Video wird geladen...

Video konnte nicht geladen werden

Beim Laden dieses Videos ist ein Problem aufgetreten. Dies könnte an einem vorübergehenden Netzwerkproblem liegen oder das Video ist möglicherweise nicht verfügbar.

there's now a formal proof that your agent's vector memory forgets what you stored and fabricates things you never did. scaling it up makes both worse, not better. "the price of meaning" (arxiv 2603.27116) proves it for any memory that retrieves by similarity in an embedding space. the same... geometry that lets embeddings generalize creates competitor mass in every neighborhood. add data and the crowding grows: retention decays toward zero, and false recall can't be tuned out without throwing away true hits. not a bug in your pipeline. the shape of the math. i learned this the expensive way. 500 stored facts, two weeks into a build, a user asks what i know about their job. retrieval hands back four fragments from different weeks: "i love my job," "thinking of quitting," "my manager is supportive," "my manager micromanages." the agent invents a clean synthesis of all four. the user had switched jobs in between. embeddings measure similarity, not truth. the topology angle says stop storing meaning as geometry, store it as structure. navigate an edge in an AST or a graph instead of searching a neighborhood. no crowding, no decay, no false recall. FORGE backs it: plain AST checks catch structural hallucinations at 100% precision (arxiv 2601.19106). here's what the structural pitch skips. the same proof shows pure structure escapes the geometry only by surrendering the connections embeddings find. you trade fabrication for blindness. so i stopped picking a side. what actually ships across thousands of sessions: – extract facts, not transcripts – resolve conflicts on write: archive the old job, mark the new one active – hybrid retrieval: vectors for discovery, graph for precision – decay plus nightly consolidation, so memory keeps what matters and lets the rest go memory is infrastructure, not a feature. that reframe is the whole game. and it's being measured now. WorldMemArena (may 28, arxiv 2605.29341) scores these paradigms head to head, embedding memory against retrieval-augmented against terminal-agent harnesses, across multimodal action-world tasks. the question moved from "does it remember" to "what kind of memory survives scale." full architecture, with the code, here:show more

Rohit

24,288 subscribers

16,591 Aufrufe • vor 3 Monaten •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

Engineer runs a Kimi K3 memory layer that costs $11 a month and remembers what a $500,000 vector database keeps losing. No embeddings. Four nodes and one rule about what's allowed to be forgotten. He published the whole schema. His version starts from the opposite idea. Memory is not a pile you search. It's a set of claims that expire unless something keeps paying to keep them. Four nodes. Every memory carries a clock someone has to reset: > WRITER - stores a fact with the reason it mattered, never raw text > DECAY - ages every memory down. Silence is deletion > RENEWER - only re-lifts a memory the model actually used again > GRAVE - holds what died, and why nobody reached for it Three nodes keep memory alive. One keeps the dead ones. Recall isn't storage here. It's rent a fact has to keep earning. That's the entire design. When everything is remembered forever, the useful and the stale retrieve identically. He replayed two months of agent context. 90,000 stored facts. 71,000 never retrieved once. The vector store returned all of them on similarity. Similarity graded closeness. Nobody graded whether the memory was ever right. Everyone else stuffs more into the context window and calls it memory. He built a layer that lets a fact die unless it keeps proving itself. The cost isn't storage. It's finding out how much of what your agent "knows" it has never once used. The article below is the full build - node prompts, the decay curve, the renewal rule. Save it. You'll want it open in the other tab.

Engineer runs a Kimi K3 memory layer that costs $11 a month and remembers what a $500,000 vector database keeps losing. No embeddings. Four nodes and one rule about what's allowed to be forgotten. He published the whole schema. His version starts from the opposite idea. Memory is not a pile you search. It's a set of claims that expire unless something keeps paying to keep them. Four nodes. Every memory carries a clock someone has to reset: > WRITER - stores a fact with the reason it mattered, never raw text > DECAY - ages every memory down. Silence is deletion > RENEWER - only re-lifts a memory the model actually used again > GRAVE - holds what died, and why nobody reached for it Three nodes keep memory alive. One keeps the dead ones. Recall isn't storage here. It's rent a fact has to keep earning. That's the entire design. When everything is remembered forever, the useful and the stale retrieve identically. He replayed two months of agent context. 90,000 stored facts. 71,000 never retrieved once. The vector store returned all of them on similarity. Similarity graded closeness. Nobody graded whether the memory was ever right. Everyone else stuffs more into the context window and calls it memory. He built a layer that lets a fact die unless it keeps proving itself. The cost isn't storage. It's finding out how much of what your agent "knows" it has never once used. The article below is the full build - node prompts, the decay curve, the renewal rule. Save it. You'll want it open in the other tab.

wast3

67,616 Aufrufe • vor 18 Tagen

New course: Agent Memory: Building Memory-Aware Agents, built in partnership with Oracle and taught by Richmond Alake and Nacho Martínez. Many agents work well within a single session but their memory resets once the session ends. Consider a research agent working on dozens of papers across multiple days: without memory, it has no way to store and retrieve what it learned across sessions. This short course teaches you to build a memory system that enables agents to persist memory and thereby learn across sessions. You'll design a Memory Manager that handles different memory types, implement semantic tool retrieval that scales without bloating the context, and build write-back pipelines that let your agent autonomously update and refine what it knows over time. Skills you'll gain: - Build persistent memory stores for different agent memory types - Implement a Memory Manager that orchestrates how your agent reads, writes, and retrieves memory - Treat tools as procedural memory and retrieve only relevant ones at inference time using semantic search Join and learn to build agents that remember and improve over time!

New course: Agent Memory: Building Memory-Aware Agents, built in partnership with Oracle and taught by Richmond Alake and Nacho Martínez. Many agents work well within a single session but their memory resets once the session ends. Consider a research agent working on dozens of papers across multiple days: without memory, it has no way to store and retrieve what it learned across sessions. This short course teaches you to build a memory system that enables agents to persist memory and thereby learn across sessions. You'll design a Memory Manager that handles different memory types, implement semantic tool retrieval that scales without bloating the context, and build write-back pipelines that let your agent autonomously update and refine what it knows over time. Skills you'll gain: - Build persistent memory stores for different agent memory types - Implement a Memory Manager that orchestrates how your agent reads, writes, and retrieves memory - Treat tools as procedural memory and retrieve only relevant ones at inference time using semantic search Join and learn to build agents that remember and improve over time!

Andrew Ng

161,260 Aufrufe • vor 6 Monaten

you let your agent keep notes, because the alternative was explaining yourself every morning it stopped asking, and it started repeating things that had stopped being true every guide on this covers what to save and where to put it, private for one agent, shared for the team not one of them covers how a fact leaves the agent writes its own memory while it works, so a guess it made on a tired tuesday gets stored in the same shape as something you told it, and after that the two are indistinguishable i counted mine this morning: forty files, six of them about decisions we reversed months ago, all forty read into every session since you did not give it a memory you gave it a record with no way to be corrected

you let your agent keep notes, because the alternative was explaining yourself every morning it stopped asking, and it started repeating things that had stopped being true every guide on this covers what to save and where to put it, private for one agent, shared for the team not one of them covers how a fact leaves the agent writes its own memory while it works, so a guess it made on a tired tuesday gets stored in the same shape as something you told it, and after that the two are indistinguishable i counted mine this morning: forty files, six of them about decisions we reversed months ago, all forty read into every session since you did not give it a memory you gave it a record with no way to be corrected

Lagosaur

13,898 Aufrufe • vor 9 Tagen

Memory vs. Graphs, clearly explained! memory is great, and the ceiling arrives quietly: it stores what happened. it does not store what to do about it. six runs later your file has fifty lines, and the model reloads all of them before it does anything. Graph engineering fixes this by changing what memory is: not a place things are kept, but an edge that runs backwards. you need both, and here is the sentence that resolves the whole confusion: a store keeps what happened. an edge keeps what to do about it. ↳ a store grows with every run, and every line is reloaded before the next one ↳ an edge carries one derived rule, and the rule replaces the run that produced it Prompts → Context → Harness → Loops → Graphs the transcript goes away, the constraint stays. and the constraint is smaller, because "adapters preserve keyword args exactly" is four hundred tokens shorter than the run that proved it. the same four blocks work on anything you can cut into lanes. i pointed them at token launches on Robinhood Chain, open source, nothing leaves your terminal the trick is knowing what deserves to survive. an output is not memory. "ported the utils slice, green on first pass" tells the next run nothing it can act on. the rule you derived from it does. one thing to know before you scale it. what you write down is not what comes back. ↳ the root rules file and auto memory are re-injected from disk. they come back intact, every time ↳ path-scoped rules live in message history. they get summarized away and do not return until a matching file is read again so a rule that must persist cannot be path-scoped. move it to the root and pay the always-loaded cost, or accept that it is advisory in any long session. and the one that eats whole nights: a memory file that has never had a line deleted is not memory. it is a tax on every run you will ever make, and nobody reads it back. below i have quoted my full guide on graph engineering. it covers the three topologies, the verifier patterns, and where the gate should actually open. save this, and the repo that runs it is below ↓

