Perplexity is jumping on the open claw band wagon, or at least their own flavor of this. They said that their premium subscribers are going to have access to a new agent tool that can take 19 different AI models all working together in computer use, meaning it can take control of your screen and get jobs done for you. Now, there's a whole bunch of interesting things and features that they have added that I think might give them a bit of a competitive advantage. I'm going to be breaking that down.
But before I do, if you want to get access to all of the different AI models I talk about on the show, including Perplexity, go check out my own startup, which is AIBox.ai. I've recently overhauled the entire site, and we've recreated it from the ground up. If you've tried it in the past, I'd love for you to try it again. For $8.99 a month, you get access to over 40 of the top AI models all in one place.
You can test them against each other. For the first time ever, we're adding some music capabilities to our platform from 11 Labs, if you'd like, to be able to also generate music with AI. So go check it out. It's only $8.99 a month.
The link is in the description to AIBox.ai. All right. Let's talk about what's going on with Perplexity. So what their highest tier gets access to, and Perplexity's highest tier, I believe, is $200 a month.
So you're not going to get this. If you're just a $20 a month user, this is not a cheap use case. But it's so perplexing Max Plan. Because it's also running entirely in the cloud, they say that it can avoid a lot of security risks that have happened to local device-level agents like OpenClaw.
And essentially, this is, in their own description, a tool that, quote, unifies every current AI capability into a single system. So basically, this is a cloud-based computer using Agent. It can execute complex workflows independently. It's orchestrating 19 different AI models.
And it can even spin up what they are saying are sub-agents to tackle specific sub-problems. I think there is no hands-on demo that they've provided to anyone yet. But they've shown some examples of the workflows on their website. And basically, they're showing the system going around.
It's gathering financial, legal, statistical data. It's conducting a bunch of multi-step analysis, presenting results as a finished website or visualization. I think the product was originally slated for a live demo during a background media briefing last week. But then they canceled the demo after they identified some last minute flaws in the software.
It's kind of brutal. I mean, it really just shows how breakneck we are right now with a lot of these software. And to be fair, we've seen companies that have not done that, like Google at Google IOLS here. I remember they were demoing their glasses, this kind of new computer vision glasses that was showing Google Maps live on your Google lens in front of you.
And while they were demoing it, it kind of crashed and stopped working. So sometimes the software is not ready, but they just want to get it out there, and show it off. In this case, perplexity did pull back, and it's waiting since I guess they don't have a huge event with thousands of people there that they got to, wow, they can always show it in the future. So this launch, though, that they're going to be rolling out, I think is the next step, basically, for perplexity's evolution, perplexity is a company that kind of started out doing what Google AI is doing today in Google Search.
And that's where they got a lot of their initial user base. And as Google took over, a lot of people were saying, oh, is perplexity dead? Are they gone? I don't think so.
I think they've created a ton of really cool products. And just the fact that on perplexity, you get access, kind of like AI box, if I'm being honest. You get access to all of the different AM models in one chat interface. I think it gives them a great competitive advantage to something like chat, JPT, or Cloud, or perplexity, or Gemini, et cetera.
Initially, they were kind of this search engine, like I mentioned. But they've pivoted and done a lot of additional things. So one of those is, comma, it's an entire web browser. They're kind of positioning themselves as a new layer between users and internet.
A lot of executives argued during a recent press briefing that they did that competitors, including Google, have since kind of reshaped their own products in ways that resemble perplexity's early design choices. They say that this is kind of like, look, this validates that we're on the right track. But of course, I think, to everyone, we can see that this does give them a lot of competitive pressure. Perplexity also said that they're adjusting their business model in response to kind of where the whole AI ecosystem is going.
One of the first AI startups that was experimenting with advertising. They actually abandoned ads. Late last year, they said that the model risk kind of undermining user trust and accuracy. But that's interesting because chatGP is now about to jump into this.
And they've reported that they have 900 million weekly chat JPT users that are all going to, I mean, a lot of the free ones are going to start getting seeing kind of these ads everywhere. So it's interesting that Perplexity already tried ads, walked away from it, chat GPT is about to get into that. I think rather than go and try to chase kind of maximum scale Perplexity has said that they are targeting a narrower audience. They said they're looking for professionals that are making what they describe as a GDP moving decisions, right?
