Why it matters
Running several coding agents in parallel multiplies token spend fast. Orchestration and guardrail design determine whether that parallelism pays off or just burns budget on redundant work.
The tokenmaxxing angle
Persistent sessions and cloud sandboxes for multi-agent coding are a token-cost decision in disguise: keeping context alive across agents versus re-priming from scratch changes the bill a lot.
From the organizers
Held at the Databricks San Francisco office and organized by the Databricks Community with hosts including Jules Damji and Daniel Liden, featuring a panel with DoorDash on agent infrastructure.