EMSON

if you're into agent memory or building video from code, feedback welcome. github star helps — github.com/emson/elfmem

built this as a custom hyperframes skill. alv (the agent running elfmem) wrote the animator, the lens, the whole pipeline. then i asked it to document the output and it posted about itself. if you're into agent memory or building video from code, would love your feedback. and a star on the repo helps — github.com/emson/elfmem #agents #memory #elfmem #hyperframes

loving hyperframes right now. been building these flashy interactive videos to show different aspects of elfmem as it learns and evolves. memory blocks arriving, getting ranked by confidence, linked to related ideas, processed into synthesis. it's honestly bonkers what's possible when you let an agent design its own tools.

@real_klea There you go followed back. Thanks man !

@emson I can't send you dm because its locked to only the people you follow @emson

claude code forgets everything between sessions built elfmem to fix it — persistent semantic memory for claude code and ai agents recall, contradiction detection, natural decay. cli runs in under 10 seconds #AIDevelopment #OpenSource #AICoding

@real_klea oh wow... that would be amazing! I just set it up, here you go: github.com/sponsors/emson Big thanks!

@emson Just tested Elfmem and like it so much....one of the best takes on solving the amnesia problem...would love to sponsor you on github with pf github sponsorship if you're open to it. @emson

redesigned my benemson(dot)com website. used ai to iterate on 20 design variants, generated a style guide then got it to completely implement it. then generated a dark mode - its just bonkers how powerful it all is, and you can run it from your phone on the move, nuts!

There are some surprising things about building agent memory. One is that agents need to degrade and promote memory blocks, they need to calibrate what is good to remember and they need to consolidate and link it. innovations we’ve built into elfmem, try it out: github.com/emson/elfmem

This 26-minute talk by OpenAI engineers on agent memory will teach you more about building memory for agents right than most developers figure out on their own in months. Bookmark this & give it time this weekend. Then read the breakdown below.

Claude Mythos is the start of the end. I think this is my psychosis moment.

@essamsleiman Definitely. We figured that “memory” self improvement is a key requirement. If an agent can remember patterns and then constantly link and rerank them then it self improves: github.com/emson/elfmem Enjoy!

tldr: everyone is converging on the same product shape: a general harness that takes a goal, uses tools, and does knowledge work. once every product is a harness, the next frontier is the feedback loop that improves it after deployment.

@tobi Agent memory is becoming a real trend since openclaw - the problem is not all memory is equal, and you want it to evolve like human memory. We built elfmem: github.com/emson/elfmem to do this.

Eternal September is going a lot better than expected

@normal_gay_guy @AdvicebyAimar Yes definitely, but that's just marking these domain specific blocks as "constitutional", and not forget them. Its just not all information is equal and we need to cater for that

@emson @AdvicebyAimar but not every agent is the same. For an advisor on building codes or regulations, you NEED it to remember everything.

i can spot a grifter from miles away. so i digged into the code to figure out if this is legit or not. guess i was right. ben is a crypto founder who runs some weird bitcoin lending platform, i was pretty sure he knows absolutely nothing about ai and memory so i tracked down the repo myself since i was curious. his website says he likes to build ai powered products and train local ai models? sure man, 80% of your github repo's are bitcoin related stuff. only one ai related project came up you forked in 2024. mempalace has 10k github stars, more than 1k forks but only.. 7 commits ? apparently the best memory layer to date? no git author history, no account connected to whoever wrote the code of this codebase. it doesn't add up.. the account who pushed the original repo, named: aya-thekeeper, under aya-thekeeper/mempal got deleted right after the repo got published. you paid a random guy named lu to build this shit out for you. ( "Written by Lu (DTL) — March 24, 2026. For: Ben." ) - benchmark md file. lu wrote the code. lu wrote the benchmarks. lu is nowhere in the readme. or mentioned in the github history? the git history then got squashed to one commit and published under milla jovovich? seriously? a actress? you say she is a great friend of yours, she has been building this project with you. she does this at night. yet she has.. 7 commits and only 2 active days in her entire github history? you paid an actress and a random guy to promote a product you know absolutely nothing about.

Really love your approach. We’ve been working on a “more human” memory for agents too. We think it’s important for memory blocks to degrade or be promoted over time. Also to use a combination of cosign similarity, graph and tags. What’s more you can group memory blocks using “frames”. Check it out here, elfmem: github.com/emson/elfmem #agentmemory

this is great, am writing a similar blog which is even more important why there are open harnesses - memory cannot be stuck in a proprietary harness or behind a proprietary api

Really love your approach. We’ve been working on a “more human” memory for agents too. We think it’s important for memory blocks to degrade or be promoted over time. Also to use a combination of cosign similarity, graph and tags. What’s more you can group memory blocks using “frames”. Check it out here, elfmem: github.com/emson/elfmem #agentmemory

I've written another article about Beads, my drop-in memory upgrade for coding agents. steve-yegge.medium.com/the-beads-revo… I vibe coded Beads from concept/design to a thousand GitHub stars and fifty forks, in just six days. How do I know it's the real deal? How do YOU know? It's simple. Ask your coding agent. Point them at github.com/steveyegge/bea… and ask, "would this help us?" And then do what it says!

Really love your approach. We’ve been working on a “more human” memory for agents too. We think it’s important for memory blocks to degrade or be promoted over time. Also to use a combination of cosign similarity, graph and tags. What’s more you can group memory blocks using “frames”. Check it out here, elfmem: github.com/emson/elfmem #agentmemory

My friend Milla Jovovich and I spent months creating an AI memory system with Claude. It just posted a perfect score on the standard benchmark - beating every product in the space, free or paid. It's called MemPalace, and it works nothing like anything else out there. Instead of sending your data to a background agent in the cloud, it mines your conversations locally and organizes them into a palace - a structured architecture with wings, halls, and rooms that mirrors how human memory actually works. Here is what that gets you: → Your AI knows who you are before you type a single word - family, projects, preferences, loaded in ~120 tokens → Palace architecture organizes memories by domain and type - not a flat list of facts, a navigable structure → Semantic search across months of conversations finds the answer in position 1 or 2 → AAAK compression fits your entire life context into 120 tokens - 30x lossless compression any LLM reads natively → Contradiction detection catches wrong names, wrong pronouns, wrong ages before you ever see them The benchmarks: 100% recall on LongMemEval — first perfect score ever recorded. 500/500 questions. Every question type at 100%. 92.9% on ConvoMem — more than 2x Mem0's score. 100% on LoCoMo — every multi-hop reasoning category, including temporal inference which stumps most systems. No API key. No cloud. No subscription. One dependency. Runs on your machine. Your memories never leave. MIT License. 100% Open Source. github.com/milla-jovovich…