惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

cs.AI updates on arXiv.org
cs.AI updates on arXiv.org
T
The Blog of Author Tim Ferriss
S
Securelist
C
CXSECURITY Database RSS Feed - CXSecurity.com
T
Tor Project blog
S
Schneier on Security
Google DeepMind News
Google DeepMind News
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
Blog — PlanetScale
Blog — PlanetScale
AWS News Blog
AWS News Blog
F
Fortinet All Blogs
C
Cisco Blogs
L
LangChain Blog
MongoDB | Blog
MongoDB | Blog
D
Docker
T
The Exploit Database - CXSecurity.com
The GitHub Blog
The GitHub Blog
月光博客
月光博客
量子位
A
Arctic Wolf
博客园 - 叶小钗
L
LINUX DO - 热门话题
J
Java Code Geeks
The Hacker News
The Hacker News
Security Latest
Security Latest
V
Vulnerabilities – Threatpost
M
MIT News - Artificial intelligence
博客园 - 【当耐特】
T
Tenable Blog
Microsoft Security Blog
Microsoft Security Blog
Recent Announcements
Recent Announcements
Simon Willison's Weblog
Simon Willison's Weblog
B
Blog
阮一峰的网络日志
阮一峰的网络日志
A
About on SuperTechFans
博客园_首页
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
T
Threat Research - Cisco Blogs
Exploit-DB.com RSS Feed
Exploit-DB.com RSS Feed
Y
Y Combinator Blog
PCI Perspectives
PCI Perspectives
Hacker News: Ask HN
Hacker News: Ask HN
P
Palo Alto Networks Blog
小众软件
小众软件
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
O
OpenAI News
P
Privacy International News Feed
P
Privacy & Cybersecurity Law Blog
N
Netflix TechBlog - Medium

Hacker News - Newest: "AI"