Memory vs. Graphs, clearly explained! memory is great, and the ceiling arrives quietly: it stores what happened. it does not store what to do about it. six runs later your file has fifty lines, and the model reloads all of them before it does anything. Graph engineering fixes this by changing what memory is: not a place things are kept, but an edge that runs backwards. you need both, and here is the sentence that resolves the whole confusion: a store keeps what happened. an edge keeps what to do about it. ↳ a store grows with every run, and every line is reloaded before the next one ↳ an edge carries one derived rule, and the rule replaces the run that produced it Prompts → Context → Harness → Loops → Graphs the transcript goes away, the constraint stays. and the constraint is smaller, because "adapters preserve keyword args exactly" is four hundred tokens shorter than the run that proved it. the same four blocks work on anything you can cut into lanes. i pointed them at token launches on Robinhood Chain, open source, nothing leaves your terminal the trick is knowing what deserves to survive. an output is not memory. "ported the utils slice, green on first pass" tells the next run nothing it can act on. the rule you derived from it does. one thing to know before you scale it. what you write down is not what comes back. ↳ the root rules file and auto memory are re-injected from disk. they come back intact, every time ↳ path-scoped rules live in message history. they get summarized away and do not return until a matching file is read again so a rule that must persist cannot be path-scoped. move it to the root and pay the always-loaded cost, or accept that it is advisory in any long session. and the one that eats whole nights: a memory file that has never had a line deleted is not memory. it is a tax on every run you will ever make, and nobody reads it back. below i have quoted my full guide on graph engineering. it covers the three topologies, the verifier patterns, and where the gate should actually open. save this, and the repo that runs it is below ↓

Hanako

47,766 Aufrufe • vor 8 Tagen

ANTHROPIC KILLED THE VECTOR DATABASE AND REPLACED IT WITH A FOLDER no embeddings. no black-box retrieval. no special memory tools. the agent just reads and writes a folder the insight: your agent already knows how to use a filesystem. so Anthropic made long-term memory a directory of small markdown files, mounted straight into the agent's sandbox at /mnt/memory the agent manages it with the tools it already has, bash, read, write, edit, grep, and Anthropic auto-injects a note into the system prompt so the agent always knows the memory is there and where to look the result: > every change is an immutable version: full audit trail, point-in-time recovery, redaction > read-only or read-write per store, so many agents can safely share one memory > workspace-scoped and persistent, memory that survives every session end most teams bolt on a vector DB and pray retrieval works. Anthropic went the opposite way: compiled knowledge as plain, human-readable files you can open, diff, and edit yourself and there's a second layer. Anthropic's "Dreaming" runs offline, reviews past sessions and the store, and rewrites it into a cleaner, higher-signal version. the agent literally consolidates memory in its sleep. bookmark this. the most production-grade agent memory right now is just a well-organized folder, and Anthropic proved it.

ANTHROPIC KILLED THE VECTOR DATABASE AND REPLACED IT WITH A FOLDER no embeddings. no black-box retrieval. no special memory tools. the agent just reads and writes a folder the insight: your agent already knows how to use a filesystem. so Anthropic made long-term memory a directory of small markdown files, mounted straight into the agent's sandbox at /mnt/memory the agent manages it with the tools it already has, bash, read, write, edit, grep, and Anthropic auto-injects a note into the system prompt so the agent always knows the memory is there and where to look the result: > every change is an immutable version: full audit trail, point-in-time recovery, redaction > read-only or read-write per store, so many agents can safely share one memory > workspace-scoped and persistent, memory that survives every session end most teams bolt on a vector DB and pray retrieval works. Anthropic went the opposite way: compiled knowledge as plain, human-readable files you can open, diff, and edit yourself and there's a second layer. Anthropic's "Dreaming" runs offline, reviews past sessions and the store, and rewrites it into a cleaner, higher-signal version. the agent literally consolidates memory in its sleep. bookmark this. the most production-grade agent memory right now is just a well-organized folder, and Anthropic proved it.

NO1ennn

13,359 Aufrufe • vor 1 Monat

I built the graph engineering pipeline instead of just reading about it. had to share what i found THE ENTIRE GRAPH ENGINEERING PIPELINE—ROOTED IN REAL, OPEN-SOURCE BUILDING, NOT SLIDES. Representation. Ontologies. Entity extraction. Relationships. Events. QA gate. Fusion. Embeddings. Actual memory your agent can navigate instead of hand-waving. And the best part: the end-to-end path already exists in public repos. Strwythura lays out the full ontology workflow, including entity resolution plus embedding-boosted GraphRAG. llm2kg tackles entity + relation extraction aligned to an ontology, persists it in Neo4j with vector embeddings, and ships a ReAct agent that queries the graph directly. No “coming soon.” Just working code you can point at your own data today.

I built the graph engineering pipeline instead of just reading about it. had to share what i found THE ENTIRE GRAPH ENGINEERING PIPELINE—ROOTED IN REAL, OPEN-SOURCE BUILDING, NOT SLIDES. Representation. Ontologies. Entity extraction. Relationships. Events. QA gate. Fusion. Embeddings. Actual memory your agent can navigate instead of hand-waving. And the best part: the end-to-end path already exists in public repos. Strwythura lays out the full ontology workflow, including entity resolution plus embedding-boosted GraphRAG. llm2kg tackles entity + relation extraction aligned to an ontology, persists it in Neo4j with vector embeddings, and ships a ReAct agent that queries the graph directly. No “coming soon.” Just working code you can point at your own data today.

CyrilXBT

12,800 Aufrufe • vor 3 Tagen

context engineering vs graph engineering. every few months the list gets a new word and everyone treats it as a replacement for the last one. these two are not on the same list. one decides what the model sees this turn, the other decides what exists at all. the cleanest way to tell them apart is to ask what a single unit of work looks like. > context engineering is the window the window opens empty, every single time. you assemble what goes in it. the prompt, the docs, the history, the tool results. the assembling is the work. the window only grows. it never shrinks on its own, so eventually something gets dropped. usually from the middle. usually without telling you. then the turn ends and the window is thrown away. not archived, thrown away. the next turn opens empty again and you re-explain what you already explained. good context engineering is knowing what to leave out, not what to pack in. the unit of work is one window. > graph engineering is the structure the same material arrives from the same sources. instead of packing it into a window, you pull entities out of it, resolve the duplicates into one node, and write typed edges between them. nothing here is stored as text you hope to find again. it is stored as a thing with a name and its connections to other things. when the turn ends, the graph is still there. the next turn does not start from zero. it starts by querying what already exists, and the query walks edges instead of guessing at similarity. good graph engineering is deciding what counts as the same thing twice. the unit of work is one relationship. > they are not alternatives the graph is what refills the window. context engineering decides what fits. graph engineering decides what there is to choose from. remove the graph and every session starts blind. remove the context work and the best structure in the world arrives as an unreadable dump. that also tells you which one broke. the answer drifted from what you actually said, or forgot something from this same session. that is the window. the answer is coherent but invents a connection that does not exist, or cannot join two facts it has clearly seen. that is the structure. people debug the prompt because the prompt is the easiest thing to edit. it keeps taking the blame for failures that live a layer down. save this - then read the full breakdown below

context engineering vs graph engineering. every few months the list gets a new word and everyone treats it as a replacement for the last one. these two are not on the same list. one decides what the model sees this turn, the other decides what exists at all. the cleanest way to tell them apart is to ask what a single unit of work looks like. > context engineering is the window the window opens empty, every single time. you assemble what goes in it. the prompt, the docs, the history, the tool results. the assembling is the work. the window only grows. it never shrinks on its own, so eventually something gets dropped. usually from the middle. usually without telling you. then the turn ends and the window is thrown away. not archived, thrown away. the next turn opens empty again and you re-explain what you already explained. good context engineering is knowing what to leave out, not what to pack in. the unit of work is one window. > graph engineering is the structure the same material arrives from the same sources. instead of packing it into a window, you pull entities out of it, resolve the duplicates into one node, and write typed edges between them. nothing here is stored as text you hope to find again. it is stored as a thing with a name and its connections to other things. when the turn ends, the graph is still there. the next turn does not start from zero. it starts by querying what already exists, and the query walks edges instead of guessing at similarity. good graph engineering is deciding what counts as the same thing twice. the unit of work is one relationship. > they are not alternatives the graph is what refills the window. context engineering decides what fits. graph engineering decides what there is to choose from. remove the graph and every session starts blind. remove the context work and the best structure in the world arrives as an unreadable dump. that also tells you which one broke. the answer drifted from what you actually said, or forgot something from this same session. that is the window. the answer is coherent but invents a connection that does not exist, or cannot join two facts it has clearly seen. that is the structure. people debug the prompt because the prompt is the easiest thing to edit. it keeps taking the blame for failures that live a layer down. save this - then read the full breakdown below

