Really enjoyed this! The point that agentic AI needs boundaries, verification and the right level of autonomy really resonated with me.
But I’m curious about one thing: as we add stronger testing, review and rollback loops around agents, do you think the 40% failure rate would drop significantly, or are we simply moving the failure point elsewhere?
Swati Rawat
Really enjoyed this! The point that agentic AI needs boundaries, verification and the right level of autonomy really resonated with me.
But I’m curious about one thing: as we add stronger testing, review and rollback loops around agents, do you think the 40% failure rate would drop significantly, or are we simply moving the failure point elsewhere?