How-ToAI AgentsOctober 26, 2026

Arize AI's Xander Song demos Harbor for testing agents on real tasks

Read original source →youtube.com

Harbor is an open source framework that evaluates agents on realistic tasks using real data inside isolated Docker sandboxes. It structures runs around tasks, verifiers and rewards, trials and jobs, plus oracles that validate the setup.

People · Xander Song

1 source

More stories today

Open the live feed