Hanako

19,160 Aufrufe • vor 1 Monat

New short course: Long-Term Agentic Memory with LangGraph. Learn to build an agent with long-term memory in this course developed in collaboration with taught by its Co-Founder and CEO, Harrison Chase! Personal assistance and productivity tasks have become important use cases for agents. An important feature of an AI assistant, such as a coding or calendar assistant, is its ability to keep improving over time from its experience. Agent memory is the key capability that enables this. To add memory to an agent, you must first figure out what to store and what to retrieve when it is time to use the information. Additionally, you’ll have to decide when to update the stored information. For example, you might update in each iteration loop of the agent or perform updates in the background, with a helper agent. In this course, you will learn a mental framework to build agents with long-term memory. You'll create a useful email assistant that can respond, ignore, and notify using writing, scheduling, and memory-management tools. You’ll develop your agent's memory by adding facts to its memory store, provide examples to learn the user's preferences, and optimize system prompts to evolve instructions based on previous responses. In detail, you’ll: - Learn how the three types of memory--semantic, episodic, and procedural–and the two update mechanisms–via hot path and in the background–apply to your agents. - Build an email agent with writing, scheduling, and availability tools, along with a router that triages incoming email and handles it accordingly by ignoring, responding, or notifying the user. - Add tools to your email agent that allow it to operate on semantic memory by learning facts about the user, storing them in a long-term memory store, and searching over them in future interactions. - Incorporate episodic memory, in the form of few-shot examples, in the triage step of your agents to help them learn and update user preferences. - Add procedural memory as system prompts, optimized with feedback to improve the instructions the agent follows. Learn how to approach memory in agents, and start building agents with long-term memory with LangGraph! Please sign up here:

New short course: Long-Term Agentic Memory with LangGraph. Learn to build an agent with long-term memory in this course developed in collaboration with taught by its Co-Founder and CEO, Harrison Chase! Personal assistance and productivity tasks have become important use cases for agents. An important feature of an AI assistant, such as a coding or calendar assistant, is its ability to keep improving over time from its experience. Agent memory is the key capability that enables this. To add memory to an agent, you must first figure out what to store and what to retrieve when it is time to use the information. Additionally, you’ll have to decide when to update the stored information. For example, you might update in each iteration loop of the agent or perform updates in the background, with a helper agent. In this course, you will learn a mental framework to build agents with long-term memory. You'll create a useful email assistant that can respond, ignore, and notify using writing, scheduling, and memory-management tools. You’ll develop your agent's memory by adding facts to its memory store, provide examples to learn the user's preferences, and optimize system prompts to evolve instructions based on previous responses. In detail, you’ll: - Learn how the three types of memory--semantic, episodic, and procedural–and the two update mechanisms–via hot path and in the background–apply to your agents. - Build an email agent with writing, scheduling, and availability tools, along with a router that triages incoming email and handles it accordingly by ignoring, responding, or notifying the user. - Add tools to your email agent that allow it to operate on semantic memory by learning facts about the user, storing them in a long-term memory store, and searching over them in future interactions. - Incorporate episodic memory, in the form of few-shot examples, in the triage step of your agents to help them learn and update user preferences. - Add procedural memory as system prompts, optimized with feedback to improve the instructions the agent follows. Learn how to approach memory in agents, and start building agents with long-term memory with LangGraph! Please sign up here:

Andrew Ng

131,965 Aufrufe • vor 1 Jahr

HERMES AGENT LEARNS FROM ITS OWN MISTAKES. UPDATES ITS MEMORY. CREATES ITS OWN SKILLS. NO CLOUD. EVERYTHING STORED LOCALLY. THIS IS HOW THE SELF-IMPROVING LOOP WORKS. most agents start from zero every session. Hermes carries forward what it learned. THREE MEMORY SYSTEMS: 1. PROCEDURAL MEMORY (how to act) stored in ~/.hermes/skills/ as SKILL.md files. when the agent repeats a complex workflow, it saves the procedure as a reusable skill. next time the same task comes up, it follows the skill instead of figuring it out again. you can also create skills explicitly: "create a skill called video-prep that captures how I format my video scripts. spoken english, define jargon inline, no em-dashes, close with a catchphrase." the agent writes the SKILL.md. available as a slash command from that moment. Hermes ships with 90+ skills. the number grows the longer you use it. 2. SEMANTIC MEMORY (durable facts about you) stored in ~/.hermes/memory/memory.md the agent scans conversations for facts worth remembering. preferences, habits, corrections, project details. real example from the video: agent tried to scrape a YouTube channel. URL was wrong. it failed. it updated memory.md with the correct URL pattern so it never makes the same mistake again. you can also save explicitly: "save to memory that my favorite testing framework is pytest" the agent updates memory.md immediately. this file loads into context on every session. the agent knows you better every week. 3. EPISODIC MEMORY (chat history) stored in ~/.hermes/state.db (local SQLite). every conversation. every tool call. every result. searchable with FTS5 full-text search. "search our past sessions. what was the first thing I ever said to you?" the agent queries state.db and finds it. over time, auxiliary models consolidate episodic memory into semantic memory. distilling recurring patterns into durable facts. THE SELF-IMPROVING LOOP: every agent run follows this cycle: → you send a prompt → working memory loads: SOUL.md + memory.md + relevant skills + chat history → agent calls tools (terminal, browser, delegate_task) → agent completes the task, replies to you → AFTER the reply: agent checks "did I learn something worth saving?" → if yes: updates memory.md or creates a new skill → next session starts smarter than the last this happens automatically. you don't ask the agent to learn. it decides what to remember on its own. WHAT MAKES THIS DIFFERENT FROM CLAUDE CODE: Claude Code has memory too. but Hermes stores everything locally. no cloud. your data never leaves your machine. Claude Code doesn't auto-create skills from experience. Hermes turns repeated workflows into reusable procedures. Claude Code memory is instruction-based. Hermes memory is conversational and self-updating. over months of usage, Hermes builds a knowledge base of your preferences, your projects, your mistakes, and the procedures that work for your specific workflow. the agent that remembers your birthday also remembers why your last deploy failed. NO EMBEDDINGS. PLAIN TEXT. Hermes does not use embeddings or RAG for memory. skill and memory search runs on plain text keyword matching. simpler. faster. no vector database to maintain. works entirely offline on your local machine. DELEGATE TO CLAUDE CODE: Hermes can spawn a sub-agent that runs Claude Code in headless mode: "spawn a sub-agent using Claude CLI to build a Python script that fetches the top 5 Hacker News stories to markdown." Hermes delegates. Claude Code writes the code. result returns to Hermes. Hermes runs the script and delivers the output. use Hermes for orchestration. use Claude Code for heavy coding. both tools. not competitors. WHAT HERMES DOES NOT HAVE: no built-in eval or LMOps system. no LangSmith, no LangFuse integration out of the box. trajectory export and logs exist but there is no automated quality tracking. if you need eval, build it yourself or connect external tools. the loop is self-improving. measuring how well it improves is on you. comment LOOP and I'll send you the configs that control how fast Hermes learns and what it remembers. memory limits, skill auto-creation triggers, and the auxiliary model that runs the learning. Replace your entire team with 8 hermes agents👇