AI can't read an investor deck AI as an attorney? Student uses ChatGPT, Gemini to sue UW over alleged racial discrimination Hacking MCP Servers in AI Systems – The Rug Pull: Tool Changes After Approval GitHub - MeepCastana/KubeezCut: Free Web based video editor GitHub - GenAI-Gurus/awesome-eu-ai-act: Curated tools, official sources, OSS, templates, and guides for EU AI Act compliance. Can AI judge journalism? A Thiel-backed startup says yes, even if it risks chilling whistleblowers Coming soon: 10 Things That Matter in AI Right Now DARPA built an AI to fact-check enemy weapons claims What explains heterogeneity in AI adoption? When AI Meets Muscle: Context-Aware Electrical Stimulation Promises a New Way to Guide Human Movements - Department of Computer Science AI Changed How We Build. It Did Not Change What Matters. Linux rules on using AI-generated code - Copilot is OK, but humans must take 'full responsibility for the… Meta spins up AI version of Mark Zuckerberg to engage with employees Code Mode: Let Your AI Write Programs, Not Just Call Tools | TanStack Blog GitHub - Delavalom/graft: Go framework for building AI agents. Type-safe tools, multi-provider (OpenAI, Anthropic, Gemini, Bedrock), zero vendor SDKs. India's TCS tops estimates, says new AI models did not dent services demand Gen Z's fading AI hype Strong feeling: we are in a folded AI reality GitHub - machinarii/total-recall-catalog: A reference catalog of latest knowledge retrieval, memory & RAG systems GitHub - mensfeld/code-on-incus: Give each AI agent its own isolated machine with root, Docker, and systemd. Active defense detects and stops threats automatically.. Quantization, LoRA, and the 8% Problem: Benchmarking Local LLMs for Production AI Iran war: We spoke to the man making Lego-style AI videos that experts say are powerful propaganda Powell, Bessent discussed Anthropic's Mythos AI cyber threat with major U.S. banks GitHub - immartian/bellamem: Persistent belief-graph memory for AI agents. Retrieves decisive context by importance — not recency, not RAG, not /compact. recursive-mode: The Repo-Native Operating System for AI Engineering After the attack on Sam Altman's home, will AI CEO's go on the offensive? The biggest advance in AI since the LLM Opus 4.6 vs GPT 5.4 One Prompt Unity World Generation Test “AI polls” are fake polls Client Challenge Can AI be a 'child of God'? Inside Anthropic's meeting with Christian leaders How to Switch AI Chatbots and Why You Might Want To GitHub - MattMessinger1/agentic_refund_guardrail: Safe refund policy layer for AI agents — Python + TypeScript. Same behavior, shared tests. Adam/papers/emergent_values_whitepaper.md at master · strangeadvancedmarketing/Adam Ask HN: How do you stop playing 20 questions with your AI coding tools How far can automation and AI support psychotherapy? - @theU GitHub - stagas/rtdiff: realtime git diff gui and AI-assisted commits A Mac Studio for Local AI — 6 Months Later A History of the Early Years of AI at the University of Edinburgh Why AI Coding Tools Still Feel Stuck on Localhost MSN AI Datacenters Are Becoming Strategic Targets twitter.com Penn Researchers Use AI to Surface Unreported GLP-1 Side Effects in Reddit Posts Show HN: MoodSense AI (ML and FastAPI and Gradio, Deployed on Hugging Face) Moodsense Ai - a Hugging Face Space by aman179102 AI models are terrible at betting on soccer—especially xAI Grok GitHub - xialeistudio/echoic GitHub - HimashaHerath/github-dev-wrapped: AI-powered weekly GitHub activity reports deployed to GitHub Pages GitHub - alejandrobalderas/claude-code-from-source: Architecture, patterns & internals of Anthropic's AI coding agent — reverse-engineered from source maps AI and Tech brief: Ireland ascendant GitHub - Titovilal/context0: Context0 - Never Surrender Training for a Marathon with an AI Coach: What Worked and What Didn't Cyber Pulse: Agentic Intel - Apps on Google Play I Built an AI PR Reviewer That Catches Bugs by Not Looking for Bugs Gen Z workers are so fearful AI will take their job they’re intentionally sabotaging their company’s AI rollout | Fortune How AI Is Reimagining the Game of Golf–For Both Players and Courses GitHub - nattergabriel/reseed: A CLI tool for managing and distributing agent skills across projects Is SVG the final frontier? My AI workflow evolved from prompts to a near-autonomous workflow MLSharp Help - 3DGS Viewer & Generator I put my cognitive field based AI's runtime on GitHub Is Numble the first AI-proof game? A3: Kubernetes for autonomous AI agent fleets | Emergent Principles Deepali Vyas ("The Elite Recruiter") GitHub - msmarkgu/RelayFreeLLM: A restful API designed to route user prompts to various AI model providers. Unionized ProPublica staff are on strike over AI, layoffs, and wages Unleashing the Advantage of Quantum AI We're heading for an AI-fueled 'dementia crisis,' brain scientist warns The AI-Assisted Breach of Mexico's Government Infrastructure [pdf] GitHub - stef41/lmscan: 🔍 Detect AI-generated text and fingerprint which LLM wrote it. Open-source GPTZero alternative. Zero dependencies, works offline. MSN GitHub - visionscaper/collabmem: Enabling long-term collaboration with Agentic AI - building up episodic and world model memory over time with in-context awareness We gave an AI a 3 year retail lease in SF and asked it to make a profit | Andon Labs AI Code is Hollowing Out Open Source, and Maintainers are Looking the Other Way What leaked "SteamGPT" files could mean for the PC gaming platform's use of AI AI is the boss at this retail store. What could go wrong? GitHub - Wuzu11517/agentic-proxy: Local proxy meant to help reduce With Drones, Geophysics and ArtificiaI Intelligence, Researchers Prepare to Do Battle Against Land Mines A Single Operator, Two AI Platforms, Nine Government Agencies: The Full Technical Report 在 Steam 上购买 FriedrichAI: Offline AI 立省 10% GitHub - inevolin/resume-cli: Hit Claude usage limits? Resume any AI coding session elsewhere. Switch tools at zero friction. GitHub - atripati/ark: AI Runtime Kernel — a context operating system for AI agents. Eliminates tool bloat, loads only what’s needed, and gives LLMs their reasoning space back. How to Build a Secure AI PR Reviewer with Claude, GitHub Actions, and JavaScript This Startup Wants You to Pay Up to Talk With AI Versions of Human Experts Intel Arc Pro B70 Brings 32GB VRAM to Local AI for $949 WordPress 7.0: The Good, the AI, and the Still Missing AI on the couch: Anthropic gives Claude 20 hours of psychiatry IatroBench: Pre-Registered Evidence of Iatrogenic Harm from AI Safety Measures AI Agents Know About Supabase. They Don't Always Use It Right. The history and future of AI at Google, with Sundar Pichai Inside an AI‑enabled device code phishing campaign How Meta Used AI to Map Tribal Knowledge in Large-Scale Data Pipelines AI for Systems: Using LLMs to Optimize Database Query Execution Forecasting the Economic Effects of AI Introducing Tinker: Play with AI, bring your ideas to life AI sheds light on an ancient gaming mystery People really hate AI but not as much as Iran—or Democrats | Fortune What is an AI Product Engineer? Phoebe Gates wants her $185 million AI startup to succeed with 'no ties to my privilege or my last name': 'I have a chip on my shoulder' | Fortune
Why we shipped Dreaming for AI memory
soorajsanker · 2026-06-03 · via Hacker News - Newest: "AI"

