Hermes AI + Obsidian Just Got Simpler

Today I'm answering the latest questions about the Agent OS and these answers are going to make your AI way more powerful. First up, how to give your AI a memory that never forgets your business. Every chat, every idea, save forever. And it all comes down to one simple folder.

I'll show you mine running live and how we've stored over a thousand memories inside. Then I'll show you the setup where clients can book appointments by just calling a phone number the AI can answer, talk out loud and drop the book in straight into your calendar. And stick around because there's a two-minute fix later in this video that saves you hours of headaches. Let's get into it.

So these are all the latest questions from the AI Profit Boardroom. Each day what I do is sit down, answer them, give you a video tutorial like you see today. So if you want me to answer yours, feel free to post inside the AI Profit Boardroom. Link in the comments description or just go to the AIProfitBoardroom.com And the first question we have is from David.

So David says, do you need g-brain or hindsight for memory? And the answer is no. So if you're using, for example, Hermes or any single agent, you don't need g-brain, hindsight, anything like that. What I personally use is Obsidian.

Obsidian is really powerful, free app you can use. It can sync across all your devices and it's pretty easy to store stuff. So if we have a look as an example here, we've got the memory system as you can see. So this is the memory galaxy.

It stores all of my ideas and basically every time I interact with my agents, they do something useful, they log it into the system. Now this works in two ways as a positive feedback loop. So number one, what it actually does is that every agent will write into that vault, that Obsidian vault. And the great thing about that is that basically when we're using it, it can update it for us.

I don't need to organize it myself. We don't need to get like make it super messy. It can just answer everything. And that saves me a lot of admin.

And it also means that I create a system that I don't even have to think about. It's just running without me, which is beautiful. Now the other thing that we have is that we have the agent that can pull the data from that Obsidian memory, right? It can pull those memories out and then we can ask you like, what do you remember about my business from the Obsidian vault?

And from there it can use that to pull in memories. So basically you have one vault, all your agents plugged into it, Hermes as well, Claude as well, every chat auto logs inside it, every agent can read it back, nothing resets. And the thing that I would say here is like, you don't need more memory tools. So when you're using this sort of thing, some people think more memory tools equals better memory, but that's actually backwards.

One shared brain beats three partial one, because you want like focus. You want to keep your agents focused. And also this will reduce the amount of tokens used. If you have three different memory systems plugged into Hermes, it's going to confuse it, but also it's going to make it more difficult and time consuming to use Hermes because it's using up loads of tokens across three different memory systems.

So we've got another question here, which is about voice booking. So Jay Lane asks, you know, can clients book on the phone? Let's say for example, you've got some agents or some clients who don't want to use an app and they want to be able to book stuff on the phone. And basically Jay Lane is looking for a good live agent tool that can do this.

And he's wondering, you know, if Hermes and a voice will do the trick. So what you can actually do is you can connect, for example, Hermes to your phone, or you can have, for example, Hermes live on your website. And basically that can be connected to Google Workspace API. So method number one is that you could use Telegram, you could use WhatsApp, you could have that connected on your website.

And then with Google Workspace API connected to your calendar, Hermes can book in slots for you. If you say, okay, anytime a client requests an appointment, allow them to book it in via chat. So that's one method. And then that will actually go to your Google calendar and the booking gets saved.

The other option is that you could, for example, have Tailscale and the agent operating system connected to a voice agent. The client calls it. So you can, for example, we have a full tutorial on this. You can connect your voice agent to something like 11 labs.

And with 11 labs, Hermes can connect to it. People can call it. Obviously you would limit what is connected to for security, but basically you could allow it to access your calendar as well. And then you give it custom instructions in terms of how to answer people and that sort of thing.

So here's an example. We've got Hermes Apollo inside our agentic operating system. You can speak to it, you can ask it to do things for you, and then it can actually operate, for example, your calendar or anything else. So that same technology could answer a phone, it hears, it replies, and it can write what's agreed inside your tools that it's connected to.

If you want to connect a voice agent to Tailscale as well, we also have that system right here. So however you want to connect it. Now, three options. But if you want the simplest one, you just go with Telegram, WhatsApp, connect Hermes to Google Workspace, and then it can book calendar slots for you.

And that way, the phone number is the app, so it's non-technical to use. Anyone can use it. Anyone can write a message or anyone can call a number, but your agent can handle it for you. Next up, we've got a question about Obsidian, which is the memory vault I mentioned earlier.

And someone asks, can the vault live on an old machine? And the answer is absolutely yes, it can. Because all it is, is a folder. So it's just a vault, which is a folder of files.

And there's a few ways you could link this, right? Number one, is you sync the folder between machines. You can use Obsidian Sync, or you can use a synced folder, like for example, iCloud or Drive or whatever you want. The other option is that you could allow it to reach the old machine directly via Tailscale.

It's free, takes about 10 minutes, and that basically allows you to operate your agent OS from anywhere, which means you can connect it to an old laptop. And then the third option is that you can connect your memory system to an MCP. So for example, if we have a look at this memory system here, there is an MCP for Obsidian. You can just ask Hermes or whatever you want to set up with it.

