← Back to search
NEW Hermes Voice Agent is ABSOLUTELY INSANE!
The AI Firehose · 2026-11-09 · 8 min
Show full episode description
🚨 NEW Hermes Voice Agent is ABSOLUTELY INSANE! 🤯🔥 Imagine having your own JARVIS-style AI assistant that can SEE your screen, TALK to you, REMEMBER your projects, and actually GET THINGS DONE! 🚀 That's exactly what I built with Hermes Apollo , a powerful voice agent running inside my AgentOS! And it's WAY more than a normal AI voice assistant. Here's what makes it so powerful 👇 🎙️ Real-Time AI Voice – Talk naturally with Hermes and get spoken responses without waiting for the entire answer. 👀 Screen Awareness – Share your screen and ask Hermes questions about exactly what you're looking at. 🧠 Shared AI Memory – Connect Hermes Apollo to Obsidian so your other AI agents can access the same knowledge. 🔍 SEO Keyword Research – Ask Hermes to find keyword opportunities using your voice. 💻 Voice-Powered App Building – Describe what you want, and the agent can help build it. 🤖 Multiple AI Agents – Manage marketing, SEO, video creation, and other workflows from one system. 🏢 Interactive Agent Office – Watch your AI agents work and see what each one is doing. 🎯 Mission Control – Everything runs from ONE connected dashboard! And here's the GAME CHANGER… I can ask Hermes Apollo to find SEO keywords, then have my SEO agent use that information to create website content. Different agents. One shared memory. One connected workflow. 🤯 No more constantly repeating instructions or switching between disconnected tools! 🎥 Watch the full video to see Hermes Apollo in action and learn how to get the complete AgentOS setup. 🚀 Want the ready-to-install Hermes Apollo AgentOS? Check out AI Profit Boardroom: aiprofitboardroom.com 💬 Would you use an AI assistant that can see your screen and remember your projects? 👇 Let me know in the comments! #HermesAgent #HermesApollo #AgentOS #AIVoiceAgent #AIAgents #AIAutomation #NousResearch #ArtificialIntelligence
✨ Episode Outline — click any point to jump to it in the episode
Problem solved
How to run a screen-aware, streaming voice agent (
Hermes Apollo) inside an agent OS that shares memory with every other agent via
Obsidian.
Benefits
- Streaming voice: Hermes starts talking while still thinking, no walkie-talkie wait
- Sees your screen so you say 'look at this' instead of copy-pasting
- No API key needed; runs through an existing ChatGPT account with ChatGPT voice
- Shared Obsidian memory means every agent knows what the voice agent heard
- Wake words run locally; say 'stop' to interrupt; max task steps raised 90 to 500
Use cases
- Asked by voice to find SEO keywords: returned 'AI automation', 'AI workflow automation', 'AI automation agency', 'AI automation for small business'
- Opened Obsidian notes by voice and read back current projects; confirmed what was open on screen
- Built a to-do list app purely by talking to Hermes Apollo
- Video agent turns a typed news headline into a finished short video
- SEO agent pulls site numbers, finds keywords, writes content, and publishes from one screen
KPIs / results
- Hermes 'Herald' release v0.2.0.0 on August 3 added streaming voice; wake words shipped July 29
- Max steps per job raised from 90 to 500
- Julian spends 3-4 hours/day improving the agent OS
- 4 coaching calls per week inside AI Profit Boardroom
Tools / build
- Hermes Apollo voice agent
- Agent OS with mission control and pixel-office view
- Obsidian shared memory
- ChatGPT voice (via existing account)
- Video agent, SEO agent, marketer agent, daily updates agent
New Hermes Voice Agent is ABSURD. Good morning, sir. How can I help? That's my AI agent talking out loud. I asked it to open Google and it did. I asked it to find SEO keywords and it came back with AI automation, AI workflow automation, AI automation agency and AI automation for small business. Then I asked it to open my Obsidian Notes and it read off the projects I'm working on right now. Then I asked, can you see my screen? It said, yes, sir, I can see it. Then it told me exactly what was open. This is Hermes Apollo. It's a voice agent I built inside my agent OS. It sees my screen, it answers right away and it remembers what I'm working on. By the end of this video, I'll show you how to get this exact setup on your own computer. So what is Hermes? Hermes Agent is an open source AI agent made by a lab called Noose Research. Open source means anyone can use it and build on it. It has memory, it can learn skills and now it has a real voice. On August 3rd, Noose Research put out the Herald release. That's version 0.2, 0.0. It added streaming voice, which means the agent starts talking while it's still thinking. Before that, voice mode was like a walkie talkie. You talked, you waited for the whole answer. Then you listened to one long audio file. Now it talks back the way a person does. The update also added wake words. You say a word like, hey Siri, and the agent wakes up. The listening happens on your own computer. You can even give different agents different wake words. And when you say stop, it stops. They also let Hermes take more steps on one job from 90 up to 500. That means it can keep going on big tasks for a lot longer. Three things make it different. The first is cool. The second is handy. The third is why I use it every day. First, it can see. You share your screen and it understands what's on it. In the demo, it told me I had the Hermes page open with a chat about SEO keywords. When I said, give me a minute, it said, I'll wait while you take a moment. No more copying and pasting. You just say, look at this. For example, open the AI profit boardroom sales page and say, tell me the three spots where a new visitor might get confused. Quick pause before I show you the next part. If you want this exact setup, you can get it inside the AI profit boardroom. You don't just get a guide. You get the actual agent OS zip file with Hermes Apollo built in ready to install. We've also made a 30 day agent OS roadmap. Day one, you install the zip. Day two, you connect Hermes to your obsidian memory. So every agent knows your business. Day