Perhaps the market won't even give us six months. We are also juggling research, development, product, design, and operations. However, under this pace, it is difficult to ensure academic results, code quality, product experience, and even design taste. Fortunately, the team atmosphere is very good, and everyone wants to make this happen. You can also look forward to our upcoming release of Raven, a Memory-first, ultra-strong self-evolving Agent.
@elliotchen100
-
EverOS Completes Major Update: Wiki and Reflection Features Launched
By
–
Recently the update speed has been very fast. The last two important features of this version are now online. With Wiki and Reflection, it's now complete. In one sentence: EverOS's goal is not to be a memory API, but to be an open-source, local-first, Markdown-native, evolvable Agent Memory OS. 1. Markdown is the source of truth
All long-term memory first is standard. -
Tips for Automatically Cleaning Duplicate Browser Tabs with Codex
By
–
Here's a little browser use trick I've been loving recently – it's very smooth. I'm the kind of heavy user who opens dozens or hundreds of tabs at once, and they often end up duplicated. Later, I added a trigger task in Codex. Every time I lock my screen or go offline, Codex automatically calls browser use to clean up and close redundant, duplicate tabs – completely hands-free. This trick works with Codex and CC.
-
Impeccable: A Design Understanding and Real-Time Modification Add-on for AI Coding Agents
By
–
Many people may not know what Impeccable is for. Simply put, it's a "design add-on" for AI coding agents. It's not just about having the agent write out the page, but enabling it to understand design language, project context, visual inspection, and to look and modify in the browser in real time. I've used it myself. Except for the name being hard to pronounce, everything else is pretty smooth. What it solves is not "AI
-
The Core of AI Agent: Context, Harness, and Loop
By
–
这个讲得非常好，推荐大家看一下这个视频。
— 艾略特 (@elliotchen100) 23 juin 2026
很多人做 AI 还停在“怎么写 prompt”，但 Agent 真正能开始干活，是从 context、harness 到 loop 这一整套东西接起来：知道什么、能用什么、怎么判断下一步、怎么反复推进。
把这几个概念串起来看，很多 AI 产品的差距就很清楚了。 https://t.co/vrKCKQ5jvcThis is very well said. I recommend everyone to watch this video. Many people working on AI are still stuck on 'how to write prompts', but an Agent can truly start working only when the whole set of things from context, harness, to loop are connected: knowing what, being able to use what, how to judge the next step, how to iteratively advance. By linking these concepts together, the gaps between many AI products become clear.
-
AI Great Replacement Era: How Can Ordinary People Survive?
By
–
The era of AI great replacement has arrived. How can ordinary people survive? (Translation) Over the past two years, many people have mixed feelings when reading AI news. On one hand, it feels great: what used to take days can now be done in minutes. On the other hand, there is a creeping anxiety: if I can use these tools, and my boss can too, and the company can too, then what am I really worth? Dan Koe
-
Benchmark for testing agent self-evolution
By
–
Allow me to present our Benchmark, which is specifically designed to test the self-evolution capability of Agents, which is absolutely crucial. This year, we have constantly explored the direction of self-improvement / self-evolution, because we have
-
New Product Raven: Self-Evolving Agent OS
By
–
我又来剧透了
— 艾略特 (@elliotchen100) 21 juin 2026
从龙虾到爱玛氏，从爱玛氏到什么？
没错，我们马上要发布一个全新的产品：Raven。
它不是又一个「会调用很多工具」的 Agent。
它是一个会自我进化的 Agent OS。
大多数 Agent 的学习，停在技能层：多学一个工具，多记一条流程，多写一段 prompt。
Raven 不一样。… pic.twitter.com/KoO8KGMrPCHere I go again with a spoiler. From Lobster to Aimashi, from Aimashi to what? That's right, we are about to release a brand new product: Raven. It's not just another Agent that 'can call many tools'. It's a self-evolving Agent OS. Most Agents' learning stops at the skill level: learn one more tool, memorize one more process, write one more prompt. Raven is different.
-
Memory Becomes a Must-Have for Agents; Wiki and Dreaming Launching Soon
By
–
Easter egg: Wiki and Dreaming will be launched very soon. This weekend in Singapore, after in-depth conversations with a group of friends working on overseas Agent products, I had a very intuitive feeling: everyone's demand for Memory has shifted from "nice-to-have" to a must-have. When we started working on Memory, it was actually from a more engineer-oriented perspective:
on one hand, it can replace part of RAG — not everything needs to be stuffed into a knowledge base and retrieved again. -
EverMind’s Two Papers on AI Memory Confusion Accepted at KDD 2026
By
–
The biggest problem with AI memory may not be 'forgetting', but 'confusion'. In multi-person collaboration, what A said to B, decisions later changed by C, temporary consensus in a group chat — none of these can be compressed into a unified memory. This time, two memory-related works from EverMind have been accepted at KDD 2026.