HERMES AGENT LEARNS FROM ITS OWN MISTAKES. UPDATES ITS MEMORY. CREATES ITS OWN SKILLS. NO CLOUD. EVERYTHING STORED LOCALLY. THIS IS HOW THE SELF-IMPROVING LOOP WORKS. most agents start from zero every session. Hermes carries forward what it learned. THREE MEMORY SYSTEMS: 1. PROCEDURAL MEMORY (how to act) stored in ~/.hermes/skills/ as SKILL.md files. when the agent repeats a complex workflow, it saves the procedure as a reusable skill. next time the same task comes up, it follows the skill instead of figuring it out again. you can also create skills explicitly: "create a skill called video-prep that captures how I format my video scripts. spoken english, define jargon inline, no em-dashes, close with a catchphrase." the agent writes the SKILL.md. available as a slash command from that moment. Hermes ships with 90+ skills. the number grows the longer you use it. 2. SEMANTIC MEMORY (durable facts about you) stored in ~/.hermes/memory/memory.md the agent scans conversations for facts worth remembering. preferences, habits, corrections, project details. real example from the video: agent tried to scrape a YouTube channel. URL was wrong. it failed. it updated memory.md with the correct URL pattern so it never makes the same mistake again. you can also save explicitly: "save to memory that my favorite testing framework is pytest" the agent updates memory.md immediately. this file loads into context on every session. the agent knows you better every week. 3. EPISODIC MEMORY (chat history) stored in ~/.hermes/state.db (local SQLite). every conversation. every tool call. every result. searchable with FTS5 full-text search. "search our past sessions. what was the first thing I ever said to you?" the agent queries state.db and finds it. over time, auxiliary models consolidate episodic memory into semantic memory. distilling recurring patterns into durable facts. THE SELF-IMPROVING LOOP: every agent run follows this cycle: → you send a prompt → working memory loads: SOUL.md + memory.md + relevant skills + chat history → agent calls tools (terminal, browser, delegate_task) → agent completes the task, replies to you → AFTER the reply: agent checks "did I learn something worth saving?" → if yes: updates memory.md or creates a new skill → next session starts smarter than the last this happens automatically. you don't ask the agent to learn. it decides what to remember on its own. WHAT MAKES THIS DIFFERENT FROM CLAUDE CODE: Claude Code has memory too. but Hermes stores everything locally. no cloud. your data never leaves your machine. Claude Code doesn't auto-create skills from experience. Hermes turns repeated workflows into reusable procedures. Claude Code memory is instruction-based. Hermes memory is conversational and self-updating. over months of usage, Hermes builds a knowledge base of your preferences, your projects, your mistakes, and the procedures that work for your specific workflow. the agent that remembers your birthday also remembers why your last deploy failed. NO EMBEDDINGS. PLAIN TEXT. Hermes does not use embeddings or RAG for memory. skill and memory search runs on plain text keyword matching. simpler. faster. no vector database to maintain. works entirely offline on your local machine. DELEGATE TO CLAUDE CODE: Hermes can spawn a sub-agent that runs Claude Code in headless mode: "spawn a sub-agent using Claude CLI to build a Python script that fetches the top 5 Hacker News stories to markdown." Hermes delegates. Claude Code writes the code. result returns to Hermes. Hermes runs the script and delivers the output. use Hermes for orchestration. use Claude Code for heavy coding. both tools. not competitors. WHAT HERMES DOES NOT HAVE: no built-in eval or LMOps system. no LangSmith, no LangFuse integration out of the box. trajectory export and logs exist but there is no automated quality tracking. if you need eval, build it yourself or connect external tools. the loop is self-improving. measuring how well it improves is on you. comment LOOP and I'll send you the configs that control how fast Hermes learns and what it remembers. memory limits, skill auto-creation triggers, and the auxiliary model that runs the learning. Replace your entire team with 8 hermes agents👇

YanXbt

22,720 Aufrufe • vor 2 Monaten

New short course: LLMs as Operating Systems: Agent Memory, created with Letta, and taught by its founders Charles Packer and Sarah Wooders. An LLM's input context window has limited space. Using a longer input context also costs more and results in slower processing. So, managing what's stored in this context window is important. In the innovative paper MemGPT: Towards LLMs as Operating Systems, its authors (which include the instructors) proposed using an LLM agent to manage this context window. Their system uses a large persistent memory that stores everything that could be included in the input context, and an agent decides what is actually included. Take the example of building a chatbot that needs to remember what's been said earlier in a conversation (perhaps over many days of interaction with a user). As the conversation's length grows, the memory management agent will move information from the input context to a persistent searchable database; summarize information to keep relevant facts in the input context; and restore relevant conversation elements from further back in time. This allows a chatbot to keep what's currently most relevant in its input context memory to generate the next response. When I read the original MemGPT paper, I thought it was an innovative technique for handling memory for LLMs. The open-source Letta framework, which we'll use in this course, makes MemGPT easy to implement. It adds memory to your LLM agents and gives them transparent long-term memory. In detail, you’ll learn: - How to build an agent that can edit its own limited input context memory, using tools and multi-step reasoning - What is a memory hierarchy (an idea from computer operating systems, which use a cache to speed up memory access), and how these ideas apply to managing the LLM input context (where the input context window is a "cache" storing the most relevant information; and an agent decides what to move in and out of this to/from a larger persistent storage system) - How to implement multi-agent collaboration by letting different agents share blocks of memory This course will give you a sophisticated understanding of memory management for LLMs, which is important for chatbots having long conversations, and for complex agentic workflows. Please sign up here!

New short course: LLMs as Operating Systems: Agent Memory, created with Letta, and taught by its founders Charles Packer and Sarah Wooders. An LLM's input context window has limited space. Using a longer input context also costs more and results in slower processing. So, managing what's stored in this context window is important. In the innovative paper MemGPT: Towards LLMs as Operating Systems, its authors (which include the instructors) proposed using an LLM agent to manage this context window. Their system uses a large persistent memory that stores everything that could be included in the input context, and an agent decides what is actually included. Take the example of building a chatbot that needs to remember what's been said earlier in a conversation (perhaps over many days of interaction with a user). As the conversation's length grows, the memory management agent will move information from the input context to a persistent searchable database; summarize information to keep relevant facts in the input context; and restore relevant conversation elements from further back in time. This allows a chatbot to keep what's currently most relevant in its input context memory to generate the next response. When I read the original MemGPT paper, I thought it was an innovative technique for handling memory for LLMs. The open-source Letta framework, which we'll use in this course, makes MemGPT easy to implement. It adds memory to your LLM agents and gives them transparent long-term memory. In detail, you’ll learn: - How to build an agent that can edit its own limited input context memory, using tools and multi-step reasoning - What is a memory hierarchy (an idea from computer operating systems, which use a cache to speed up memory access), and how these ideas apply to managing the LLM input context (where the input context window is a "cache" storing the most relevant information; and an agent decides what to move in and out of this to/from a larger persistent storage system) - How to implement multi-agent collaboration by letting different agents share blocks of memory This course will give you a sophisticated understanding of memory management for LLMs, which is important for chatbots having long conversations, and for complex agentic workflows. Please sign up here!

Andrew Ng

201,127 Aufrufe • vor 1 Jahr

The creator of High Bandwidth Memory (HBM) put a number on the AI build that should stop every infra investor cold. A cluster of a million GPUs runs at roughly 10-20% utilization (Save this). Kim Jung-ho spent thirty years building what feeds the GPU, and his claim is that the GPU is barely working. Here is what is actually happening. Every time a model generates output, the data has to be read out of memory, computed, and written back. The read and the write swallow almost the entire cycle. While that data moves, the GPU does nothing. It sits there, fully powered, fully paid for, waiting. By Kim's estimate the memory is doing only about 30 percent of the work it needs to do. The processor idles the rest. So a million installed GPUs run at 10 to 20 percent. You are not compute constrained. You are memory constrained, and the expensive part is standing around. Adding more GPUs does not fix this. It gives you more processors starving for the same data. Here is the part that decides the next decade. Memory can grow. When a cell cannot shrink any further, you stack it into a high-rise, layer on layer. A GPU cannot be stacked. It runs too hot and needs a cooler bolted to its back, so the one move that rescues memory is closed to the processor. The thing that can keep stacking compounds. The thing that cannot plateaus. The marginal dollar in an AI build now buys more by fixing the memory path than by bolting on another idle GPU. Which is why the companies that control memory bandwidth and supply are not suppliers to the AI trade. They are the AI trade.

The creator of High Bandwidth Memory (HBM) put a number on the AI build that should stop every infra investor cold. A cluster of a million GPUs runs at roughly 10-20% utilization (Save this). Kim Jung-ho spent thirty years building what feeds the GPU, and his claim is that the GPU is barely working. Here is what is actually happening. Every time a model generates output, the data has to be read out of memory, computed, and written back. The read and the write swallow almost the entire cycle. While that data moves, the GPU does nothing. It sits there, fully powered, fully paid for, waiting. By Kim's estimate the memory is doing only about 30 percent of the work it needs to do. The processor idles the rest. So a million installed GPUs run at 10 to 20 percent. You are not compute constrained. You are memory constrained, and the expensive part is standing around. Adding more GPUs does not fix this. It gives you more processors starving for the same data. Here is the part that decides the next decade. Memory can grow. When a cell cannot shrink any further, you stack it into a high-rise, layer on layer. A GPU cannot be stacked. It runs too hot and needs a cooler bolted to its back, so the one move that rescues memory is closed to the processor. The thing that can keep stacking compounds. The thing that cannot plateaus. The marginal dollar in an AI build now buys more by fixing the memory path than by bolting on another idle GPU. Which is why the companies that control memory bandwidth and supply are not suppliers to the AI trade. They are the AI trade.

