MODULE 01 · INCLUDED IN FOUNDATIONS
Architecture: how the whole thing works
What this module is
Every AI streamer is one loop running forever. Once you can draw it from memory, every later module is just filling in a box.
What you'll build
- The eight-step loop diagram (chat in → selector → brain → safety → voice → body → stage → chat out)
- A latency budget for each step
- Your choice between assembling an open-source framework or building the loop yourself
Lessons
- 01The loop
- 02The latency budget (under 3 seconds end to end)
- 03Your two build options: assemble (Open-LLM-VTuber, AIRI) or build (five Python files)
Checkpoint
Draw the loop from memory on paper, with the latency target next to each arrow.