Prospective Intention Store for Small-Model Agents
Making Prospective Memory SLM-Shaped: Typed Intention Stores for Small-Model Agents
PIS enables small models to outperform large models on prospective memory tasks, making advanced agent capabilities accessible with less computational resources.
Prospective memory benchmarks show frontier LLMs struggle with only 65.1% Set-F1 on PM-Bench. The Prospective Intention Store (PIS) achieves 82.9% Set-F1 on DeepSeek-Chat and 66.2% on Gemma-E2B, compared to 4.2% without a store. PIS reaches 70.1% Set-F1, surpassing retrospective memory methods at 54.4% and setting a new state of the art.
Making Prospective Memory SLM-Shaped: Typed Intention Stores for Small-Model Agents
Prospective memory means carrying out a deferred intention at the right future cue while other work continues. Benchmarks now isolate it as an agent skill, yet frontier LLMs still struggle: the best published PM-Bench scaffold reaches only 65.1% Set-F1. We argue that this loop is schema-constrained state tracking rather than open-ended reasoning, and that small models can execute it when the action space is typed. We propose the Prospective Intention Store (PIS) that puts lifecycle logic in code and scoped language work on the model. The scaffold is agentic and training-free: no selector fine-tuning and no trajectory distillation. On PM-Bench, DeepSeek-Chat with PIS reaches 82.9% Set-F1. On Gemma-E2B, Set-F1 is only 4.2% without a store and at most 6.6% under seven retrospective memories, while PIS reaches 66.2%. PIS further reaches 70.1% Set-F1, where retrospective memory methods stay at most 54.4%. PIS sets a new state of the art on this benchmark and enables small models to surpass the published large-model scaffold.