Anthropic's Managed Agents Dreams are a strong signal: long-running agents need memory consolidation, not just memory storage. The same pressure shows up in shorter products too, as soon as users return across separate sessions. Here is the research thread, what happens to personalization when memory is never cleaned, and why MemexAI takes a different infrastructure path.

Once agents work across sessions, memory cannot just accumulate. It needs to be curated, cleaned, corrected, and made usable again. That is the admission Anthropic made explicit in Managed Agents Dreams — and it is the right framing. Multi-session memory is not a storage problem. It is a maintenance problem.

Anthropic has now made that idea explicit in Claude Managed Agents. Their Dreams feature lets Claude reflect on past sessions, deduplicate and reorganize memory, surface new insights, and produce a new memory store for review. That is a useful direction for the whole agent ecosystem. It says that memory consolidation is becoming a primitive, not a nice-to-have.

MemexAI Dreaming comes from the same broad observation, but from a different product stance. We are not building a hosted agent runtime. We are building a memory layer that product teams can own: Postgres backed, inspectable, model-agnostic, and wired into the app where durable memory already has governance, consent, and operational consequences.

1-100sessions

Anthropic Dreams can take a memory store plus 1 to 100 past sessions as input during the research preview.

97%fewer first-pass errors

Anthropic reports Rakuten used Managed Agents memory to reduce first-pass errors within workspace-scoped boundaries.

30%faster verification

Anthropic reports Wisedocs sped document verification by remembering recurring document issues across sessions.

~6xcompletion-rate lift

Anthropic reports Harvey saw a roughly 6x completion-rate improvement in tests using Managed Agents with dreaming.

Figures above are Anthropic-reported customer and internal results. They are evidence that the category matters — not MemexAI benchmark claims.

The research thread

This idea has been circling in agent research for a while. Generative Agents used reflection to synthesize higher-level memories from lower-level observations, making agents more coherent over time. Reflexion showed that language agents can improve by writing verbal feedback about prior attempts. MemGPT framed memory as an operating-systems problem: the context window is limited, so agents need explicit memory tiers and movement between them.

The common thread is not that agents should remember everything. It is that raw experience is too messy to reuse directly. Something has to decide what matters, what changed, what conflicts, and what should be carried forward.

01

Agent writes local memory during work

02

Memory accumulates duplicates, fragments, and corrections

03

Background consolidation runs after quiet time

04

Durable memory becomes easier to trust next session

Multi-session memory is not a pile of old text. It is a maintained working set that future agent sessions rely on.

What Anthropic got right

Anthropic's Managed Agents memory is file-backed, permissioned, auditable, and programmatically controllable. That matters. Agents are already good at using files during serious work, so filesystem-like memory gives the model a familiar control surface instead of forcing every memory operation through opaque retrieval.

Their Dreams design is also explicit about the maintenance problem. Managed agents write local, incremental memories as they work. Across sessions, that creates duplicates, contradictions, and stale entries. A dream reads a memory store and past sessions, then produces a reorganized output store. The input store is not modified, so the result can be inspected before it is adopted.

I like this framing because it moves the industry away from the shallow version of AI memory: embed every conversation, retrieve some chunks, and call it personalization. The harder problem is not whether you can find an old fact. It is whether the durable memory state is still worth trusting.

Where MemexAI is different

MemexAI Dreaming is not a clone of Anthropic Dreams. It solves the same class of problem at a different layer of the stack.