And then you have the old machine with your vault, a folder of files, three ways to sync it, which is sync the folder via Cloud, Tailscale Direct, MCP or Login. And then that can link to the same machine, and you have the same system linked together inside of memory. So it's the same vault, just two machines synced together as a folder, reaching the old machine, and it all links together and updates automatically. Another question we got here from Thomas, which was, I completed the installation of the agent OS, however, when I try and open it, I can't access it.

So there's a couple options for this. Number one, it may be that it's not switched on. If the local host isn't switched on when you're setting up the agent OS system, you can actually just run the start command. So you can tell Claude or Hermes or whatever set up for you to run that and make sure it's running.

If it's already switched on, but you just can't access it on, say, localhost colon 3000, ask it which local address, which local host it is set up on, and then the agent will tell you which local host is set up the agent OS on, and you can just get access to it. So two different options right there. Now I've got a question from Greg, which is basically, I've been trying to get my memory to sync in two different workstations. So running the same instance of Claude on the personal workstation and also the agent OS workstation.

So trying to sync them together. So you want to pick like one sync engine, if that makes sense. So I see here that you said, I think I have to ask the agent to write a file at the end of each session to the Obsidian vault. Basically, the way you want to sync them together is with one login.

So I can log into Obsidian and then it's synced directly on two different machines or whatever machines I have it logged in. And that way we don't get any conflicting copies. So what we can do is we can go to our settings, go to sync, remote vault, and then just choose which remote vault we want access to. That will send us to this link, and then we can have everything synced together as you can see right here.

So that's the way that you would sync them. Just have a login, you configure the vaults, and you make sure that everything is synced properly with the sync controls inside Obsidian. And that way two machines don't run out of sync. By the way, if you want your questions answered like this in a video tutorial every single day, every question on this page came from a member inside the AI Profit Boardroom, and they get a personal video answer.

And that's just a normally like daily rhythm we do inside the AI Profit Boardroom. So your questions get answered with video tutorials. You also get the full agent operating system. You get weekly live coaching calls, a room of 3,700 founders, and the prompts, ESPs, and daily new tutorials as they ship.

Feel free to get it, link in the comments description, or just go to the AIProfitBoardroom.com. We've got a question here, pretty cool, about the best office setup ideas, right. So this is my setup as you can see. So I have a walking treadmill, got a nice view out the window.

I have a Mac Studio connected to a desktop monitor. We have a teleprompter at the top. We have Hue lights I can control with my phone. Nice little economic mouse right there.

Apple keyboard as well. Some guitars and awards and all that sort of thing just to make it cozy. I think if I had to choose for content, it's like I'm going to focus on the microphone first, then we're going to make sure we have good lighting, then we are going to add a second monitor. You just want to have a nice space that you enjoy operating from.

Also what I find these days is when you're operating clawed, a lot of the time you're just waiting. So what's the best thing to do during waiting? For me personally, I like to get the steps in, which is why I stand right here. Got a question about avatars.

So you can see David Ogden was asking about Gemini and avatars. So number one, they actually mentioned that they heard images are being flagged on Pinterest. Now for me personally, I'm not an actual expert on Pinterest, but what I would say is just warm up the account properly. That's what I do on every single platform.

And the reason for that is you just need to make it look natural and feel natural. So if you start a brand new account and then you're creating loads of content, that doesn't look natural. If you gradually warm it up, you interact with other people, you take it step by step each day, that's much better. And also bear in mind, even if you create a lot of content on one platform from day one, probably not going to help you a lot.

But if you go into Pinterest and then you gradually build it up, that's going to be much better to the point where you can just give it to in a virtual system. And then the next question they actually had was about creating an AI avatar, but they'd prefer it to be faceless, which is totally fine. So if you prefer it to be faceless, the good thing about that is like, it's actually kind of a benefit because if you don't show your face, then you don't need to sync the audio API with the video API. And so if you don't need to sync it, you get more options.

If you get more options, you can use something that sounds more realistic. So for example, if you're using your voice only, audio only, then you could go with something like Phish. Phish sounds a lot more human when it comes to faceless audio. Way better, right?

Whereas for example, a lot of these other platforms where you generate the face avatar, not that flexible in terms of the audio API you can use. So that unlocks a lot of options. And if you prefer to do that, totally fine. We've also had a member, Todd, who joined recently.

So welcome, Todd. Fantastic to see you here. Thanks for introducing yourself as well. So I recommend starting here for the start here post inside the AIprof forum.

That's the best place to get the most out of this. And that just gives you a path in terms of where to start and what to do next. So thanks so much for watching. If you have any more questions like this, feel free to post them inside the AIprof forum.

You can also direct message me personally inside the community. I answer your questions, give you help and support whenever you need it. Inside the calendar, you can drop a week of coaching calls. Inside the classroom, you can get access to all my best trainings.

Inside the map, you can meet people in your local area. And bear in mind, I do personal tutorials like this for you just to help you as much as I can because I really care about your results. So hope to see you inside there. Cheers for watching.

Bye-bye.