---
title: 'Claude AI SEO: Live Ranking Session + DnB Vibes'
source: 'https://youtube.com/watch?v=ngdNoImKW8s'
video_id: 'ngdNoImKW8s'
date: 2026-07-28
duration_sec: 3346
---

# Claude AI SEO: Live Ranking Session + DnB Vibes

> Source: [Claude AI SEO: Live Ranking Session + DnB Vibes](https://youtube.com/watch?v=ngdNoImKW8s)

## Summary

The video discusses the current state of AI agents, backlinks, and SEO, highlighting the importance of backlinks in the era of AI-generated content. The presenter showcases Chinese AI models like GLM 5.2, Kimi K2.7, and N2, and demonstrates how to use them within an agent operating system to build various applications.

### Key Points

- **Backlinks remain crucial** [09:04] — Despite AI making content creation easy, backlinks are more important than ever as a differentiator. Building backlinks cannot be fully automated yet.
- **AI agent for backlink outreach** [13:21] — An AI agent can handle backlink outreach by targeting small, realistic websites. The presenter's system sets up full outreach campaigns automatically.
- **Three Chinese AI models introduced** [41:13] — GLM 5.2, Kimi K2.7, and N2 are powerful Chinese AI models. N2 is temporarily free as an API, Kimi K2.7 will be open source, and GLM 5.2 is currently the best of the three.
- **Kimi K2.7 capabilities** [42:10] — Kimi K2.7 can code quickly, create videos, and build games. It has a 'fast mode' and can create content that rivals human quality.
- **Judge agent for quality control** [46:35] — Using a judge agent within the agent OS allows the system to critique its own work and iterate until completion, ensuring quality control.
- **Agent OS as key differentiator** [50:22] — An operating system for agents is essential to organize and maximize the potential of multiple AI models. Without it, users cannot easily coordinate or track outputs.
- **Fusion method outperforms benchmarks** [52:10] — Using a panel of agents with a fusion API from OpenRouter, a judge fuses answers to produce better results than single models like Claude, as shown on the Draco score.

### Conclusion

The presenter emphasizes that organizing AI agents within an operating system unlocks their full potential, and Chinese models like GLM 5.2 are now competitive with or superior to Claude for coding tasks.