three, you turn on Hermes Apollo and give it his first job by voice like finding keywords for your website. There are four coaching calls every week where you can ask about your own setup live and plenty of members already run this exact stack link in the description or go to AI profit boardroom.com. Second, it doesn't need an API key. An API key is like a special pass. Most AI tools make you set up first. Hermes Apollo skips that. It runs through the chat GPT account I already use with one of the newest chat GPT voice updates. I didn't have to set anything extra up and you can change the voice whenever you want. Third, and this is the big one. It remembers and it shares what it remembers. Everything I say to Hermes Apollo goes into my memory system in Obsidian. Obsidian is a notes app. Think of it as one big notebook. They're all my agents read. So if I tell the voice agent something, my other agents know it too. It works the other way as well. My daily updates agent writes into that notebook every morning and Hermes Apollo can read it. So I can say, what did the daily updates agent find today? And it tells me, you might think I already talked to chat GPT with my voice. True. But chat GPT doesn't know what your other agents did today. And it can't see your screen and your notes together. Hermes Apollo can because it lives in the same system as everything else. Here's what it looks like day to day. You open Hermes Apollo and see every pass voice chat. Click one to pick up where you're going. Left off. Scroll down and you see what you've built with it. Like the to-do list app we made just by talking. You can shrink it into a pop-up that works on any tab. So while the text agent does a long job, you talk to the voice agent. Two jobs at once. Now, to be honest, it isn't perfect. In the demo, it started reading out a long list and I had to say, stop. Voice is great for quick jobs. For long reports, you still want them on a screen. That's why agent OS gives you both. Now for the part most people don't expect, the voice agent is one piece. Behind it is a whole team of agents working all the time. One part of agent OS looks like a little pixel office, kind of like an old video game. You watch your agents walk around and they're real agents doing real work. Click the marketer and you see what it's working on. Click Julian's agent and you see its full history. You can give any agent a job or add new ones whenever you want. There's an engineering office for building and testing. A main office where Julian's agent works. And a web designer agent chilling on the sofa until it gets a job. It looks silly, but it fixes a real problem. When you have lots of agents, you lose track of who's doing what. Here, you just look. For example, tell the marketer, write three welcome posts for new AI profit boardroom members. Then there's the video agent. Julian had an idea. What if he typed in a news headline and an agent made the whole video? So he built it. You give it the idea and it puts the video together. You could say, take today's top AI headline and make a short video on how boardroom members could use it. SEO agent. It pulls in the latest numbers from your websites. You pick a site from a list and see how it's doing. From the same screen, you can find keywords, write content and put it live. It's all hooked into the same memory. The keywords Hermes Apollo found at the start. The SEO agent can see them. So you could say out loud, find keywords people search before joining an AI community and then turn them into a boardroom page with the SEO agent. Everything sits under one main screen called mission control. Now you're probably thinking this looks great, but I could never build it. Honestly, you could, but Julian spends three to four hours a day improving the system. So you don't have to build it. You can just get it set up inside the AI profit boardroom, go to the classroom. Then the new daily update section. The full agent OS zip file is there freshly updated, download it and install it. What if I get stuck? There are video tutorials for every step of the setup made from questions. Members have already asked. You can also post a question and the team will help you. So where is this going? Wake words came out on July 29th. Streaming voice came out on August 3rd. That same update let Hermes answer voice notes on apps like WhatsApp and it let agents talk to other agents, put those together and you can see what's next. You'll send a voice note from your phone while you're out. Your voice agent will hear it, hand the job to the right agent and tell you when it's done. So what should you do now? If you're new to agents, get one agent doing one job, then add voice. If you already use AI, the big step is memory. Get your agents sharing one notebook, if you run a team list, the daily jobs a voice agent could do from one sentence like checking reports, finding keywords or writing first drafts. A lot of people feel behind with AI right now and that's normal, but the people doing well aren't chasing every new tool. They picked one system and kept adding to it. That's why voice fits agent OS so well. It didn't replace anything. It made everything else easier to use right now. Most business owners still haven't figured this out. They're still typing every question. And starting from zero every time that window won't stay open forever inside the AI profit boardroom. You can download the full agent OS zip with Hermes Apollo built in ready to install today. You get daily tutorials on Hermes voice, the memory system and every agent in agent OS. You get four coaching calls every week where you can bring your setup and get it fixed live. A prompt library of voice prompts. You can say straight to Hermes Apollo and a member map to find other agent OS users near you. Link in the description. Or go to AI profit boardroom.com. And if you want the full process, the SOPs and over 100 AI use cases like this one, join the AI success lab. It's our free community links in the comments and description. You'll get all the video notes from this episode plus access to over members.