★ Cost-savings live — real Slime Gateway usage (content-free, aggregates only)
cut vs naive (cumulative)
–
vs hurling everything at the cloud
saved (cumulative)
–
actually paid
–
questions
–
loading…
These are aggregates of real in-house AI usage (cost, tokens, Claude rate, savings only — no conversation content, no document names). Full chart: cost-down dashboard.
① Pick a setup — three ways to use local AI
② Route — here's how your question flows
D · Direct
Stable past answers, calculator, cache replay. No LLM ($0).
→
μ · Damp
Watches stability via norm. Holds back drift, stops runaway output.
→
R · Residual
Only what truly needs generation goes on.
→
☁ External AI
Claude / Gemini, etc.
③ The materials — all open (zero internal data)
④ Ask it something
🟡 for laughs
Type a question below and hit Send. The D→μ→R route plays out, then it answers with a joke. Real answers live on the service page (only the real one bills).