This Week in Startups
00:12:51 · Lon Harris
[00:12:51] S1: At least two, if not three. I mean, even just, like, if you're thinking about the content alone, the amount of writing, the amount of research, like, not even getting into the editing and post production and everything. Yeah, it would need to be a two or three person team larger.
00:13:05 · Jason Calacanis
[00:13:05] S0: I'm to say three. So now we're at 12, 13, 14 people in the company. And there are things we're doing now that we just couldn't have done. There's software we're making that is growing the business. You put all that together, I think we're getting about $2,000,000 in value and $40,000 in spend. So now you kind of do some math there. You're talking about 50 times the value. So for every dollar we spend, we're getting $50 out. So I want to just let that sink in. A dollar invested in AI today, if you're doing it properly, is $50 of output. Now, can you monetize that? Is it profitable? Or are you just spinning your wheels? Well, that's up to you to define the goal, right, Lon? And what we've done now is we've defined the goal. We're hiring for 10 positions here at the company. You say, wait a second, why aren't you just automating them? Well, things are growing so fast. There are things that are uniquely human, like being on air. Video editing, still uniquely human. Having taste, running a community, running events. That's in the real world. You still need somebody to go to the venue and greet people and all that stuff. So what we're hiring for are uniquely human people. But I have to say this technology since the bots came out and in the last ninety days... I believe the last ninety days are the equivalent of the first two years of using this software. That much has changed because we were trying to do this stuff with OpenClaw. Then I tried to do a Perplexity computer, and I tried to do a cloud co work and projects. We tried to do it originally with ChatGPT. I mean, we've been down this road, you know, Slackbot, everything.
00:14:41 · Lon Harris, Jason Calacanis
[00:14:41] S1: It's the... I mean, people say it's the agents, and I think that that could... Sometimes gets considered as agents are inherently more powerful or whatever. It's that the agents have context and access. That's that's what... OpenClaw, the promise was you're gonna have this agent that act independently and is smart and can do things. And it's true. I don't mean to say it's not true, but it was very difficult, especially if it was in a virtual server somewhere, for it to get into Slack, get into Notion, get into email, get into your calendar, get into your message. [00:15:10] S0: Did that happen? Well, one, all the services out there built connectivity. Right? The second thing that happened in all this is that they put a virtual computer inside of Muse and inside of Grok Bot. I think Claude might have it. Claude hasn't released their agent technology [00:15:28] S1: that you can hook it into your Google Chrome, and then it can look at everything in your browser. [00:15:32] S0: It's No. But they're adding the virtual [00:15:34] S1: Right. [00:15:34] S0: They're gonna machine that the agent can use. Once you do that, and I've gotten comfortable with it inside of Grok, I don't know if you've seen this, there's a vault. Have you started using the vault yet? [00:15:43] S1: Absolutely, 100%.
00:15:43 · Lon Harris, Jason Calacanis
[00:15:43] S1: Absolutely, 100%. [00:15:45] S0: Okay, so there's a vault. So booking flights is also another one of these things that is a back and forth, back and forth. You're giving it to a human, etcetera. It's just hard scheduling another one of these. So what I'm doing now is I'm having my Athena assistants use Grok Bot. I got them... I have two Athena assistants. They're going use the same instance. And then I'm having them do every task they do. They have to build a Grok Bot for it. So now I sucked in all the Notion pages of my travel preferences, my miles, numbers, whatever. And I just started giving it goals. Now they're using that too. Boom. This stuff is all just getting done and it actually finishes the task. And that was the problem previously. It didn't finish the task. People didn't know how to give it goals. It wasn't connected. There wasn't a computer. [00:16:35] S1: Would time out and get confused. And then you'd like, What happened to that? And be like, Oh, you're right. I didn't run that. I'm sorry.
00:16:42 · Jason Calacanis
[00:16:42] S0: It was brittle to use a Chamath term that he likes to use. There was a brittleness to it. And now it's not brittle. It's the opposite. And so my advice for everybody with the Grok Bot is you're going to need to do the assessment like I did. If Grok Bot was a thousand dollars a month and I think we've got four or five people on it and don't clip this and send it to Elon, but if it was a thousand dollars a month of the four people I have using it, I probably would have three of the four keep using it for a thousand dollars a month. Because even then, my $40,000 a year bill went to $200,000 it would be 10 to one. So every dollar I spend, I get 10. So I think what's going to happen is this is kind of like Uber and Lyft when they were discounting rides to get people to use the software. They're going to start boiling the frog. They're to start raising the prices and take the losses out. Elon's just the type of person who will tell you that's what he's doing. Claude, and I haven't heard Dario just come out and be like, by the way, we're losing money on this. It's not sustainable. It's growing too fast. We need to change the price a little bit. We got to change the deal. Elon's just upfront about it. Anybody who's complaining, look at what you're getting out of it. Please go do some simple math. Are you getting $50.30, $20 for every dollar you spend? Just increase your spend. Just go to the ultra plan, go to the $100 plan. And you really want this to... I do think some people felt rug pulled because their browsers, their bots stopped working, they might've been dependent on that. This is why you need to give everybody the status bar that shows usage. I agree. Agree.
00:18:21 · Lon Harris, Jason Calacanis
[00:18:21] S1: Countdown clock. I figured it out because I got so frustrated. We were doing... I had Grok Bot doing, like, a long form research project, and so I was running out of compute even before all of these changes just because it was combing YouTube and doing so much research for me. So I discovered you go to cursor.com/dashboard/spending. That's where you could actually keep a running eye on how much usage your Grok Bot has left before restarting the next week. I usually mostly use it up by the end of the week, but I'm gonna keep an eye on it now. But anyway, that's where to go. But I agree. Having it in Grok Bot itself so you could just monitor it more closely would be very helpful. I think people don't realize that if you do give it those long range assignments, it can suck up compute pretty quickly. [00:19:05] S0: I am on a, ultra plan, and it, yeah. I I don't know how I got my Ultra plan. It might be because I'm part of the all in organization and I have that icon next to my name. And so somebody's paying for my Ultra. I don't know what Ultra cost. I think that might be 2 or 3 or four. [00:19:26] S1: I think that's the 200. [00:19:27] S0: 200? Okay. I have Ultra. I'm at 51% weekly usage, and it resets on November 3. So they gotta just let everybody know what they're spending and don't make you search for it. This. [00:19:40] S1: Yeah. Cursor.com/dashboard/spending, folks. That's where to go.