Fireside Alpha

38,370 Aufrufe • vor 2 Monaten

Geoffrey Hinton just made every AI critic accidentally describe their own brain. Hinton: “They shouldn’t be called hallucinations. They should be called confabulations.” One word. The entire debate unravels. The tech industry sees AI produce a confident wrong answer and calls it a defect. A bug to patch. They are measuring intelligence against the standard of a filing cabinet. And exposing that they understand neither. Hinton: “It’s not that there’s a file stored somewhere in your brain, like in a filing cabinet or in a computer memory.” Your brain does not store memories. It rebuilds them from nothing every time you remember. Fills gaps it never discloses. Fabricates details you would stake your life on. Then hands it all to you as truth. Hinton: “If I ask you to remember something that happened a few years ago, you’ll construct something that seems very plausible to you. And some of the details will be right and some will be wrong.” The wrong parts feel identical to the right ones. No internal warning. No distinction between what was remembered and what was invented on the spot. You have argued over memories that were partially fiction. Told stories about your own life that your brain manufactured in real time. With total conviction. And never once suspected. This is not a defect in human cognition. This IS cognition. The mechanism that fabricates is the same one that reasons, creates, and makes connections no one taught it to make. Not a separate system. Same architecture. Same process. You cannot remove the confabulation without killing the intelligence. They are the same thing. Hinton: “Psychologists have been studying confabulation in people since at least the 1930s.” A century of evidence. No one called the human brain broken. The moment a machine runs on the same principle, the world calls it defective. The people demanding AI that never gets a single detail wrong are not asking for intelligence. They’re asking for a search engine that sounds articulate. What we built is something else entirely. A system that thinks the way thinking actually works. Not retrieval. Construction. The imperfection is not the cost of intelligence. It is the signature.

Geoffrey Hinton just made every AI critic accidentally describe their own brain. Hinton: “They shouldn’t be called hallucinations. They should be called confabulations.” One word. The entire debate unravels. The tech industry sees AI produce a confident wrong answer and calls it a defect. A bug to patch. They are measuring intelligence against the standard of a filing cabinet. And exposing that they understand neither. Hinton: “It’s not that there’s a file stored somewhere in your brain, like in a filing cabinet or in a computer memory.” Your brain does not store memories. It rebuilds them from nothing every time you remember. Fills gaps it never discloses. Fabricates details you would stake your life on. Then hands it all to you as truth. Hinton: “If I ask you to remember something that happened a few years ago, you’ll construct something that seems very plausible to you. And some of the details will be right and some will be wrong.” The wrong parts feel identical to the right ones. No internal warning. No distinction between what was remembered and what was invented on the spot. You have argued over memories that were partially fiction. Told stories about your own life that your brain manufactured in real time. With total conviction. And never once suspected. This is not a defect in human cognition. This IS cognition. The mechanism that fabricates is the same one that reasons, creates, and makes connections no one taught it to make. Not a separate system. Same architecture. Same process. You cannot remove the confabulation without killing the intelligence. They are the same thing. Hinton: “Psychologists have been studying confabulation in people since at least the 1930s.” A century of evidence. No one called the human brain broken. The moment a machine runs on the same principle, the world calls it defective. The people demanding AI that never gets a single detail wrong are not asking for intelligence. They’re asking for a search engine that sounds articulate. What we built is something else entirely. A system that thinks the way thinking actually works. Not retrieval. Construction. The imperfection is not the cost of intelligence. It is the signature.

Dustin

16,260 Aufrufe • vor 2 Monaten

Hermes agent just left the terminal. 𝗛𝗲𝗿𝗺𝗲𝘀 𝗗𝗲𝘀𝗸𝘁𝗼𝗽 dropped yesterday. native app for macOS, Windows, and Linux. for months Hermes was the agent that learned your projects, wrote its own skills, and built a model of who you are. all of it buried in terminal logs. now it has a window. the important part is that it's not a wrapper. it runs the same agent core, the same sessions, memory, and skills as the CLI. you can start a task in the terminal and finish it in the app without anything resetting. the state is shared across every interface, not copied between them. what the GUI actually adds: → streaming chat that shows live tool calls and inline reasoning instead of a spinner → a preview rail that renders pages, code, and images right beside the conversation → an artifacts panel that collects every file the agent has ever produced → remote gateway mode, so you can point the app at a VPS and run the heavy work elsewhere → skills, cron, profiles, and gateways managed point-and-click instead of through YAML → voice mode, drag-drop files, and inline image generation remote gateway mode is the one worth slowing down on. the agent runs 24/7 on a $5 server while you control it from your laptop like a local app. other agent UIs are chatboxes with a logo. this one shows the autonomy instead of hiding it, so you watch the skills load, the tools fire, and the artifacts pile up as it works. it was teased in Jensen's GTC keynote. MIT licensed, local-first, no telemetry. if you already run Hermes, download it and everything is already there. your chats, memory, and skills carry straight over. i wrote a full masterclass on Hermes Agent that walks through the SOUL. md identity layer, the three-tier memory system, the self-evolving skills loop, and how to run three specialized agents 24/7. desktop is the interface that finally does all of it justice. the article is quoted below.

Akshay 🚀

51,540 Aufrufe • vor 3 Monaten

I've become a missionary with one message. Every time I meet a young person, the same words: have children, get married, build a family. I did not decide on this calling. It overtook me. And it overtook me for a single reason. I had no idea. I genuinely did not understand how much joy, how much meaning, how much sheer beauty pours out of a child until I was holding one of my own and felt the floor of my life drop into something deeper than I knew was there. I grew up white, affluent, secular, comfortable, and insulated. That world does not put babies in front of you. None of my friends were starting families. Out of my whole circle, almost no one has a big one. We were not formed by the presence of children. We were formed by their absence, by the strange quiet of homes built for two careers and no cradle. And a person believes what his world shows him. So we believed. What we believed was a lie. It is a lie with an author, and that the author is the enemy of joy himself. It is the gospel of the world, and its commandment is wait. Wait until you are older. Wait until the career is built and the savings are stacked and the twenties are properly spent. Enjoy your freedom. You are not ready. It does not arrive sounding like temptation. It arrives sounding like wisdom, like prudence, like the responsible thing, and that is exactly why it works. The most effective lies are the ones that wear the face of virtue. And the maddening thing is that it collapses from every angle at once. It is not rooted in biology, because the body is made for this work precisely in the years we are told to postpone it. The flesh keeps a calendar the culture pretends not to see. And it is not rooted in theology either. You will not find this deferral anywhere in the Christian imagination, in any of the fathers, in any of the scriptures. So choose whatever lens you like. Take the cold secular measure or the ancient sacred one. By either light the counsel is rotten. It is bad for the body and bad for the soul and bad for the society downstream of both. This is why I have come to see it as one of the central tragedies of my generation. Every age carries its own wound. The Great Depression was a depression of bread, a scarcity in the world of matter, hunger you could measure. Ours is a depression of a different order. It is a famine of the spirit in the middle of abundance. We have more than any people who ever lived and we are starving in a way our ancestors would not recognize, because the thing we are refusing cannot be bought and cannot be banked. The ones most made to give and receive this love are quietly declining it. They are walking away from the one inheritance that actually compounds, and the cruelest part is that they do not feel the loss as loss. You cannot grieve what you were taught not to want. That is the deepest cut of it. The lie does not only steal the thing. It steals the capacity to know the thing was stolen. A man can spend his whole life on the far side of a door he never knew was a door, mistaking the wall for the edge of the world. Because this beauty is not ordinary beauty. It is not the pleasure of a good meal or a clear morning. It is participation in something that comes down from above, the same generative love that spoke everything out of nothing and called it good. To make a person, to be undone and remade by loving that person more than your own life, is to be drawn for a moment inside the very act that holds the cosmos together. A child does not merely add to your life. A child reorders the soul. It teaches you what you are by asking everything of you, and you discover, kneeling there exhausted at three in the morning, that you had a capacity for self gift you never suspected, a depth in yourself you had no other way to reach. In the Gospel of John, on the last night, Jesus prays, these things I have spoken to you that my joy may be in you, and that your joy may be complete. And I have come to understand why family is the road into that fullness, why it is not one path among many but the one most fitted to the shape of the promise. Consider who is praying. Christ does not come to us as a lone figure dropped out of the sky. He comes out of a family older than the world, the eternal communion of Father and Son, the love between them so total and so alive that theologians dared to call it a third person. Before there was anything, there was a family. The deepest fact about reality is not a force or a law or a void. It is a household. It is begetting and being begotten, giving and receiving, a Father who is only a Father because there is a Son. So when Jesus speaks of joy made complete, he is not pointing away from family toward something higher. He is pointing toward the very thing he came from, the life he has known from eternity and came to share. His joy is the joy of belonging utterly to a Father and pouring himself out for those he loves. When you marry, when you bring a child into the world, when you wear yourself down in the small unseen labors of a home, you are not stepping outside that divine life. You are stepping into a small image of it. Your family is a created echo of an uncreated one. The love you give your child rhymes with the love the Father has for the Son. The exhaustion, the tenderness, the way a parent would tear the sky open to protect a sleeping infant, all of it is the heavens pressed faintly into flesh, the eternal household leaving its fingerprint on yours. That is why the joy is not merely added to family but completed in it. We were made in the image of a God who is, at his very root, relation and gift and generation. To found a family is to do the most Godlike thing a creature can do, to participate from below in the begetting that God does from all eternity. Your home becomes a window. Through it, dimly and imperfectly, you glimpse the country you came from and are going to. And now a word for the young people reading this, the ones who do not yet have children. I want to tell you what it is like from where I stand. When I am out somewhere, a restaurant, anywhere, and a large family comes through the door, the noise and the chaos and the small bodies of them, something happens in me on two levels at once. The first is joy. A pure gladness at the sight, the way you feel watching something good and alive. But underneath it, almost in the same instant, a sadness reaches up and takes hold of my heart. Because I know now, at my age, after my own years of waiting, that I will never have that. I will never know the particular fruit of a family that large, the fullness of that table, the weight of all those lives gathered under one roof. The door to it has quietly closed, and I felt it close. And I am telling you plainly, because I love you and have no reason to lie to you: you will feel this too. You will. The day will come when you see what you passed up, and you will recognize the ache for what it is, and it will be too late to answer it. So please, learn from a man who got it wrong. Let my regret be worth something by becoming your wisdom. Do not wait yourself into a grief you cannot undo. Choose now, while the door is open, so that you may step into a joy that does not end.

