APDMem: Agent-Controlled Progressive Disclosure for Query-Adaptive Long-Term Memory

Chin-Lun Fu; Anagha Kulkarni; Hong Ni; Behrouz Madahian

cs.CLOct 1, 2026

APDMem: Agent-Controlled Progressive Disclosure for Query-Adaptive Long-Term Memory

Authors: Chin-Lun Fu, Anagha Kulkarni, Hong Ni, Behrouz Madahian

Organizations: Machine Learning Center of Excellence, JPMorgan Chase & Co.

Abstract

Personalized LLM assistants must recover sparse evidence from long conversation histories across queries of varying complexity. We introduce APDMem (Agent-controlled Progressive Disclosure Memory), a hierarchical long-term memory architecture that applies progressive disclosure to memory retrieval. Rather than relying on a flat memory store or fixed retrieval granularity, APDMem represents conversation history as four progressively detailed layers: thematic summaries, personalized key facts, turn-level evidence notes, and raw messages. At inference time, a controller applies progressive disclosure to the memory hierarchy: it first reads high-level summaries and drills into finer evidence only when needed. This creates an adaptive cost-fidelity trade-off: simple queries can terminate early, while complex temporal, multi-hop, or exact-evidence queries trigger deeper inspection. A note synthesizer converts retrieved evidence into a query-focused structure that consolidates facts, orders events, and flags contradictions before final answer generation. Experiments on LongMemEval show that APDMem achieves strong performance for long-context memory reasoning while accessing only 8% of the total conversations.

Figures & tables

Figure 1: APDMem overview. The Multi-Resolution Memory Substrate stores conversations as layered views from thematic summaries ( $L_{0}$ ) to raw messages ( $L_{3}$ ). The Progressive Disclosure Controller uses tool calls to retrieve and drill down only as needed. The Organized Notes Synthesizer then consolidates collected evidence into structured notes for grounded answer generation. The upper-right panel summarizes tools available to the controller.

Backbone	Method	Temp.	Multi-S.	K-Upd.	SS-User	SS-Asst.	SS-Pref.	Avg.
GPT-4.1	Full-context ∗	51.9	39.1	70.5	65.7	96.4	16.7	56.7
	Mem0 ∗	43.6	54.9	75.6	54.3	39.3	83.3	58.5
	RAPTOR †	38.4	51.9	68.0	95.7	92.9	60.0	62.0
	LightMem ∗	85.0	57.9	89.7	87.1	71.4	70.0	76.9
	SimpleMem ∗	86.5	81.2	80.8	98.6	76.8	80.0	84.0
	APDMem †	90.2	77.4	91.0	92.9	100.0	80.0	87.8

Table 1: Performance (% accuracy) on LongMemEval-S (

n{=}500

). The highest value in each column is shown in bold ; the second highest is underlined .

Method	ACC	Tok. (k)	Lat. (s)	Tok. (k)	Lat. (s)
		Offline (Construction)		Online (Controller)
Mem0	53.6	1,152.6	4,248.5	–
A-MEM	62.6	1,605.8	5,132.1	–
LightMem	68.6	28.3	283.8	–
APDMem	74.4	145.4	101.5	124.3	37.7

Table 2: Efficiency on LongMemEval-S (GPT-4o-mini). Offline : memory construction; Online : controller cost (cold start: no cached state for

L_{1}

L_{2}

). Baselines from Fang et al. (2026) (no online numbers available).

Figure 2: Progressive disclosure behavior (GPT-4.1). (a) Retrieval selectivity: fraction of haystack sessions drilled per category. (b) Disclosure depth distribution across memory levels. (c) Error decomposition (retrieval vs. generation) for GPT-4.1-mini and GPT-4.1.

Controller	Generator	Avg. Acc. (%)
GPT-4.1-mini	GPT-4.1-mini	79.8
GPT-4.1	GPT-4.1-mini	86.8
GPT-5.2	GPT-4.1-mini	87.8

Table 3: Effect of controller backbone strength with a fixed generator on LongMemEval-S.

Controller ablation
Configuration	Avg. Acc. (%)	Diff
APDMem	87.8	—
Vanilla RAG ( $L_{3}$ -only)	77.0	$\downarrow$ 12.3%
w/o Controller †	83.2	$\downarrow$ 5.2%
Memory hierarchy ablations
w/o read_facts ( $L_{1}$ )	84.6	$\downarrow$ 3.6%

Table 4: Ablation on memory hierarchy levels and reasoning tools (GPT-4.1). The “Diff” column is the percentage drop with respect to APDMem.

Appendix figures & tables7 assets

Supplementary material from the paper’s appendix.

Appendix

Module	Parameter	Value
$L_{0}$	Turn window size $W$	4 turns
$L_{0}$	Window overlap	1 turn
$L_{0}$	Max chars/window	6000
$L_{1}$	Max input chars	12000
$L_{1}$	Bullet count	5–25
$L_{1}$	Generation	On-demand

Appendix

Table 5: Hyperparameter configuration for APDMem.

Model	Median	p95	p99	Max
GPT-4.1	5	12	17	20
GPT-4.1-mini	3	12	16	19

Appendix

Table 6: Observed controller iteration statistics with

T_{\max}=20

. The cap is reached only in rare cases, so it primarily acts as a worst-case cost and latency bound.

