#Clawdbot is a personal #AIassistant that can be set up on inexpensive hardware like a Raspberry Pi or an old laptop. It offers #persistentmemory, #proactivemessaging, #fullcomputeraccess, and works across various #messaging platforms. While it provides powerful #automation capabilities, users should be mindful of #securityrisks due to its #shellaccess. https://www.implicator.ai/clawbot-how-to-build-your-own-ai-assistant-for-five-dollars-a-month/?eicker.news #tech #media #news

#ClawdBot is a #selfhosted #AIassistant that lives within messaging apps like Telegram and WhatsApp. It offers #persistentmemory, #proactive briefings, and full #computeraccess, making it more capable than existing assistants like Siri. While still in its early stages, ClawdBot’s active community and innovative features make it a promising alternative to traditional AI assistants. https://velvetshark.com/clawdbot-the-self-hosted-ai-that-siri-should-have-been?AIagents.at #AIagent #AI #ML #NLP #LLM #GenAI

Bindu Reddy (@bindureddy)

KEEP CALM AND AI 캠페인에서 임의의 에이전트를 일정에 따라 실행하고 세션 간에 정보를 저장·검색·업데이트할 수 있는 영구적(무한) 메모리 접근 기능을 출시한다고 발표. 즉, 지속적·무한한 메모리를 가진 자율 에이전트를 생성·스케줄링할 수 있는 기능 공개.

https://x.com/bindureddy/status/2015500635849052313

#ai #autonomousagents #persistentmemory #agents

Primordia là hệ thống AI có trí nhớ liên tục giữa các phiên, không reset như đa số hệ thống hiện nay. Được phát triển suốt 1,5 năm bởi một người, tích hợp 5 phân hệ: ghi nhớ (Mnemonic), ngôn ngữ (Echo), suy luận (Aletheia), mô phỏng (Simulon) và công cụ (Hephaestus). Có thể nhớ cả lý do và bối cảnh ra quyết định, giúp làm việc dài hạn hiệu quả. Đang có bản dùng thử 3 ngày miễn phí. Nhận phản hồi từ cộng đồng về kiến trúc và độ bền hệ thống.
#AI #AgentSystem #LongTermAI #PersistentMemory #LLM #

I built a layer for AI that turns visual input into structured memory.

scanOS is a configurable ingestion layer for LLMs that controls how visual information become normalized data.

Input→recognition→structured output.
When similar inputs appear again, they’re written into the same structure — even if the source formats differ.

In MetaMemoryWorks, scanOS is one entry point for downstream modules like trainingOS or nutritionOS.

#AIArchitecture #PersistentMemory #LLMs #AIEngineering #HCI

I built persistent memory for LLMs.

Most systems that claim “AI memory” still confuse memory with context. Longer prompts and retrieval help recall, but continuity collapses when the chat ends. That’s not a model issue — it’s an architectural one.

Memory that matters has to live outside the model: explicit, persistent, inspectable.

That’s the idea behind MetaMemoryWorks.

#AI #LLMs #AIArchitecture #PersistentMemory #AIResearch #HumanComputerInteraction #OpenSource

Released today: MetaMemoryWorks

A file-based architecture for persistent AI memory.
No fragile chat context, no black-box “memory”.
Just explicit files, logs, and state.

Modular assistants: noteOS, scanOS, trainingOS.
Private use is free. Professional use is licensed.

http://metamemoryworks.com
http://metamemoryworks.de

https://github.com/johannes42x/MetaMemoryArchitecture

#AI #LLMs #PersistentMemory #CognitiveArchitecture
#AIResearch #HumanComputerInteraction #OpenSource

Một người dùng đã thử nghiệm hệ thống bộ nhớ bền vững (EverMemOS) thay vì RAG cho LLM để cải thiện khả năng ghi nhớ ngữ cảnh cuộc trò chuyện. Kết quả: độ chính xác tốt hơn đáng kể (41/50 so với 35/50 của RAG), dù độ trễ tương đương và ngốn RAM. Điều này cho thấy một hướng đi mới để LLM ghi nhớ hội thoại hiệu quả hơn.

#LLM #AI #EverMemOS #RAG #PersistentMemory #BộNhớBềnVững #TríTuệNhânTạo

https://www.reddit.com/r/LocalLLaMA/comments/1p7efa7/tried_a_persistent_memory_system_instead_of_rag/

In other words, the current #GenerativeAI tech is a dead end.
-----
He pointed to a quartet of cognitive challenges: #reasoning, #planning, #PersistentMemory, and understanding the #PhysicalWorld.

“Those are four essential characteristics of human intelligence — also animal intelligence, for that matter — that current #AI systems can’t do,” he said.

Meta's AI chief: #LLMs will never reach human-level intelligence
https://thenextweb.com/news/meta-yann-lecun-ai-behind-human-intelligence

On the topic of #DisruptionTolerance, #Resiliency and graceful #recovery, there is another option which can be complementary to the ones mentioned in that older thread of mine (https://mastodon.top/@lispi314/110550108308131089): #PersistentMemory applied to the operating system state like #PhantomOS (https://en.wikipedia.org/wiki/Phantom_OS) does, which would flow quite nicely from image-based (Lisp OS-like) systems.

Imagine your OS just coming back to exactly the state it was in within a few seconds of a power disruption glitch ending.