Search papers, labs, and topics across Lattice.
This paper challenges the conventional approach of using LLMs for proactive agent triggering by proposing a more efficient temporal-graph-learning (TGL) model that operates directly on structured event streams. The TGL model predicts per-event trigger probabilities and per-entity routing scores, bypassing the need to convert structured data to text for LLM processing. Results show TGL significantly improves F1 scores across various backbones (mean +16.7) and achieves substantial speedups (4-83x) compared to LLM-based triggers, with a smaller memory footprint.
Ditching the always-on LLM for proactive agents can boost speed by up to 83x and improve F1 scores by 16.7, simply by processing structured event data directly with a temporal graph learning model.
Proactive agents read user activity as text and call an LLM on every event to decide whether to act. But user activity is not natively text: it is a structured event stream of (actor, verb, object, timestamp) tuples that the operating system already maintains in graph form. Rendering the structure as text and asking an LLM to recover it is a round-trip the system never had to take. We treat the always-on signal as graph updates rather than text and use a small temporal-graph-learning (TGL) model as the encoder: one forward pass yields a per-event trigger probability and a per-entity routing score, and only the downstream agent (turning a small structured handoff into a fluent user-facing sentence) is an LLM call, invoked only when the trigger fires. TGL improves F1 on each of 14 backbones (mean +16.7, up to +46.0); in trigger-architecture comparisons, one TGL checkpoint gives the strongest trigger AUCs and the most stable deployed threshold. It runs at 11.13 ms per event on a GPU server and 13.99 ms on a consumer laptop, approximately 4--7x and 12--83x faster than every single-forward LLM-as-trigger configuration tested in each regime, with an approximately 220 MiB BF16 resident footprint deployable on-device alongside the privacy-sensitive activity stream it consumes.