## Transcript

[music] Heat. Heat.
Heat. Heat. N. [music]
>> Heat. Heat. [music]
baby. [music]
>> Oh, [music]
>> Welcome [music] here. Thanks for joining.
second. Actually, I'll show you what we've been doing recently.
>> Hey, hey, welcome here. [music]
[music] >> Heat. Heat.
>> Will Fable come back? I think it's coming back this week, mate. Just give coming back this week, mate. Just give it a bit of time.
>> right, let's test out this music [music] agent.
nicely. [music] With AI SEO, are back [music] links still relevant? Yeah, 100%. I mean, to
be fair, at this point, anyone [music] can create content, right? So, backlinks are more important than ever. So, let me give you an example that we So, let me give you an example that we go over [music] to HFS here.
boardroom right you see the the back links [music] that were built to this website over time. So that's increasing over time right now if you look at the over time right now if you look at the trajectory of this website it matches
built. So this is the same website that we're checking out here. Biggest reason for this is like it's a differentiator. Like if you don't have backlinks pointing to your content, especially if you're creating it with AI, very
difficult to rank because you have no authority. And so anyone can create content now. Anyone can target low competition keywords now. But the biggest difference between everything is how many backlinks you can get because
you can't really automate that at this point. And there'll probably come a time, [music] but honestly, like people have been talking about it for years and never really fully automated it. So, it's interesting. It's an interesting
time where, you know, most most people don't know how to to get back links even don't know how to to get back links even in the [music] world of AI.
[music] >> Hey.
done inside Hermes in the agent OS right [music] we've set up a system where you can basically handle back links and do [music] outreach with your AI agent. So we've
taught it a skill where [music] we give it the page that we want to build back links to and we say like reach out to five for backlinks and then it goes from there and uh then it can set up the full
outreach campaign [music] as you can see. been done. So [music] yeah, you can get been done. So [music] yeah, you can get it inside the agent OS system.
>> Make this more attention grabbing and more interesting. more interesting. [music]
>> Hello. Hello. [music]
Heat. Heat. [music]
>> [music] >> Heat. Heat.
>> So for those sort of things, you want to do more PR, [music] which you can't do more PR, [music] which you can't really automate that much. So I would recommend that you don't I mean for outreach like you don't want
[music] to target websites like those right you there's there's better ways to So you want to target [music] like smaller, more realistic websites who actually link to you. [music]
>> Happy mate. [music] Thanks for the questions.
Nice. That was a great event at Estonia. Really enjoyed it. >> It kind of feels like coding with Claude is is a new version of that.
Yeah, it kind of feels like Claude is a new version of that, you know.
[music] here.
[music] >> Heat. Heat.
>> Heat. Heat. [music]
autonomous Chinese AI agents that I'm using to build and automate basically anything. And one of them is actually completely free which is Nex NGI. That's free temporarily as a free API on a router. We also have for example Z AI
which is going to be open source and Kim K 2.7 which is also going to be free local and open source as well. So these are three different powerful Chinese models and if you're wondering okay like are they actually any good? Can you use
them etc. We've plugged all three into our agent operating system. So I'm going to basically build and automate anything. So for example this is GLM 5.2 some pretty amazing stuff as you can see
right here in terms of like open world games. Uh we created like for example full websites, all sorts of interesting things with this and it's actually really impressive. Now we also have for example
Kimmy code and Kimmy K2.7. And one of the best things about this is that it's actually got this new fast mode which is really powerful as well. So basically really powerful as well. So basically with this you can code stuff out really
quickly. But not just that you can actually for example even create videos like you can see with AI. This is not me. This this one here and the AI fully me. This this one here and the AI fully created it, edited it, scripted it and
it's is better than what most of my team could do. So this is with Kimmy K 2.7. create anything good or these are not good. You can clearly see like this is actually a really good video that was fully created with AI. And you can see
another one right here. This was a game that we created. All sorts of interesting stuff. So these are are pretty powerful and and probably the best versions of Chinese AI that we've ever seen. Bear in mind like they're,
you know, particularly Kimmy and GLM 5.2 like these are open source projects which is insane when you think about that. Now on top of that what we've also got is free clawed code and that is using
is free clawed code and that is using N2. So we have N2 which is the free API the free Chinese API from Nex AGI. This is a 250 well 262K context window. Uh
and it's free as an API I think for the next two weeks something like that. And it's pretty impressive like you can use it inside frequ
with it and then it just goes off and builds stuff and you can see some which is pretty insane when you think about it as well. So it can create all sorts of things.
one system, like all of this becomes way more powerful. Okay, so
let me show you some other stuff that we built here. game that we built out, which is pretty crazy. It can create
which is pretty crazy. It can create like whole worlds as you can see here. like whole worlds as you can see here. Can create like these 3D games as well, which is pretty amazing. And this is just really like a visual
example, but it's fun to play. And you couldn't do this with with these models powerful. Here's the other one that I showed you before. Here's another one kind of like with the snow effect, but open world. Again,
we built with the AI and this is with GLM. So, we called it GLM OS and basically it created multiple different apps and then if you type If you close it and then open it up again, it's got what we've written a
second ago. So, this is pretty amazing. So if you're looking for like, you know, So if you're looking for like, you know, the best free coding model, give N2 a easy to set up. And the other thing about all of this, this is pretty
interesting, too, is like if you're using Claude, well, normally inside Claude, for example, you can't plug that into Hermes or Open Claw. They actually stopped all that, right? But with these three
Hermes. So you can have a whole team of AI agents. And we've actually set up different profiles for each of these AI models. Right? So for example, GLM 5.2, we have a separate Hermes agent profile just for that API. So we can switch
between it. We can get it to do stuff. We can get that working with Kim K2.7. We could even set one up for Quen 3.7 as well. And then, you know, you've got and they're all working together. Now,
here, you can see that we can set up teams of agents working together. So, we set up this one for Kimmy and then it went off and built this for example. And if we open up that, it went off and just had a team of agents create this video.
that video I showed you before. We did the same, for example, with GLM as well. different agents building together. And one of the other interesting things one of the other interesting things about GLM in particular is that it's it