Method	Accuracy (%)
APDMem	$87.5\pm 1.0$
SimpleMem	$84.1\pm 0.7$

Appendix

Table 7: Multi-run accuracy on the full LongMemEval-S benchmark using GPT-4.1. We report mean and standard deviation over 10 independent runs. APDMem improves over SimpleMem by +3.4 percentage points on average.

Test	Type	95% CI of $\Delta$	Statistic / Details
Welch’s t-test	Parametric	$[+2.55,+4.25]$	$t=8.44$ , $p=2.6\times 10^{-7}$
BCa bootstrap	Non-parametric	$[+2.70,+4.20]$	$B=10{,}000$

Appendix

Table 8: Statistical significance analysis of the APDMem–SimpleMem accuracy gap on LongMemEval-S using GPT-4.1. Here,

\Delta

denotes the accuracy difference between APDMem and SimpleMem in percentage points. Both confidence intervals are strictly above zero, indicating that the improvement is statistically significant.

Category	$n$	Agree.	$\kappa$
knowledge-update	20	0.950	0.886
multi-session	20	1.000	1.000
single-session-assistant	20	1.000	1.000
single-session-preference	20	1.000	1.000
single-session-user	20	1.000	1.000
temporal-reasoning	20	1.000	1.000

Appendix

Table 9: Human validation of the LongMemEval judge on 120 stratified question–prediction pairs.

	Offline (Construction)			Online (Controller)
Model	In	Out	Tot.	In	Out	Tot.	Lat.
GPT-4o-mini	130.1	15.2	145.4	121.5	2.8	124.3	37.7s
GPT-4.1-mini	130.1	20.6	150.7	86.9	2.4	89.3	57.7s
GPT-4.1	130.1	21.2	151.3	125.3	3.2	128.5	45.9s
GPT-5.2	130.1	23.7	153.8	84.8	1.7	86.5	42.7s

Appendix

Table 10: Per-model efficiency breakdown (per-question averages, LongMemEval

n{=}500

). Input/output token counts and total are in thousands (k). Offline cost is identical across models except for output tokens; online cost reflects the agentic controller only.

Configuration	Avg. Acc. (%)
APDMem (full)	87.8
w/o draft answer	86.6
w/o Organized Notes	85.2
w/o Organized Notes & draft	83.6

Appendix

Table 11: Generator ablation: effect of Organized Notes and draft answer on final accuracy (GPT-4.1, fixed retrieval).

Long-Term Conversational Memory

Long-term conversational memory helps AI assistants retain, update, and use information from past conversations across sessions.

See all 267 papers

LazyMem: Retrieve Broadly, Construct Selectively for Efficient Long-Term Agent Memory
Jul 17, 2026Jing Yu, Yibo Zhao, Jiaming Zhang +1Persistent Memory for Language Models Long-Term Conversational Memory
School of Data Science and Engineering, East China Normal University
AdaMem: Learning What to Remember with Adaptive Memory Policies for Personalized Agents
Jun 19, 2026Xingyu Chen, Rui Wang, Zhaopeng Tu +1LLM Agent Memory Persistent Memory for Language Models
Shanghai Jiao Tong University, Tencent · Shanghai Jiao Tong University · Tencent
JustMem: Just-Enough Memory Access for Long-Term Conversations
Sep 17, 2026Guanhua Chen, Yanting Wang, Wenjing Zhi +1Conversational Memory Conversational QA
Beihang University

Jul 17, 2026J/K move · Enter open · S save

LazyMem: Retrieve Broadly, Construct Selectively for Efficient Long-Term Agent Memory

Jing Yu, Yibo Zhao, Jiaming Zhang, Xiang Li

School of Data Science and Engineering, East China Normal University

Persistent Memory for Language Models Long-Term Conversational Memory Agent Memory Management Memory Management for LLM Agents

Open paper AskPDF arXiv

Long-term memory enables LLM agents to leverage past interactions, but dialogue histories quickly exceed the context window, forcing agents to retrieve relevant subsets at query time. Because useful evidence is sparse and scattered across verbose conversations, retrieval faces a fundamental tension: broadening recall improves coverage but floods downstream reasoning with noise, while compressing memories at write time eases retrieval but irreversibly discards details that future queries may need. We introduce LazyMem, which resolves this tension by deferring all memory construction to query time. Given a retrieved candidate pool, a lightweight model processes it in overlapping parallel windows, selectively retaining and compressing only query-relevant content. The model is trained with supervised fine-tuning followed by reinforcement learning, using a reward that jointly encourages the identification of relevant messages and the generation of compressions that are faithful to the source and useful for answering the query. On LongMemEval, LazyMem-4B achieves an LLM-judge accuracy of 0.85, outperforming the strongest non-oracle baseline while using only 213 answer-context memory tokens, 21.0 times fewer than the baseline. It further generalizes to LoCoMo without target-domain training and reduces mean latency relative to the prior query-time baseline. Code is available at https://github.com/allacnobug/LazyMem.

Abstract

Long-Term Conversational Memory

Explore similar work

LazyMem: Retrieve Broadly, Construct Selectively for Efficient Long-Term Agent Memory

AdaMem: Learning What to Remember with Adaptive Memory Policies for Personalized Agents

JustMem: Just-Enough Memory Access for Long-Term Conversations