So they're like, look, we're not going for the whole world. We're not trying to boil the ocean like opening eyes. We're just going for people making GDP moving decisions. And if you see a lot of their branding, you'll notice that they do a lot of work with highlighting news or financial analysis.
I think during a recent briefing they did, their executives really were kind of emphasizing the enterprise subscriptions and a lot of the deep research as some of their core priorities. They kind of publicly downplay all of the monthly active users. They say, you know, this is not something that we ever talk about. We're not actually, this is a quote from them.
They said quote, quote, quote, we're not actually on a mission to get as many users as possible. Now, if they had an insane amount of monthly active users, I'm sure they'd be touting it, but it doesn't seem like it's maybe the strongest thing about the company. And so it feels like that's why they might not be sharing it out so publicly. I think they're trying to kind of reinforce their position right now.
They said that they recently introduced a new benchmark for some complex research tasks. They're calling it Draco. They say that their deep research product outperforms things like Gemini. Perplexity also said that they have built their own AI optimized search index, which is basically reducing the reliance on third party APIs.
I think kind of the core thing to perplexity strategy right now is that they believe that the future is multimodal rather than betting on a single LLM like CHET-GPT or Claude perplexity is basically routing your question across a bunch of different third party LLMs. And they pick the best one for answering that specific task. They actually have a cool feature that I've recently seen, which is like a, I think it's called like court or something where you basically ask a question to the court. And it pitches the question to like all the top AI models.
They all give their response. And then it has them all deliberate between each other with the best response. And then it gives you like, based off of Groc, Gemini, Chachibiti, and Thropak, this is like the best response. Which I think is a pretty interesting, you'd hate to ask a question that one particular model doesn't have a good data set for it, right?
Like we know that anthropic is the best for code. And opening I might be the best for getting real world info. And Google's the best for things that need API integrations to a lot of data. Like they all have these kind of different strengths.
And sometimes I feel like I asked the wrong question to Chachibiti. I get a lot of creative writing or something like that. I'm like, oh, I really should be asking Claude. Anyways, perplexity is a tool for this, which I think is quite interesting.
So right now, according to all their executives, the visual outputs in the late 2025 were most frequently handled by Gemini Flash software engineering tasks were done by Claude Sonnet. And medical research was all given to GPT. So this is kind of interesting. These different areas that are funneling different directions.
Their system right now can automatically choose what model based on cost and performance. In some cases, perplexity is running some modified open source Chinese bill LLMs to answer different questions more cheaply. Some people have criticized that in the past because it weren't really clearly disclosed. But when they're usually running open source stuff, even if it's Chinese, it's not like they're sending data back to Chinese servers, typically unless they have an API.
If they're running it on their own cloud, then I think it's not really security risk. A lot of their executives argue that they're handling all this transparently. They say that this type of orchestration can significantly improve efficiency, which to be honest, I think that they might be onto something there. You can also, I think they have a whole bunch of other products that they're launching soon.
There's a comment browser that is slated to arrive on iOS next month, which should be interesting. And they're also planning on hosting their first developer conference in March, I believe, to promote third party use of their API, which I already have their API tied into AI box. It was a good way for us to get access to AI plus the internet. So it's something that's interesting for me.
Internally, they say that their focus has kind of shifted from daily query counts to revenue metrics. I think that has not been something that their users have not noticed. I think there's some people that have been complaining on Reddit and stuff about tighter rate limits on the free and paid tier. A lot of people over perplexity right now say that the free tier has not been degraded.
I think if you look at all this though, perplexity computer, this new feature that they're going to be rolling out here soon, is basically showing they have a really big ambition. They're trying to take on what Claude Code or Open Claw is doing. They're moving beyond just kind of being an AI search tool. And I think that they actually made the first seven distraction with perplexity comment, their browser, which is an AI agent browser that could actually do tasks for you.
So I think this isn't a new direction for them. I think this is just them kind of following through. And I will give perplexity credit, which is that it feels like they're faster than a lot of the other frontier companies. They beat OpenAI before OpenAI was able to release their Atlas browser and before Chrome was able to kind of do some of their features and even before anthropic was able to.
So if they're doing a great job, this is a company I'll definitely keep following in the future. Thanks so much for tuning into the podcast. I will catch you guys on the next episode. Make sure to go check out AIBox.ai if you want to get access to all of these models, including everything from perplexity, link is in the description.
It is $8.99 a month.