can look at its own work and critique it. So, what we did with Hermes agent and GLM 5.2 is we had multiple agents. So like we've got an editor agent, we have for example a writing agent, a research agent, but most importantly we
have a judge agent. And so the judge can look at the work, look at what's been created, and then actually give feedback and tell the team to iterate until it's completed and the job is actually done properly. And so the great thing about
that is if you're looking for a system that has quality control, well, you can because you can always create a judge agent, plug them into a camb board like you can see over here. And these are actual use cases and and just to recap
here instead of actual use cases like stuff you would actually build. We've shown you the video agent the the games that you can create. It can create websites. You could create tools for you. So for example like the
app we showed you with N2 that we built earlier. Like this is pretty powerful stuff. And also not just that but you can use the coding plants that you get with these agents and plug them into Hermes or OpenClaw or whatever you want.
get the API for that and plug it into free clawed code, which means you get combined with N2. And then you can even voice control it as well and build cool additionally, on top of that, what we actually have over here is the idea
factory. And so when we've got our agents talking together over here inside the group chat, they can drop ideas. So we can have, for example, like GLM 5.2 two or we could have N2 or we could have a group chat with all of them and Kimmy
K217 and then they can drop ideas into our pipeline here and we just go off and our pipeline here and we just go off and build them. And once that's done, it will create a plan. If we click shape it, then we can approve or reject it. It
And finally, once that's done, we actually have a full gallery of can see here, like it's it's pretty wild, you know? This is a full gallery say, "Well, can you actually build stuff with it?" Yes, you can see all of these
apps right here. Can you actually create awesome content with it? Yes, you saw the video before. So, this is not like just a demo. This is stuff you can actually use right now and is ready to go, which is mind-blowing when you think
having like good systems. And when and then you got Claude working together, you got Hermes, you got together, you got Hermes, you got anti-gravity, etc., you can create and
you want. And that's really really all like power of this as well. Really and you can see some screenshots of what were built pretty amazing like the the graphics on all this stuff as well. Pretty interesting.
complex to build these sort of systems around this stuff? So, we actually have 3,600 members inside the air profit boarding, which is where we give away our agent operating system. And you can see there's 182 pages of wins and people
building stuff like this. So, people get awesome results. So, I know like if I everyone inside the community can't code and they can build this, then I know who have never touched AI before have set up agent operating systems that help
the key really is like for example, if you were using any of these, like if you were using, for example, Kimmy directly at kimmy.com, you wouldn't really get to find what you've done previously? How are you going to find all the all the
have it coordinating with GLM 5.2? How are you going to create a pipeline of automatically and then view inside the gallery? You can't really do that inside way of doing things is to make sure that
you have a operating system. Without that, you can't really get the most out differentiator as well. And I think that's why some people love GLM for It's like some people have great systems and they've organized it very well and
then other people have not not really organized it and so they just have like you know Kimmy in a separate tab and you know they forget about it or they etc. Right? It's the same for example if we can just quickly go inside the manage
section. We can set up a new model with GLM 5.2 or we can set up a new agent GLM 5.2 or we can set up a new agent profile with Kim K2.7. Then everything and we can go back and see what we've built as well. So we actually have a
separate workspace for GLM 5.2. We have a separate one for Kim K2.5. We have a separate one for example for N2 as well where we can preview what we've built where we can preview what we've built which is super useful.
And that's the way that I recommend running this, just keeping it organized.
I've tested, my favorite is GLM 5.2. The benchmark, this is so new, the benchmarks haven't even come out for GLM 5.2. You can see how Kimmy and N2 score on SW Bench verified. Obviously, Claude
can't use it anymore. And that's the interesting timing about this. The other pretty interesting, too, is like, and this was proven on benchmarks, so you know, you make up your own mind, but proven to work to get better results.
You could have a panel of agents. So, you could have Kim K27, you could have you could have Kim K27, you could have GLM 512, you can have N2 as a panel that works together. And then when you ask or request it to build something, it's
going to use all those models using a fusion API which is from open router. And then a judge actually looks at all the answers, judges it and fuses it into one answer. And that is beating on benchmarks fables results which is
Draco score. So if you look at the Draco score here um this method of using fusion with open route API and you can select those models I was talking about you could actually potentially and test it yourself like don't believe me don't
yourself itself you can potentially get better results in Fable just doing that as well to power all this up and just have them working camera.
much smaller context window than GLM 5.2. And that's what I've noticed as well. It's just GLM 5.2 is is the best out of all three of these. So, if you're If you're looking for a free one, I would go with N2. If you've got an
amazing setup, then I think Kimmy K217 is going to be open source pretty soon combination, see what you like and and just go from there. That's what I'd just go from there. That's what I'd recommend actually.
the full system that I've talked about today, you saw how Kimmy builds stuff, you've seen how powerful GLM is, you've seen how to use N2 for free with free claw code. If you want my full agent operating zip file with Kimmy, GLM N2
use to build all this stuff, the Obsidian memory setup, coaching calls these agents, you can get that inside the AI profitable border, and you get this whole system. We actually update it daily. So if you go to the link in the
airoffer.com, you can get this whole ask questions, get help and support. I answer those personally, too. Inside the lessons. And if you go to the daily update section, you'll find the agent OS
30-day road map on how to implement it. And also, we've got the last update date here, so you know when it was last updated. We do la uh add new updates daily just to help you and give you a tool that just gets better and better
these new lessons as well that come out. So we've got a full tutorial on GM 5.2 and Kim K7 plus the fusion method that I was talking about before. Inside the so you can ask questions, get help and support in real time. And inside the
map, you can actually meet people in your local area who are using AI agents like you. So feel free to get that link in the comments description or go to the aiprofit.com. Thanks for watching. Cheers. Bye-bye.