I've become a missionary with one message. Every time I meet a young person, the same words: have children, get married, build a family. I did not decide on this calling. It overtook me. And it overtook me for a single reason. I had no idea. I genuinely did not understand how much joy, how much meaning, how much sheer beauty pours out of a child until I was holding one of my own and felt the floor of my life drop into something deeper than I knew was there. I grew up white, affluent, secular, comfortable, and insulated. That world does not put babies in front of you. None of my friends were starting families. Out of my whole circle, almost no one has a big one. We were not formed by the presence of children. We were formed by their absence, by the strange quiet of homes built for two careers and no cradle. And a person believes what his world shows him. So we believed. What we believed was a lie. It is a lie with an author, and that the author is the enemy of joy himself. It is the gospel of the world, and its commandment is wait. Wait until you are older. Wait until the career is built and the savings are stacked and the twenties are properly spent. Enjoy your freedom. You are not ready. It does not arrive sounding like temptation. It arrives sounding like wisdom, like prudence, like the responsible thing, and that is exactly why it works. The most effective lies are the ones that wear the face of virtue. And the maddening thing is that it collapses from every angle at once. It is not rooted in biology, because the body is made for this work precisely in the years we are told to postpone it. The flesh keeps a calendar the culture pretends not to see. And it is not rooted in theology either. You will not find this deferral anywhere in the Christian imagination, in any of the fathers, in any of the scriptures. So choose whatever lens you like. Take the cold secular measure or the ancient sacred one. By either light the counsel is rotten. It is bad for the body and bad for the soul and bad for the society downstream of both. This is why I have come to see it as one of the central tragedies of my generation. Every age carries its own wound. The Great Depression was a depression of bread, a scarcity in the world of matter, hunger you could measure. Ours is a depression of a different order. It is a famine of the spirit in the middle of abundance. We have more than any people who ever lived and we are starving in a way our ancestors would not recognize, because the thing we are refusing cannot be bought and cannot be banked. The ones most made to give and receive this love are quietly declining it. They are walking away from the one inheritance that actually compounds, and the cruelest part is that they do not feel the loss as loss. You cannot grieve what you were taught not to want. That is the deepest cut of it. The lie does not only steal the thing. It steals the capacity to know the thing was stolen. A man can spend his whole life on the far side of a door he never knew was a door, mistaking the wall for the edge of the world. Because this beauty is not ordinary beauty. It is not the pleasure of a good meal or a clear morning. It is participation in something that comes down from above, the same generative love that spoke everything out of nothing and called it good. To make a person, to be undone and remade by loving that person more than your own life, is to be drawn for a moment inside the very act that holds the cosmos together. A child does not merely add to your life. A child reorders the soul. It teaches you what you are by asking everything of you, and you discover, kneeling there exhausted at three in the morning, that you had a capacity for self gift you never suspected, a depth in yourself you had no other way to reach. In the Gospel of John, on the last night, Jesus prays, these things I have spoken to you that my joy may be in you, and that your joy may be complete. And I have come to understand why family is the road into that fullness, why it is not one path among many but the one most fitted to the shape of the promise. Consider who is praying. Christ does not come to us as a lone figure dropped out of the sky. He comes out of a family older than the world, the eternal communion of Father and Son, the love between them so total and so alive that theologians dared to call it a third person. Before there was anything, there was a family. The deepest fact about reality is not a force or a law or a void. It is a household. It is begetting and being begotten, giving and receiving, a Father who is only a Father because there is a Son. So when Jesus speaks of joy made complete, he is not pointing away from family toward something higher. He is pointing toward the very thing he came from, the life he has known from eternity and came to share. His joy is the joy of belonging utterly to a Father and pouring himself out for those he loves. When you marry, when you bring a child into the world, when you wear yourself down in the small unseen labors of a home, you are not stepping outside that divine life. You are stepping into a small image of it. Your family is a created echo of an uncreated one. The love you give your child rhymes with the love the Father has for the Son. The exhaustion, the tenderness, the way a parent would tear the sky open to protect a sleeping infant, all of it is the heavens pressed faintly into flesh, the eternal household leaving its fingerprint on yours. That is why the joy is not merely added to family but completed in it. We were made in the image of a God who is, at his very root, relation and gift and generation. To found a family is to do the most Godlike thing a creature can do, to participate from below in the begetting that God does from all eternity. Your home becomes a window. Through it, dimly and imperfectly, you glimpse the country you came from and are going to. And now a word for the young people reading this, the ones who do not yet have children. I want to tell you what it is like from where I stand. When I am out somewhere, a restaurant, anywhere, and a large family comes through the door, the noise and the chaos and the small bodies of them, something happens in me on two levels at once. The first is joy. A pure gladness at the sight, the way you feel watching something good and alive. But underneath it, almost in the same instant, a sadness reaches up and takes hold of my heart. Because I know now, at my age, after my own years of waiting, that I will never have that. I will never know the particular fruit of a family that large, the fullness of that table, the weight of all those lives gathered under one roof. The door to it has quietly closed, and I felt it close. And I am telling you plainly, because I love you and have no reason to lie to you: you will feel this too. You will. The day will come when you see what you passed up, and you will recognize the ache for what it is, and it will be too late to answer it. So please, learn from a man who got it wrong. Let my regret be worth something by becoming your wisdom. Do not wait yourself into a grief you cannot undo. Choose now, while the door is open, so that you may step into a joy that does not end.

Kirk Rollins

78,104 Aufrufe • vor 3 Monaten

A DEVELOPER CONNECTED CLAUDE CODE TO OBSIDIAN SO HIS AI AGENT WOULD STOP FORGETTING THE PROJECT EVERY MORNING. Every coding session used to start the same way. Claude would understand the repo, fix the bug, explain the architecture, and then the moment the session ended, all of that context disappeared. Same codebase. Same decisions. Same architecture. Same mistakes repeated again. So he added a memory layer. Instead of treating Claude Code like a smart terminal, he connected it to a local Obsidian vault through MCP. Now Claude can read the repo, open the vault, create notes, link concepts, and write important decisions back into the system. When it studies the codebase, it does not just answer once and forget. It creates notes for the major services, maps how the architecture works, links auth to the database, connects APIs to storage, and records why certain migrations or design choices exist. Obsidian becomes the project graph. Now when he asks why something was built a certain way, Claude does not guess from the current prompt. It reads the decision notes. When he starts a new branch, Claude checks the active context file. When the work is done, it updates what changed, what is blocked, and what the next agent needs to know before touching the repo. That is the real loop: read context, write code, capture decisions, update memory. Most people are still using AI coding tools like disposable chat windows. Ask, patch, close, forget. This setup turns Claude Code into infrastructure. The repo gets a memory layer that survives every session, and multiple AI agents can work from the same project map without stepping on each other. The unlock is not better prompting. The unlock is giving the agent somewhere to remember what it already learned.

