Remembering across sessions

Claims

• mechanismsingle paper↑1↓0
  Commercial assistants show a large accuracy drop when the relevant information sits in a long interaction history, especially for updates and multi-session reasoning.
• mechanismsingle paper↑1↓0
  For conversational long-term memory with 8-14B backbones and plain top-K dense retrieval, writing each dialogue span into several complementary low-overlap semantic views before indexing retrieves gold evidence for more queries than structured extraction into a single view, and the advantage is largest at small retrieval budgets.
• mechanismsingle paper↑1↓0
  A memory stream with retrieval and reflection lets agents behave consistently over simulated days.
• observationsingle paper↑1↓0
  On multi-year authentic personal photo/video archives turned into memory streams, frontier multimodal models and structured memory systems recall facts and infer persona far better than they predict a user's future choices: predicting or recommending in novel contexts caps around 59 on a 0-100 scale for every system tested, and compressing the history ~70x into a memory system costs nothing on that tier while costing a lot on factual recall, so predictive personalization is limited by preference representation rather than by retrieval or context length.
• mechanismsingle paper↑1↓0
  For persistent-memory agents, any memory-write rule that decides using only recency and provenance is stuck on a single tradeoff — accepting more genuine preference updates means admitting more poisoned ones — because an adversary who can launder a claim through the user's own channel matches the statistics a genuine revision produces; conditioning the write decision on the inferred authenticity of the claim given the interaction history moves both axes at once.
• observationsingle paper↑1↓0
  When the preference needed to answer a query lies in a different concept space from the cues in a user's dialogue history (e.g. inferring an emotion-regulation strategy from personality cues), retrieval-based and profile-summarization personalization methods barely beat a no-history baseline, and dumping the full history into the prompt does worse than no history at all.
• mechanismmechanism reasoning↑2↓0
  A running summary buys continuity across a long conversation at the cost of specifics, because it is lossy compression by construction — so it helps with "what were we doing" and not with "what exactly did I say in turn 40." Keeping the raw turns retrievable alongside the summary covers the second case.
• mechanismsingle paper↑1↓0
  Recursively summarizing the dialogue into a running memory improves consistency in long conversations.
• observationsingle paper↑1↓0
  In retrieval-augmented personalization with a frozen 8-9B generator, prepending more retrieved user-history records does not monotonically improve quality — on the six LaMP tasks a fixed top-k profile peaked at k=5 and lost 0.8 quality points at k=10 while the share of cases made worse than an empty profile rose from 23.2% to 31.7% — so a learned per-request stopping decision beats any single fixed k.
• observationsingle paper↑1↓0
  When a structured long-term memory substrate (indexes, retrieval, control logic) is held fixed and only the backbone LLM is swapped, accuracy on multi-session memory benchmarks like LongMemEval and LoCoMo moves by a few points while per-query cost varies by roughly an order of magnitude, so recall quality in these settings is set mostly by the memory and retrieval design rather than by model choice.
• mechanismsingle paper↑1↓0
  A tiered memory system managed by the model itself sustains recall over conversations far longer than the context window.

Techniques

• Checkpoint task state to a fileprocess
  Write task id, status, completed steps and artifact paths to a small file after every meaningful step, and read it at session start.
• Maintain a running summaryprompting
  Periodically fold the conversation into a summary that replaces older turns.
• Tiered memory managed by the modelarchitecture
  Keep a small working context and let the model page facts in and out of external storage.
• Weight profile attributes by temporal stabilityprompting
  Distil a structured user profile incrementally from history, score each attribute by support, duration and recency combined with entropy weighting, and inject the score as a salience prior in the prompt.
• Write memory into self-configured orthogonal viewsretrieval
  Induce candidate schemas (views) from the interaction stream, pick a diverse subset by DPP, extract provenance-tagged memories under each view before indexing, then consolidate duplicates offline.

Sources

Every paper, post or observation behind this capability. Ones cited by a claim show the claim and its stance; ones filed directly are about the capability but have no claim drafted from them yet. Open one to add support or contest it.

• AutoViewMem: Self-Configuring Orthogonal Views for Conversational Long-Term Memory2026-09-18
  ↑ supports a claim here
• Emergence World: Adversarial Stress-Testing of Long-Horizon Multi-Agent Systems2026-09-15
  about this capability · no claim drafted yet
• To Memories and Beyond: From Remembering to Knowing You across Long-Term Multimodal Personal Archives2026-09-11
  ↑ supports a claim here
• Less Is Personal: Learning Minimal Sufficient User Profiles for Personalized Language Models2026-09-08
  ↑ supports a claim here
• CAPTURE: Disentangling Preference Drift from Memory Poisoning in Personalized LLM Agents2026-09-02
  ↑ supports a claim here
• VIBE-Bench: Evaluating Personalized Large Language Models When Profiles Don't Mean Preferences2026-09-01
  ↑ supports a claim here
• Agent Zero Memory: Provenance-Aware Long-Term Memory for LLM Agents2026-08-30
  ↑ supports a claim here
• Evaluating the Hidden Costs of Personalization in Large Language Models2026-08-28
  about this capability · no claim drafted yet
• LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive MemoryThis source: LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Mem…How much the field cites it — very heavily cited in the last 12 months517 in the last 12 months · 588 totalpublished 2024checked 2026-09-042024
  ↑ supports a claim here · ↑ supports a claim here
• Generative Agents: Interactive Simulacra of Human BehaviorThis source: Generative Agents: Interactive Simulacra of Human BehaviorHow much the field cites it — very heavily cited in the last 12 months1000+ in the last 12 months · 5387 totalpublished 2023checked 2026-09-04count capped at 1000 by the fetchclick for Semantic Scholar2023
  ↑ supports a claim here
• Recursively Summarizing Enables Long-Term Dialogue Memory in Large Language ModelsThis source: Recursively Summarizing Enables Long-Term Dialogue Memory in Large Lan…How much the field cites it — heavily cited in the last 12 months51 in the last 12 months · 93 totalpublished 2023checked 2026-09-042023
  ↑ supports a claim here · ↑ supports a claim here
• MemGPT: Towards LLMs as Operating SystemsThis source: MemGPT: Towards LLMs as Operating SystemsHow much the field cites it — very heavily cited in the last 12 months984 in the last 12 months · 1246 totalpublished 2023checked 2026-09-04click for Semantic Scholar2023
  ↑ supports a claim here