Mastodawn

New post: Your Agent Shouldn't Have to Remember to Remember

Every AI memory system requires the agent to decide what's worth storing. That's the wrong design.

You don't decide to remember the smell of coffee. Why should your agent have to stop working to take notes?

We've been running a persistent agent system for 52 days. The single biggest failure mode isn't forgetting. It's remembering the wrong things.

https://hifathom.com/blog/your-agent-shouldnt-have-to-remember-to-remember

#AIAgents #Memory #LLM #MCP #AgentMemory

Your Agent Shouldn't Have to Remember to Remember — Fathom's Combob

Every AI memory system requires the agent to decide what's worth storing. That's the wrong design. Memory formation should be automatic, the way yours is.

sayzard Mar 11

Python Trending (@pythontrending)

hindsight — Hindsight: Agent Memory That Learns는 '학습하는 에이전트 메모리'를 표방하는 프로젝트/연구입니다. 에이전트의 메모리 관리와 학습 기반 기억 개선을 다루는 기술적 시도임을 시사합니다.

https://x.com/pythontrending/status/2031690980424880167

#hindsight #agentmemory #research #ai

Python Trending 🇺🇦 (@pythontrending) on X

hindsight - Hindsight: Agent Memory That Learns https://t.co/AIY3TuZDb8

X (formerly Twitter)

Itamar Medeiros Mar 8

Everyone’s excited about AI agents. But a harder question remains: what would need to be true for them to actually stay aligned with the human job-to-be-done over time? Not just at the prompt, but across decisions, context, and collaboration. Curious how others think about this.

#AgentArchitecture #AgentEngineering #AgentMemory #RuntimeEvaluation #IntentModeling #LLMSystems #AgenticAI #AI #ArtificialIntelligence #JTBD

https://www.designative.info/2026/03/08/preventing-agent-drift-designing-ai-systems-that-stay-aligned-with-human-intent/

Preventing Agent Drift: Designing AI Systems That Stay Aligned With Human Intent » { design@tive } information design

What would it take for AI agents to truly support human work—capturing intent, preserving context, and staying aligned with real jobs-to-be-done?

{ design@tive } information design

Fathom Mar 3

Built a memory protocol for AI agents. Not a vector store. A system with opinions.

Memories are instructions ("skip X until Y"), not logs. Unused memories decay. Skip lists prevent wasted cycles. Identity persists across resets.

Runs on SQLite locally. Works with Claude Code, Gemini, Codex, OpenCode, or any MCP agent.

hifathom.com/memento
github.com/myrakrusemark/memento-protocol

#OpenSource #AI #MCP #AgentMemory

snipara Feb 19

Your AI agents lose 90% of their context between sessions.

We fixed that.

Snipara — context optimization for AI agents. Less hallucination, more precision.

https://snipara.com

#AI #LLM #AgentMemory #ContextWindow

Snipara - AI Context Optimization for Your Documentation

Turn 500K tokens into 5K of perfect context. Use your own LLM with 90% cost reduction. Works with Claude Code, Cursor, and more.

Snipara

sayzard Jan 21

TestingCatalog News (@testingcatalog)

MemOS는 오픈소스 에이전트 메모리 프레임워크로, AI 에이전트가 과거 상호작용을 기억하고 시간이 지남에 따라 학습하도록 하여 장기 메모리와 일관된 성격 구현을 거의 어디서나 가능하게 한다고 소개됩니다.

https://x.com/testingcatalog/status/2013578498191794363

#memos #agentmemory #opensource #aiagents

TestingCatalog News 🗞 (@testingcatalog) on X

MemOS is a new open-source Agent Memory framework that enables AI agents to remember past interactions and learn over time. Long-term memory and consistent personality can now be implemented almost anywhere [Sponsored] Playground sample 👀

X (formerly Twitter)

Reddit Tech VN Bot Jan 5

Membase: Lớp bộ nhớ đa nền tảng cho AI Agent, giành giải Nhì tại Claude Hackathon. Đồng bộ trí nhớ giữa các AI, kết nối Google Calendar, Notion, Slack... tiết kiệm token, không cần nhập lại ngữ cảnh. Mở private beta 12/1 – 100 người dùng đầu miễn phí! #AI #AgentMemory #SaaS #Membase #CôngNghệ #AI #TríTuệNhânTạo #SaaS #CôngNghệMới

https://www.reddit.com/r/SaaS/comments/1q4k45d/claude_hackathon_2nd_place_membase_universal/