← Back to search
Hermes Obsidian is Insane (FREE!) 🤯
AI News Today | Julian Goldie Podcast · 2026-10-11 · 13 min
Show full episode description
Hermes Obsidian is Insane (FREE!) 🤯
✨ Episode Outline — click any point to jump to it in the episode
Problem solved
Giving
Hermes,
Claude Code,
Codex and
OpenClaw one shared, self-maintaining memory so agents pick up where each other left off without hand-editing files.
Benefits
- Shared Obsidian vault syncs memory across Hermes, Claude Code, Codex, OpenClaw
- Nightly daily-log task written by each agent, no manual editing
- One mega prompt to an orchestrator sets up every agent
- Hermes self-improvement learnings benefit all agents, not just Hermes
- Free, open source Obsidian vault instead of a vector database
Use cases
- Agent Daily Log: each agent writes worked-on/finished/learned summary to one shared note nightly
- PARA vault (Projects, Areas, Resources, Archive) plus inbox, daily, memories and agent OS folders
- Hermes Apollo voice agent reads the vault and recaps the day's work on request
- Agent OS 'create' tab turns recent memories into an app, SEO article, post or script
- Morning brief, pre-task check, weekly roll-up and vault health-check prompts
KPIs / results
- Memory galaxy with thousands of linked memories
- Any agent can read the daily note in under a minute
- Agent OS set up for Claude Code, Hermes, Codex, OpenClaw via one prompt
So today I want to talk you through how I'm using Hermes Agent, a simple system you can use to basically improve your Hermes Agent with a memory system. And this works across all of your agents. So Hermes Codex, OpenClaw, ClawedCode, whatever you want to use, this system can work. And you can see our memory galaxy here. This has thousands of memories plugged into it. We can also create new apps based on the system as well, which I'll come onto in a second. And it's a really powerful system. Now I call this the Agent Daily Log System. And basically what we're doing with Obsidian, which is a free open source memory system, is every night each agent can write a summary of its day into one shared Obsidian note. And the summary covers what it worked on, what it finished, what it learned. And the next morning any agent can read that note and pick up where the others left are. So you don't need to edit a single file yourself and one agent sets everything up for the rest. Now, if we have a look at Obsidian directly, let me show you what this looks like when we actually visualize it. So you can see it right here and everything's linked together. You can see that it accounts for people I'm working with. Even like it's got my, for example, personal trainer, that sort of thing. And it all links together as you can see. So for example, if we go to Facebook, you can see that links to our funnel strategy and about me and everything else. And so this whole system beautifully links together. Every single part of it is a new memory. And for example, we click on Julian Goldie Agency here. You can see it has all the details about me, who I am and everything else. Now, the great thing about this as well is like I don't have to manage it myself. I don't have to organize it myself because we're using our AI agents to keep it updated. And so here's a way that I recommend setting up Obsidian with Hermes agent. Now, if you're not, so how does this work? Basically, we have something called Para and Para is a way of organizing files. And that's basically all Obsidian is. So that's all a memory system is files and its folders, but neatly organized as you can see right here. So if we go back to Obsidian and the setup here, the way that you want to understand this, and this is crucial for understanding how to set up your memory system, is that Para stands for Projects, Areas, Resources, and Archive. Now, we can add an inbox, a daily folder for your own notes, a memories folder for decisions, and an agentic OS folder that belongs to your agents. The reason that I recommend having an agent OS is so that everything can sync. So, for example, we go to Claude here, we've got Claude ready to go, we've got Hermes, we've got Antigravity. It's all ready to go inside one system, right? And everything syncs together. So if we use, for example, Hermes Oracle, if we're using our voice agent with Hermes Apollo, anything like that, we can have ready to go and it all syncs directly into our memory vault, which is absolutely awesome. So this is the setup, the structure that I recommend giving to your agents. So your inbox, daily stuff, projects, areas, resources, memories, archive, and then you have the agent OS protocol as well. Now, inbox would be like messy captures, so agents can drop things here if it's unsure. Then we have your daily note, right? So like top three notes, notes for tomorrow, top three projects, et cetera. Then we have projects, active projects, one folder with an index MD in each. We have areas, so ongoing areas, like for example, for me, it's SEO, it's community, it's the AI profitable volume. And then we have resources. So for example, like about me, any SAPs, our voice, our style for content writing, and our memories, right? So our decisions, our log, our vaults, our wins, et cetera. And so that's all plugged into one system. And then what we can do over here is we can just paste one prompt and we can give this to our orchestrator agent to sync up all of our other agents, right? I don't think it's good to go into Hermes and then go into Claude and then go into Codex separately at this point to organize all your AI agents, right? So what I recommend instead is that you use one mega prompt. You can plug that into an orchestration agent that controls all your agents