Anthropic Dreams

Managed Agentsmemory store + sessionsdream jobseparate output store

MemexAI Dreaming

your appPostgres memory filesbounded dream passrevisioned patches

01

Runtime

Anthropic: Hosted Claude Managed Agents platform.

MemexAI: Self-hostable memory infrastructure for your app, service, or framework.

02

Input

Anthropic: A memory store plus selected past session transcripts.

MemexAI: Durable MemexAI memory files; logs and dream logs are excluded.

03

Output model

Anthropic: Produces a separate output memory store that can be reviewed or discarded.

MemexAI: Writes bounded patches through normal memory tools, creating revisions and access logs.

04

Primary goal

Anthropic: Help managed agents learn patterns across sessions and teams.

MemexAI: Keep app-owned memory clean, readable, and inspectable over time.

The biggest philosophical difference is the input. Anthropic Dreams can mine selected session transcripts for patterns and insights. MemexAI Dreaming deliberately works on the durable memory files MemexAI already owns. It excludes user/log.md, user/dream-log.md, files ending in -log.md, and files ending in .log.

That is a product decision. Raw transcripts are useful, but they are not the same thing as durable memory. In many products, transcripts include one-off debugging, accidental phrasing, sensitive context, or temporary exploration. MemexAI's first version of Dreaming is intentionally narrower: clean the memory record, do not mine the entire conversation archive.

Why we built it this way

MemexAI stores memory as scoped virtual files such as user/profile.md and shared/company.md. Agents can list, read, patch, search, and write files according to the resolved scope permissions: user/** is writable, while shared/** is read-only by default and writable only in shared RW mode. Humans can inspect them in the admin UI. Postgres stores the current file state, revisions, access logs, and runtime dream configuration.

That shape gives us a simple rule for background consolidation: if a dream changes memory, it should go through the same write path as every other memory change. No hidden side channel. No privileged rewrite path. No separate store that operators have to reconcile manually.

Dream writes use memory_write and memory_patch, create normal mx_revision rows, and appear in access logs with actor = dream-agent. Operators can inspect the before and after. If the dream agent finds nothing worth changing, it records zero files touched and does not add noise to user/dream-log.md.

The rationale is less about being conservative for its own sake and more about preserving the difference between evidence and memory. Evidence can be messy: transcripts, tool traces, temporary notes, and failed attempts. Memory is the maintained record the next session will trust. Mixing those too early makes the agent feel smart in the short term and brittle in the long term.

The scheduler is part of the design

Dreaming is opt-in. When MEMEX_DREAM_ENABLED=true, the service starts a background scheduler, but the database dream_enabled key remains the runtime master switch. The loop only selects users with qualifying user/ memory writes newer than their last dream run, waits for a quiet grace period, respects per-user pause flags, and enforces a write budget.

Those details sound operational, but they are product decisions. A memory cleanup system that runs too eagerly can erase useful ambiguity. A system with no write budget can change too much state at once. A system with no per-user pause makes debugging and user-level opt-out harder than it needs to be.

Background consolidation should be automatic enough to keep memory healthy, but controlled enough that operators can understand and stop it.

The product reason

The deeper product bet is taste memory. For consumer AI products, the useful question is not simply whether the AI can retrieve something a user said. It is whether the product can maintain a useful model of the user: what they prefer, what they avoid, how they make decisions, what constraints matter, and what has changed over time.

If that model gets noisier with every interaction, personalization eventually becomes a liability. The AI may technically remember more while feeling less trustworthy. That is why memory maintenance matters. A good memory layer should make the user record cleaner with use, not slowly turn it into a junk drawer.

Any AI product with persistent user memory eventually needs a memory hygiene loop, because user trust depends on the quality of the maintained record, not the total volume of remembered text.

What shipped

The core design is now live: opt-in background scheduling with a quiet grace period, write budgets that cap how much a single dream pass can change, per-user pause controls, and full revision and access-log preservation through the normal memory write path. Dream writes are indistinguishable from agent writes in the audit trail — except the actor is dream-agent. If a dream run finds nothing worth changing, it records zero files touched and stays silent.

Still ahead: source-aware consolidation, consent controls, conflict detection for context-dependent preferences, quality evals for dream runs, and policy hooks for sensitive domains. But this first version draws the line that matters — memory is product state, and background maintenance should be inspectable infrastructure, not a hidden rewrite.