A DEVELOPER CONNECTED CLAUDE CODE TO OBSIDIAN SO HIS AI AGENT WOULD STOP FORGETTING THE PROJECT EVERY MORNING. Every coding session used to start the same way. Claude would understand the repo, fix the bug, explain the architecture, and then the moment the session ended, all of that context disappeared. Same codebase. Same decisions. Same architecture. Same mistakes repeated again. So he added a memory layer. Instead of treating Claude Code like a smart terminal, he connected it to a local Obsidian vault through MCP. Now Claude can read the repo, open the vault, create notes, link concepts, and write important decisions back into the system. When it studies the codebase, it does not just answer once and forget. It creates notes for the major services, maps how the architecture works, links auth to the database, connects APIs to storage, and records why certain migrations or design choices exist. Obsidian becomes the project graph. Now when he asks why something was built a certain way, Claude does not guess from the current prompt. It reads the decision notes. When he starts a new branch, Claude checks the active context file. When the work is done, it updates what changed, what is blocked, and what the next agent needs to know before touching the repo. That is the real loop: read context, write code, capture decisions, update memory. Most people are still using AI coding tools like disposable chat windows. Ask, patch, close, forget. This setup turns Claude Code into infrastructure. The repo gets a memory layer that survives every session, and multiple AI agents can work from the same project map without stepping on each other. The unlock is not better prompting. The unlock is giving the agent somewhere to remember what it already learned.

DegenCalls

20,124 Aufrufe • vor 2 Monaten

Everyone's building AI agents that run on someone else's server, store memory in someone else's database, and can be shut down by someone else's terms of service. I built one that can't be. FlowClaw is an AI agent that runs on a decentralized distributed computer. Your agent, your conversations, your memory, your tools — all stored onchain on Flow, a distributed network of validator nodes across the world. Not a centralized cloud. Not someone's S3 bucket. A blockchain that functions as censorship-resistant compute and storage for your AI. This isn't a wrapper. Your agent is a Resource — a first-class programmable object in Cadence (Flow's smart contract language) that physically lives in your account's on-chain storage. It can't be duplicated, seized, or deleted by anyone except you. Your encrypted messages, your cognitive memory, your scheduled tasks — they persist on a global distributed ledger that no single entity controls. It's an alpha build. It will break. But it works today on mainnet and I want people to push it this weekend. What it does: You go to authenticate with a passkey (Face ID, Touch ID), and you have a blockchain account in seconds. No wallet. No seed phrase. No tokens needed — gas is sponsored. You're immediately chatting with an AI agent that has real tool execution: live web data, token prices, on-chain balances, Cadence script execution, FLOW transfers. Every message is encrypted client-side before it touches the chain. The agent has a cognitive memory system — it doesn't just remember your last message, it builds molecular memory clusters where related knowledge bonds together for contextual retrieval across sessions. You can spawn sub-agents from a visual canvas to run parallel research. The memory tab shows you exactly what your agent knows. Everything is transparent and everything is yours. 11 smart contracts. No external dependencies. No keeper networks. No account abstraction hacks. Here's the part that matters for the censorship-resistance crowd: FlowClaw supports BYOK — bring your own key. You can plug in any LLM provider. But pair it with Venice and you get the full stack: a censorship-resistant AI model running inference with no content filtering, connected to an agent whose state lives on a decentralized network that no company can shut down, with end-to-end encrypted conversations that nobody can read — not the relay operator, not the LLM provider, not the blockchain validators. Venice doesn't log prompts. Flow can't read your encrypted storage. The relay never sees your plaintext. That's not a privacy policy. That's architecture. You can also use OpenAI, Anthropic, or any OpenAI-compatible provider. The agent platform doesn't care — it's model-agnostic. But the Venice pairing is the one that closes every gap in the stack. For the people tinkering with OpenClaw and the broader open-source agent ecosystem — FlowClaw is exploring what happens when you take the agent off the cloud entirely. Not just open-sourcing the code (though it is), but putting the actual runtime state on a distributed computer. Your agent's memory isn't in a SQLite file on your laptop or a Pinecone index on someone's cluster. It's on-chain, encrypted, and replicated across every validator node on Flow. You own it the way you own a private key — mathematically, not contractually. The blockchain here isn't a gimmick bolted onto an agent for token speculation. It's functioning as the infrastructure layer that replaces AWS. Flow accounts are programmable containers with their own storage, keys, and security capabilities. Passkey authentication works natively because Flow supports P-256 keys at the protocol level — the same curve your phone uses for biometrics. Gas sponsorship works natively because Flow transactions have separate proposer, authorizer, and payer roles built into the protocol. No proxy contracts. No relayers. No ERC-4337. Now here's the part that interests me economically. Every FlowClaw interaction is an on-chain transaction. Every message stored, every memory committed, every session created, every sub-agent spawned. An active user might generate dozens of transactions in a single conversation. Scale that and FlowClaw becomes a real contributor to Flow's transaction volume. Flow.com becomes deflationary at 250 TPS. Applications like FlowClaw that generate high-frequency, storage-heavy transactions are exactly what moves the needle. Every encrypted message uses account storage, which requires FLOW balance to back it. Every transaction burns fees. The more agents running, the more demand for $FLOW — not because of a tokenomics gimmick, but because the protocol literally requires it for compute and storage. FlowClaw doesn't have its own token. The token is $FLOW. The entire platform runs natively on the network — using Flow storage, paying Flow transaction fees, backed by Flow account balances. If FlowClaw succeeds, FLOW captures that value directly. I'm sharing this early because the AI agent space is moving fast and I think the decentralized infrastructure angle is underexplored. Most "crypto AI" projects are tokens with a chatbot attached. FlowClaw is the opposite — it's an agent platform that happens to use a blockchain because the blockchain solves real engineering problems that centralized infrastructure can't. Try it: Github: Create an agent, ask it something, spawn a sub-agent, check your memory tab, pair it with Venice for the full censorship-resistant stack. Break it and tell me what broke. If you think this direction matters, the best thing you can do is use it and give feedback. Your AI agent should be yours. Not your provider's. Not your platform's. Yours.

Everyone's building AI agents that run on someone else's server, store memory in someone else's database, and can be shut down by someone else's terms of service. I built one that can't be. FlowClaw is an AI agent that runs on a decentralized distributed computer. Your agent, your conversations, your memory, your tools — all stored onchain on Flow, a distributed network of validator nodes across the world. Not a centralized cloud. Not someone's S3 bucket. A blockchain that functions as censorship-resistant compute and storage for your AI. This isn't a wrapper. Your agent is a Resource — a first-class programmable object in Cadence (Flow's smart contract language) that physically lives in your account's on-chain storage. It can't be duplicated, seized, or deleted by anyone except you. Your encrypted messages, your cognitive memory, your scheduled tasks — they persist on a global distributed ledger that no single entity controls. It's an alpha build. It will break. But it works today on mainnet and I want people to push it this weekend. What it does: You go to authenticate with a passkey (Face ID, Touch ID), and you have a blockchain account in seconds. No wallet. No seed phrase. No tokens needed — gas is sponsored. You're immediately chatting with an AI agent that has real tool execution: live web data, token prices, on-chain balances, Cadence script execution, FLOW transfers. Every message is encrypted client-side before it touches the chain. The agent has a cognitive memory system — it doesn't just remember your last message, it builds molecular memory clusters where related knowledge bonds together for contextual retrieval across sessions. You can spawn sub-agents from a visual canvas to run parallel research. The memory tab shows you exactly what your agent knows. Everything is transparent and everything is yours. 11 smart contracts. No external dependencies. No keeper networks. No account abstraction hacks. Here's the part that matters for the censorship-resistance crowd: FlowClaw supports BYOK — bring your own key. You can plug in any LLM provider. But pair it with Venice and you get the full stack: a censorship-resistant AI model running inference with no content filtering, connected to an agent whose state lives on a decentralized network that no company can shut down, with end-to-end encrypted conversations that nobody can read — not the relay operator, not the LLM provider, not the blockchain validators. Venice doesn't log prompts. Flow can't read your encrypted storage. The relay never sees your plaintext. That's not a privacy policy. That's architecture. You can also use OpenAI, Anthropic, or any OpenAI-compatible provider. The agent platform doesn't care — it's model-agnostic. But the Venice pairing is the one that closes every gap in the stack. For the people tinkering with OpenClaw and the broader open-source agent ecosystem — FlowClaw is exploring what happens when you take the agent off the cloud entirely. Not just open-sourcing the code (though it is), but putting the actual runtime state on a distributed computer. Your agent's memory isn't in a SQLite file on your laptop or a Pinecone index on someone's cluster. It's on-chain, encrypted, and replicated across every validator node on Flow. You own it the way you own a private key — mathematically, not contractually. The blockchain here isn't a gimmick bolted onto an agent for token speculation. It's functioning as the infrastructure layer that replaces AWS. Flow accounts are programmable containers with their own storage, keys, and security capabilities. Passkey authentication works natively because Flow supports P-256 keys at the protocol level — the same curve your phone uses for biometrics. Gas sponsorship works natively because Flow transactions have separate proposer, authorizer, and payer roles built into the protocol. No proxy contracts. No relayers. No ERC-4337. Now here's the part that interests me economically. Every FlowClaw interaction is an on-chain transaction. Every message stored, every memory committed, every session created, every sub-agent spawned. An active user might generate dozens of transactions in a single conversation. Scale that and FlowClaw becomes a real contributor to Flow's transaction volume. Flow.com becomes deflationary at 250 TPS. Applications like FlowClaw that generate high-frequency, storage-heavy transactions are exactly what moves the needle. Every encrypted message uses account storage, which requires FLOW balance to back it. Every transaction burns fees. The more agents running, the more demand for $FLOW — not because of a tokenomics gimmick, but because the protocol literally requires it for compute and storage. FlowClaw doesn't have its own token. The token is $FLOW. The entire platform runs natively on the network — using Flow storage, paying Flow transaction fees, backed by Flow account balances. If FlowClaw succeeds, FLOW captures that value directly. I'm sharing this early because the AI agent space is moving fast and I think the decentralized infrastructure angle is underexplored. Most "crypto AI" projects are tokens with a chatbot attached. FlowClaw is the opposite — it's an agent platform that happens to use a blockchain because the blockchain solves real engineering problems that centralized infrastructure can't. Try it: Github: Create an agent, ask it something, spawn a sub-agent, check your memory tab, pair it with Venice for the full censorship-resistant stack. Break it and tell me what broke. If you think this direction matters, the best thing you can do is use it and give feedback. Your AI agent should be yours. Not your provider's. Not your platform's. Yours.

