All labs
Lab 31
Scaling AI Systems

Agentic AI — Same Pressures, New System

The bridge the AI crowd asked for. Take one agentic system and switch on each architecture idea you learned — stateless agents + Redis memory, an event-driven agent mesh, a semantic cache for repeated LLM calls, a sharded & replicated vector store, and autoscaling under load — and watch every concept land in an AI system.

The bridge you asked for: take a customer-support AI agent and switch on each architecture idea from the workshop. Watch the same concepts — statelessness, events, caching, sharding, autoscaling — fix an AI system. Start with everything off and flip them on one by one.
Cost / 1k queries
$120
p95 latency
4.2s
Retrieval
900 ms
Memory
lost
On a spike
drops
👤 Users
conversations
📡 Agent mesh
sync, blocks
🤖 Agent workers
fixed pool
⚡ Redis memory
in-instance
🪣 Semantic cache
every call hits LLM
🧠 LLM
the easy part
🗄️ Vector store
single index
📈 Autoscaler
off
🛠️ Tools / APIs
actions
Flip the switches and watch the metrics move. Each one is the same idea from a scaling or event-driven lab — now keeping an AI agent reliable and affordable. Turn all five on to see the full harness.
What just happened