApp, Telegram, Slack, Discord, Signal or iMessage), which model provider (Claude, GPT, Gemini, OpenRouter or a local model), and where state lives. Then installation, configuration and hardening on their Mac, Windows, Linux or VPS. They receive documentation and an optional working session with the team.