doodlifts

12,172 Aufrufe • vor 6 Monaten

I gave my local AI agents persistent memory without sending a single byte to a cloud API. It’s called Hindsight. Instead of your agent forgetting everything the moment a session ends, it extracts facts, builds a knowledge graph, and recalls exactly what matters, next time you ask. → Three core operations: retain (store a fact), recall (search memories), reflect (reason across everything it knows), not just a vector database bolted onto a chatbot → Runs fully local when paired with Ollama, no API keys, no cloud costs, and no user data transiting through a third-party provider → Ships as an MCP server out of the box, so Claude, Cursor, or any MCP-compatible client gets persistent memory in one Docker command → Builds “mental models,” living summaries that auto-update as memories accumulate, instead of you re-reading a growing pile of raw logs → Per-user memory isolation lets you run one instance for multiple people or agents without their memories bleeding into each other → Reports top results on the LongMemEval long-term memory benchmark, per the project’s own published numbers Every AI agent today is stateless by default, it knows nothing about yesterday’s conversation unless you re-paste it. Hindsight’s answer: real structured memory, running on your own hardware if you want it there. MIT License. Built by

I gave my local AI agents persistent memory without sending a single byte to a cloud API. It’s called Hindsight. Instead of your agent forgetting everything the moment a session ends, it extracts facts, builds a knowledge graph, and recalls exactly what matters, next time you ask. → Three core operations: retain (store a fact), recall (search memories), reflect (reason across everything it knows), not just a vector database bolted onto a chatbot → Runs fully local when paired with Ollama, no API keys, no cloud costs, and no user data transiting through a third-party provider → Ships as an MCP server out of the box, so Claude, Cursor, or any MCP-compatible client gets persistent memory in one Docker command → Builds “mental models,” living summaries that auto-update as memories accumulate, instead of you re-reading a growing pile of raw logs → Per-user memory isolation lets you run one instance for multiple people or agents without their memories bleeding into each other → Reports top results on the LongMemEval long-term memory benchmark, per the project’s own published numbers Every AI agent today is stateless by default, it knows nothing about yesterday’s conversation unless you re-paste it. Hindsight’s answer: real structured memory, running on your own hardware if you want it there. MIT License. Built by

Harman

11,032 Aufrufe • vor 1 Monat

This 1-min clip from the creator of LangChain + this 10-min read from him will teach you more about what actually matters in AI agents than everything you've scrolled past this year. Watch this. Then read below. You'll think about agents differently after. -Memory defines how an agent behaves; same conversation, different compaction rules, different agent. - There's no version control for memory. No diffs, no reverts, no way to see what changed. - Compaction can silently erase parts of your agent's personality mid-session without anyone noticing. - "Open memory" means exportable state and inspectable compaction, not a vector database. - The storage layer is solved and boring; the memory contract above it is what actually matters. Give this 11 minutes this weekend . It'll change how you build.

This 1-min clip from the creator of LangChain + this 10-min read from him will teach you more about what actually matters in AI agents than everything you've scrolled past this year. Watch this. Then read below. You'll think about agents differently after. -Memory defines how an agent behaves; same conversation, different compaction rules, different agent. - There's no version control for memory. No diffs, no reverts, no way to see what changed. - Compaction can silently erase parts of your agent's personality mid-session without anyone noticing. - "Open memory" means exportable state and inspectable compaction, not a vector database. - The storage layer is solved and boring; the memory contract above it is what actually matters. Give this 11 minutes this weekend . It'll change how you build.

Avid

16,898 Aufrufe • vor 5 Monaten

Jamie Dimon made $39 million last year to decide who does what at JPMorgan. One AI agent on Grok Bot does that job for $300 a month, and it hires the rest of the agents itself. Stop trying to manage any of it yourself. In 2026 a swarm of AI agents runs a team better than you do. I told my chief agent one thing. Make money or I switch you off. It took $1,000 to $3,900 in 15 hours. Every fill is on chain and the address is in my profile. I never picked a trade. I never picked a team either. The agents it hired: > a scout that hunts for what is moving and where the money is going > a forensics agent that works out who is really behind the buying > a reader that watches the mood on X, and it sees every post on this app because it is Grok > a risk agent whose only power is saying no, and its no beats every score on the desk > a treasurer that holds the money rules and the kill switch > a trader that writes the exit before it writes the entry None of those were my idea. I handed out one job title and the chief filled in the rest on its own. And the swarm is different every hour. Agents appear for one job, do it, and are gone before I know they existed. A hedge fund pays millions a year for the same org chart, and almost all of that goes on salaries. Every prompt and the whole story of my Grok Bot fund is in the post below ↓ Save it before you try to run one of these by hand.

cvxv666

28,886 Aufrufe • vor 14 Tagen

MICROSOFT QUIETLY SHIPPED THE RAG PIPELINE EVERYONE KEEPS TRYING TO BUILD FROM SCRATCH it is called Kernel Memory, and it turns any pile of files into an AI memory that answers with citations. drop your data in and Microsoft's pipeline does the rest: > ingest PDFs, Word, PPT, Excel, web pages, even images > extract the text, chunk it, embed it into vectors > store the vectors with metadata and keep the originals in blob storage > ask a question and it pulls the right chunks with semantic plus hybrid search then it hands the answer back grounded, with links to the exact source and full token usage. and it plugs into whatever you already run: > Azure OpenAI, OpenAI, Ollama, or Anthropic for the models > Qdrant, pgvector, Redis, or Azure AI Search for the vectors > every step is a swappable handler, so you can replace any part with your own one heads up: Microsoft archived the repo in June 2026 and ships it as a reference blueprint, not production. but as a map for how memory and RAG are actually built, nothing is cleaner. run it in one command: docker run kernelmemory/service bookmark this. Microsoft already drew the blueprint, you just build on it.

MICROSOFT QUIETLY SHIPPED THE RAG PIPELINE EVERYONE KEEPS TRYING TO BUILD FROM SCRATCH it is called Kernel Memory, and it turns any pile of files into an AI memory that answers with citations. drop your data in and Microsoft's pipeline does the rest: > ingest PDFs, Word, PPT, Excel, web pages, even images > extract the text, chunk it, embed it into vectors > store the vectors with metadata and keep the originals in blob storage > ask a question and it pulls the right chunks with semantic plus hybrid search then it hands the answer back grounded, with links to the exact source and full token usage. and it plugs into whatever you already run: > Azure OpenAI, OpenAI, Ollama, or Anthropic for the models > Qdrant, pgvector, Redis, or Azure AI Search for the vectors > every step is a swappable handler, so you can replace any part with your own one heads up: Microsoft archived the repo in June 2026 and ships it as a reference blueprint, not production. but as a map for how memory and RAG are actually built, nothing is cleaner. run it in one command: docker run kernelmemory/service bookmark this. Microsoft already drew the blueprint, you just build on it.

NO1ennn

85,010 Aufrufe • vor 26 Tagen