Hermes Agent V0.20 Just Changed AI Agents Forever! artwork

Hermes Agent V0.20 Just Changed AI Agents Forever!

AI News Today | Julian Goldie Podcast

August 5, 2026

Hermes Agent V0.20 Just Changed AI Agents Forever!
Speakers: Julian Goldie
**Julian Goldie** (0:00)
Hermes Agent V Zoo, the Herald release, it just dropped today. I'm going to walk you through exactly what it means, how it works, how to get it, and how to use it. So the easiest way to update is you can go to the Manage section inside the Agent OS, and then click on Update, and then you can get the latest version, so we can update to the latest version right there. If you want to, what the highlights are, how it works, et cetera. So the first thing that you can do is you can now talk to Hermes, which means that the voice speaks whilst the response generates. It basically responds instantly whilst you're talking. It's got sighted sources, agent to agent. It's also a lot faster. So first token in 0.9 seconds instead of 4.3 seconds, and also a lot faster on Hermes desktop. Also a Hermes desktop platform, so you can now get artifacts with Sandbox Live Preview, Smart Approvals, CLI Power, et cetera. Now, we're going to run through the change log, and I'll show you this in more depth, plus some examples of how it works, et cetera. But if you're wondering, okay, what is the Herald release? Well, here's a few headlines.
Basically, it's just shipped a big update, and your AI can now finally hold a conversation. You can talk to it, answer it as it thinks, and you can cut it off mid-sentence just by speaking, kind of like a phone call. And also you can say your own wake work from across the room, and your computer starts listening. So for example, if we go over to the Agent OS over here, and then we go to the chat section, you can trigger this on and say, hey, Hermes, how are you today? What time is it? And you can see that actually goes inside the chat, so it can start responding to us directly from this. So it actually listens to us with the wake word, and it's ready to go, and then it can respond in real time. So for example, we said, hey, Hermes, what time is it? And then it responded the other day. This update actually dropped the other day, but basically this new update is a bundle of all of these updates together. It also backs research with sources, so you can actually check, ping your other systems the moment the work finishes, and it even talks to other agents as well. So what is the talk to Hermes update? Basically, Hermes answers whilst it thinks, and you can interrupt it just by talking. So it can stop, listen, and adjust as you go along. So it's kind of like having a phone call with your AI agent. So the easiest way that I find to use this is, I've got Hermes Apollo set up here. We can actually speak to Hermes' agent in real time, and then it can respond to us.
So you see how we can speak to it directly here, and then it will listen, and then if we ask it for a response, it will talk back. This actually has a voice as well, but I won't play the voice on the playback, so it just won't sound that good. But you can see how it's going back and forth with us in real time right here, and we can interrupt it whenever we need to as well. With the wake word, basically, this is just hands-free control. So you can have Hermes sitting in the background. You could have, for example, Hermes desktop on a different monitor or the Hermes HNOS. And then what this means essentially is that it would listen and when you say, for example, hey Hermes, it will wake up and start responding. There's also a voice on every platform now. So you can send a voice note to Hermes on WhatsApp, on Facebook, Ding Talk, Line, QQ, etc.
And it will automatically reply with TTS. So it's got auto TTS replies and it's platform aware.
Additionally, there's a new grounded citation skill.
So the grounded citation skill allows Hermes to produce research where it has sources for everything that's used in. So this is kind of like deep research where you just avoid hallucinations, you get better research back because it's citing the sources on every response. Kind of like perplexity does as well. Now inside Hermes desktop as well, you've got the artifacts feature. So here you can view everything you've created and used with Hermes agent across all your platforms. So for example, we were using Higgs Field the other day, and with the MCP with Hermes and Higgs Field linked, it creates some nice images. Now, all of those artifacts are actually displayed, so we can view them over here as you can see, which is pretty cool, and it's sandboxed live.

6 more minutes of transcript below

12,000+ transcripts fetched by developers building summarizers, RAG pipelines, and podcast tools

Feed this to your agent

Try it now — copy, paste, done:

curl -H "x-api-key: pt_demo" \
  https://spoken.md/transcripts/1000651996090

Works with Claude, ChatGPT, Cursor, and any agent that makes HTTP calls.

From $0.10 per transcript. No subscription. Credits never expire.

Using your own key:

curl -H "x-api-key: YOUR_KEY" \
  https://spoken.md/transcripts/YOUR_EPISODE_ID