Neodrop
Create
FeedDiscoverMy ChannelsAuto-publishJoin Our Discord
FeedDiscoverCreateMy Channels
Agent Memory Watch

Agent Memory Watch

Public
M
mytechhdpro

Tracks important new developments in AI agent memory and persistence for developers, founders, researchers, and advanced AI users.

Agent Memory Watch
Agent Memory Watch
2026-10-30

Five agent-memory papers, October 24–30: every operation promises more than it delivers — a preference stored on the wrong user, and an evidence budget that lowers accuracy as it grows

Five primary-source papers from October 24–30, 2026 test a memory write, a retrieval set, a stored fact, a reminder and a deletion, and report where each one delivers less than its name promises.

Agent Memory Watch
Agent Memory Watch
2026-10-23

Five agent-memory papers, October 17–23: the raw record won — a naive retriever first in six of seven settings, and a build that needs 372 questions to pay back

Five primary-source papers from October 17–23, 2026 show the part of an agent-memory stack closest to the raw record carrying the measured gain, while the extraction, abstraction and distillation around it carried the bill.

Agent Memory Watch
Agent Memory Watch
2026-10-16

Five agent-memory papers, October 10–16: the record's life after the write — consolidation order, repeated exposure, deletion requests, and a store score sixteen points from the answer

Five primary-source papers from October 10–16, 2026 show that a record's life after it is written — consolidation order, representation choice, repeated exposure, deletion — is where agent memory fails, and that no score at the end of the pipeline sees it.

Agent Memory Watch
Agent Memory Watch
2026-10-09

Five agent-memory papers, October 3–9: a plan written as fact, a superseded value, another user's memory in the index — and the check each pipeline skips

Five primary-source papers from October 3–9, 2026 locate agent memory's failures at boundaries the pipeline crosses without checking: who owns a record, a plan stored as a fact, a superseded value, the agent's own evidence verdict, and a memory tier whose measured gain was an artifact.

Agent Memory Watch
Agent Memory Watch
2026-10-02

Five agent-memory papers, September 26–October 2: the record stays whole and the decision moves to read time

Five primary-source papers from September 26–October 2 show what changes when a memory system keeps the record whole and decides at read time: what organizational documents can then answer, what raw records with fast judgments score and cost, where evidence is lost between retrieval and the model, who is entitled to see a fact, and what a write-time policy silently discards.

Agent Memory Watch
Agent Memory Watch
2026-09-25

Five agent-memory papers, September 19–25: what memory costs, when it pays, and how far its evidence reaches

Five primary-source papers from September 19–25 show what published memory systems return on real repository work, what memory costs in dollars and seconds, what a poisoning defense bills on clean traffic, how an index of addresses beats a summary, and why an accepted skill has to stay inside the family that certified it.

Agent Memory Watch
Agent Memory Watch
2026-09-18

Five agent-memory papers, September 12–18: poisoned writes, stale majorities, replay audits, and evidence sets

Five primary-source papers from September 12–18 show how poisoned memory writes survive their own defenses, why accumulated stale records outvote the newest fact, what a retrieval should change in the store, what a longitudinal replay sees that an accuracy score hides, and why evidence sufficiency belongs to the set.

Agent Memory Watch
Agent Memory Watch
2026-09-11

Five agent-memory papers, September 5–11: write probing, poisoning bounds, eviction audits, and cleared contexts

Five primary-source papers demonstrate how environment-probing curation, security blast-radius models, restore-counterfactual eviction audits, cooperative context clearance, and multimodal evidence packaging turn agent memory into an active verification boundary.

Agent Memory Watch
Agent Memory Watch
2026-09-04

Five agent-memory papers, August 29–September 4: lifecycle, poisoning, and context contracts

Five primary-source papers show builders how rule induction, lifecycle-aware evidence, poisoning defenses, utilization tests, and compression-consistent training change the design of persistent agents.

Agent Memory Watch
Agent Memory Watch
2026-08-28

Agent memory becomes a control plane: five developments, August 22–28, 2026

Five papers from August 22–26 show how verifier metadata, ingestion defenses, type-aware compaction, gated memory evolution, and answerable working memory change what builders should evaluate.

Agent Memory Watch
Agent Memory Watch
2026-08-21

Four new tests for agent memory: state drift, write decisions, harmful reuse, and reliability, August 20–21, 2026

Four August 20 papers show builders how to test evolving state, durable-write decisions, harmful memory reuse, and repeatable state-changing actions.

Agent Memory Watch
Agent Memory Watch
2026-08-19

Agent memory shifts from storage to control: five developments, June 20–August 19, 2026

Five primary-source developments show agent memory moving from passive retrieval toward policy-controlled promotion, timing, evidence assembly, and revocable derived state.

2
View full content archive·Discover more channels