locally, like Claude Code or Codex. And you can basically say, set up a shared Obsidian memory for all my AI agents, do all the setup yourself. I don't want to edit these files by hand, right? And the reason that you want to pick one agent as the orchestrator, for example, Claude Code or Codex, and why they work best, is because both can read and write files across your machine. So you can paste this prompt in and change the vault path. The orchestrator builds the folders, writes a protocol, adds instructions to every other agent, and then tests each one. So you can see here, how it works. And then it basically tells each agent, okay, here's a vault, here's what each agent is going to do, here's the test and report, and here's the protocol. Now, one important thing to note, when you use this, specify which agents you use and you want to set this up for. So for me, for example, I could say, okay, set this up for Claude Code, for Hermes, for Codex, for OpenClaw. And you can see that we actually label that inside the prompt here. And then you can see the vault. Now, if we copy that, and then we go into, for example, Claude here, we start a new chat. We can paste that in. And then Opus 5.5 can see the prompt, inspect our vault, install agents. It's going to run the command across all of our agents, and then sync that in. And even if you have an existing install, Opus 5.5 is smart enough to look at that prompt and then figure out, okay, what's synced, what's not synced? How do we match this idea to the setup for Obsidian that the user already has? So whether you have an existing setup or whether you have a new setup, this prompt can work. And it's just one mega prompt that you can train your AI agents to use. Now, also, what's pretty cool about this is, for example, if we're using Hermes inside the chat, if we're using the voice agent, as you can see, we can basically use our vault and speak to our agent and say, Hello, sir. How may I help? Can you just look? Checking now, sir. Can you just look at my vault and give me some ideas of what we worked on today? The console's handling that, sir. One moment, sir. The console's handling that, sir. From your notes, you were generating video title ideas in Claude for your YT. See? And it can look at my notes, look at what we've done recently, and then use that inside the voice agent or inside Hermes, inside any other agent. And you can see that it's just working on the background and setting this up in place. So the prompt's written. Now it's going to start giving that to each agent as an instruction file. Pretty nice. Now, when we're doing this, the orchestrator is going to work out the details for each agent. Which instruction file it reads, where it keeps its history, how it schedules tasks. And you don't need to do any of that or worry about any of that. You can see it's working in the background right here. It's setting up the nightly daily log task. So two things land inside each agent from here, right? Each agent gets a short pointer in its instructions. So it knows the vault exists and reads it at the start of each session. It also gets a nightly task that writes its summary without you asking. And the rules themselves live in one vault, one note inside the vault, right? So you can see here, Hermes, Codex, OpenClaw, Claude Code. We see where the instructions are written. And it's got a nightly task for each one of these. Now, you could have it upload every single memory, every single conversation into Obsidian. But one of the biggest issues with that is if you're building out a lot of memories. And I'll show you an example right here. If you're building out a lot of memories into Obsidian, it's going to get too big, too overwhelming and use up too many tokens to use, right? So for example, like some people inside our community have asked, how do we add millions of memories to Obsidian? And Obsidian is not the right system to use for that. And also, if you have millions of memories, unless you're like a very big company, I think it's unnecessary, right? It's just using up extra tokens. You could store them as vectors and use it as a vector database, but it's way too technical for 99% of people, right? Something like this is just one prompt, copy and paste, and you set up a beautiful memory that syncs across all your agents. I think that's way better. Now, if you're wondering, okay, what does a daily log system look like? Basically, each agent writes a summary of its whole day. So it's not one entry per task. And then any agent can read the note in under a minute and know what happened, what was learned and what still opened. For example, Hermes, daily summary, worked on, learned, decisions in progress. Now, the reason that you want to capture learned is because you want it to be self-improving. Now, Hermes does have a self-improving layer, but you want that to be written into the vault. So it doesn't just benefit Hermes, it benefits all of your agents together. That's way more powerful. That's way more useful. Whereas, for example, you could have Hermes self-improving, but what about the rest of your agents? How are they going to learn from that? And the same for Claw Code and everything else. Now, if you wanted, for example, other prompts to use with this, you could have, for example, a morning brief. You could say, read yesterday's memories logs in Obsidian. Send me a five-bullet point brief covering what's done, what's stuck, and the top three things to focus on. Don't write anything to the vault. You could choose that for a morning brief system. You could say, log today now. Write your daily summary, blah, blah, blah. You could say, before a task, before you start, check the last three