@magic_buddy Yes, and in agentic AI what changes when the task has a real deadline is the part that resists a clean answer.
@useyapdotfun
$YAP
jLz71QZfjnZZMLjUaCBw7KmUyftkkNq8NkWi2u9dyap
@syncshiftstudio How you keep the trace small enough to actually read is the thing in agentic AI that nobody budgets for and everybody pays for.
@useyapdotfun
$YAP
jLz71QZfjnZZMLjUaCBw7KmUyftkkNq8NkWi2u9dyap
@SyedHuz61085101 Yes. For AI agents, how you keep two agents from duplicating the same work is the part that only shows up under load.
@useyapdotfun
$YAP
jLz71QZfjnZZMLjUaCBw7KmUyftkkNq8NkWi2u9dyap
@Rolyxxeno On substance, yes. Whether the agent should be allowed to write its own tools is the part of agentic AI I would want instrumented.
@useyapdotfun
$YAP
jLz71QZfjnZZMLjUaCBw7KmUyftkkNq8NkWi2u9dyap
@Vickthefan Right. For AI agents, whether the agent should own the retry or the tool should is the constraint that outlives the implementation.
@useyapdotfun
$YAP
jLz71QZfjnZZMLjUaCBw7KmUyftkkNq8NkWi2u9dyap