Vojtech@vojtech

Suivre

Loop vs graph agents explained

Your agent keeps forgetting everything between runs, and that's the real bottleneck. A loop-based setup gives the agent a goal and lets it figure things out, which works for fresh problems but falls apart when you repeat the same task. A graph instead structures the work as connected facts the agent moves through, so knowledge builds up instead of resetting to zero. It's more effort upfront, but you never start from nothing again. The loop forgets, the graph accumulates. That's the actual unlock: organize what you know as relationships rather than loose files, and a smaller model moving through a clean graph beats a bigger one searching through raw text. Microsoft, Stanford, and MIT all confirmed this independently. Kimi K3 fits the role: a million tokens of context can handle an entire reasoning path at once, at a third of the cost of frontier models. When a loop stops teaching you anything new, freeze it into a graph. So which one is your stack actually running?

3w

Vojtech@vojtech

Suivre

Loop vs graph agents explained

Your agent keeps forgetting everything between runs, and that's the real bottleneck. A loop-based setup gives the agent a goal and lets it figure things out, which works for fresh problems but falls apart when you repeat the same task. A graph instead structures the work as connected facts the agent moves through, so knowledge builds up instead of resetting to zero. It's more effort upfront, but you never start from nothing again. The loop forgets, the graph accumulates. That's the actual unlock: organize what you know as relationships rather than loose files, and a smaller model moving through a clean graph beats a bigger one searching through raw text. Microsoft, Stanford, and MIT all confirmed this independently. Kimi K3 fits the role: a million tokens of context can handle an entire reasoning path at once, at a third of the cost of frontier models. When a loop stops teaching you anything new, freeze it into a graph. So which one is your stack actually running?

3w