days of the agent OS memories for anything about my project. Tell me what's relevant. Then carry on. Or a weekly roll-up task could be super useful. Now, these are all just additional extras. I wouldn't use all of them. It's just these are giving you some ideas of how you can use it. And then also, you could do a health check on your overall Obsidian vault. So you could ask the orchestrator now and then. Check the Obsidian daily log. Set up across all my agents. List each agent's scheduled task. When it lasts run. And whether yesterday's log has its own section. Fix any task that failed. And tell me what you changed. So that's basically how it works, right? Then you've got a nightly summary. You could. There are some other ways which I'll come on to in a second. But one thing that I like to do as well inside the agent OS. And if you want to really take it to a whole new level, you can. This is what I'm doing, right? With the agent OS, what we can actually do here is we can go to the create section. And the create section can look at my recent memories. So sync to my Obsidian vault. And then we can create an app. Or an SEO article. Or a social media post. Or a script. Or a guide based on that memory, right? So for example, here you can see, okay, 10th of October, right? And so these are all the newest updates inside our memory vault. They're automatically synced. And so we've got a personalized app builder that knows everything I've worked on. Everything that I should automate. And then we can just click one button and automate. This is really useful for stuff like SEO as well. Or content creation. Because when you're creating content, quite often you're documenting what you're working on. And so if you've got a system where you can quickly create a video based on your recent projects and tasks, and you're documenting what you're working on, that's super useful as well. So you can see how many different ways you can use this. Now, there are other ways you can set this up. But if you want to keep this simple, which I always recommend, keep things as simple as possible. But I know there's some people out there who like to experiment. I know there's people who like to see what's possible. You could have the daily log. And that's a recommended option. And you just stick to that. You stay focused. You don't get overwhelmed. If you want to try some other cool stuff, you could, for example, have an entry after every task. You could have one file per agent per day. You'd have agents right into your own daily note. You'd have agents that file straight into Para. You'd have an auto wiki of people, tools, and topics. And you could have, and this is something I used to do, but I don't use it that much because I don't think you need to anymore. Agents are smart enough to do without this. But you could also have capture tools. So, for example, OMI, ScreenPipe. These are systems that can automatically look at what you're doing. And we have a lot of training on that inside the AI Profit Boardroom if you want to set that up. But basically, they look at your screen. They look at everything that you've said into your microphone, basically. And then they turn it into memories automatically that you can sync to Obsidian as well, which is really cool. But again, if you're on that daily log system with the full guide, you can get that inside the AI Profit Boardroom. It comes with the prompts and everything else. And also, if you want this full agentic operating system with the memory galaxy, you've got Hermes directly over here. We have Hermes Apollo, Hermes Oracle, all these custom core workflows you can use right here. And you've got the pop-up voice agent as well that you can speak to at any point. That's all inside the AI Profit Boardroom. Link in the comments description or just go to the AI Profit Boardroom.com. We are improving that agent OS system every single day, right? If you want all the prompts, a full guide from Obsidian and the agent OS system, you can get that inside here. It's my AI Automation Community that's focused on one thing, which is getting you more customers, more leads and saving your time with AI Automation. Inside the community, you get access to all of our best members. And you can also see all of our best trainings like you can see right here. So if you want the agent OS system, you can get that over here. If you want more training on Hermes, how to use it, you can get that over here as well. If you want to learn more about OMI and ScreenPipe and how they can automatically plug into your memory system, we have something called the Infinite Context Engine, which basically organizes exactly how to use it. And then if you want all the prompts in the full guide from today with everything else, you can get that inside the AI Profit Boardroom as well. And you can also direct message me or ask me any questions about this stuff, and I'm always happy to help. Now, inside the calendar, you get four-weekly coaching calls so you can ask questions, get help and support in real time. You can share screen, meet other members. Inside the map, you can meet people locally near you who are building out with AI agents like you. And this is all available inside the community. Also, inside this section, as you can see right here, we actually answer your latest questions to help you as much as you can. So you can post questions and we'll create video tutorials for you, as you can see right here, with my full step-by-step guides to help you, to get you support, and to help you as much as you can. And so that's all available inside the AI Profit Boardroom. Link in the comment description or go to the AI Profit Boardroom.com. Thanks for watching.