Ever wondered how "much" internet we have? While (as usually) the answer depends on...

Satisfying my lack of patience or how we shaved 75% of our model latency and saved up to 98% on cost while doing so

A peek under the hood on how we achieved 3-4x latency improvements in our AI customer support system while maintaining acceptable quality and reducing costs by orders of magnitude.

ADK Adventures: from simple agents to multi-agent mayhem (and back again?)

Is this over-engineered? Probably yes. Did it teach me ADK? Also probably yes.