Loading summary
Transcript1 lines
- [00:00]
A
Suno is rolling out watermarking and they're also going to be curbing how many downloads you can make because they're getting in a ton of lawsuits and they're trying to fight what they're calling AI music spam, which I think there's some pros and cons to this. I actually make AI generated music and I make about $2,500 a month from it. So I want to tell you a little bit about what's going on with this case. Marendale is signing a $100 million Google Cloud deal to train self improving AI and and Google Maps is turning Ask Maps, that kind of button that they have in there, into an agent that orders food, it can book hotels is going to be able to do way more than just asking Google Maps something. With Gemini, AI moderation is not always perfect and recently it has hit Reddit, discord and Tumblr on false positives. Also, a company called Naive has just raised $28.5 million to let AI agents say, spin up and run entire companies. I wanted to do a startup spotlight and they've raised almost $30 million. This is an interesting one. On the podcast I talk a lot about different models that you may not have tried, like Mistral from France, Cohere from Canada, Deepseek from China and Grok and Perplexity and even Meta's AI which just came out with a bunch of new coding features and a new coding AI. Now my question to you is, have you tried all of these different AI models? Because if you want to, it typically is going to cost you probably $20 to get a subscription to all of those different platforms. And I will be honest, some of those models are better at answering specific types of questions than others. For a lot of what I do, I use Anthropic and I use OpenAI and I use Google as some of my main powerhouses, but I use some of the other models for different tasks. Deep Seq, honestly, for example, is something that can get things done very fast and very cheaply. If you run out of your token credits, I would love for you to check out all of those different models and try them all and compare them all side by side for your next project or your next question just to see what the results are. If you want to try that, go check out AI Box AI, my own startup where you can test over 80 different AI models and you can also do a bunch of different image, audio and video models as well. You can compare all of the results side by side and pick what your favorite one is if you're working on a new ad or you need to create a bunch of new Facebook ads, for example, you can get different video models to create the same prompt. And I've tried this and been pretty impressed with Diffuser. Different things like Sora versus Google VO3 and you can actually run those different ads side by side against each other in Facebook to see which one converts the best. But you can also see them side by side. So if you want to go check it out, it is $8.99 to get started on AI box. It's super cheap and you get access to over 80 different AI models. I'm going to leave a link in the description to go try it out. Suno is now rolling out the new watermarking and they're doing audio fingerprints for their AI music generator. This is something that I use all the time. I love Suno. I've made tons of music there and I actually make twenty hundred dollars a month from my music catalog. A ton of it is either AI generated or it's just AI empowered. So I do a lot of meditation music is like my niche. I started it in college and so I'll still drop albums on there every once in a while on the weekends. And in order to do that I actually just use Suno to generate the meditation music in the background and it's my voice, I'm talking over top of it. But yeah, from that I still make 2,500 to $3,000 a month super, super passively. And it's kind of a fun side project. But something that is happening is that a lot of different lawsuits have been leveled against Suno. And to be fair, I think they did scrape out, scrape up a ton of different people's music that wasn't licensed properly. Now they're getting sued. I think they made a settlement with War Music Group. And the other part of this whole settlement is the, the goal to try to stop low quality AI songs from flooding Spotify and Apple Music with without disclosure. So without people knowing that they're AI generated, Deezer apparently has a tool that can detect AI generated music and it's removing any AI generated music from its platform that doesn't get a certain threshold of streams. So if it's a popular song they leave it. But if it's a AI generated song that has no streams, a bunch of people are just posting it. They're actually cleaning it off the platform. They just don't want it on there. So in this deal and what suno's rolling out right now. Downloads are soon going to be restricted to paying customers. There's going to be a monthly cap on how many you can download. So I think people downloading like thousands in a month and that was causing, you know, issues. When you try to upload thousands a month to Spotify, it was, you know, very spammy. So there's a monthly cap and if I'm being honest to me, I don't think this is gonna, this doesn't really bother me. I don't make an insane volume of music and I think it, it tends to be better when you make an AI generated song, which I'm not against by the way. If you make AI generated music, I think spending a lot of time doing multiple regenerations, mixing your different bits of the song together, doing covers of something. So like a lot of times I've done stuff where I'll play something, I'll upload the file of me playing. Maybe I'm not incredibly good pianist, but I might play a little melody or I had my son, he made an entire guitar album where he plucked some little tunes on a ukulele and had it turn into like this relaxing guitar album that he hits a bunch of streams from. So I think that's like a great way to go and. But you don't need to do crazy volumes for stuff like that. So when they're putting this kind of monthly download cap under it, I'm not super worried about that. That is all in the terms of this lawsuit that Suno has settled with Warner. This all kind of. They made this deal last year and it's starting to roll out now. Udio has another settlement with Warner Music Group and they are ending downloads entirely. Which basically makes Suno's approach, keeping downloads but gated behind a subscription a much better outcome for its business. With Yu Dio, you can't even download your music anymore. So Suno, you can continue to use it for generating music. I'm happy that's the platform I've invested in the most. And then they're both doing watermarking which basically lets distributors like Spotify or Apple track which ones are AI generated when they're actually uploaded. Right now there's a, there's like a little personal opt in. So when I'm uploading stuff it's like, is this partially AI generated? Completely AI generated or no AI generation at all? And you have like a disclosure on there and it's kind of voluntary but you know, there's nothing that enforces it. No one's actually able to tell and so this, they're actually going to be able to do that. I think streaming platforms right now have started to purge AI tracks that are just trying to get farmed royalties. There's a lot of spam and stuff on there. But to be fair I think you could do spam in you know, even, even before all of this came out the, the old tactic that the spammers would do is they would just go find like random obscure YouTube channels that would publish good music. Right? Maybe you're like not very famous on YouTube. They would rip your stuff and they would upload it like, like their own. It was just on YouTube, it wasn't already. They try to find something that wasn't already on like music platforms and then they would, you know, you could do there's all sorts of spam where they would like make it the same name as an artist that actually exists or they would put like keyword stuffing in the title. Like there's a whole bunch of, of stuff that those people would do and try to release thousands of songs that way. So I don't think like the spam element of it and AI gets a bad rep for it, but the spammers are going to spam and they've been spamming for a long time. Maybe this made it a little bit easier. But at the end of the day I think we, we also need separate like spam control tools and I think the platforms have been working on that and are working on that. So hopefully that will continue. Marendale is a startup that was founded by some alumni from Anthropic. They just signed a 100 million. Actually think it's even more than that but about $100 million deal with Google Cloud to train self AI systems. This deal they put together is about half the startup's entire seed funding. So I think they were able to get 200 million, they gave 100 million of that straight, straight to Google. And it is a huge bet on compute hardware to build AI that can iterate improve itself and automate research. Specifically I think they're going to be working on medicine and biology. Their software has a bunch of different stages of AI self improvement and it's going to be, it's going to be basically routing those to whatever chip works best. So Google's TPUs or Nvidia's GPUs, it's going to let the startup kind of optimize the costs and also potentially resell the orchestration layer to customers. It's something they're going to be able to make money from. The co founders are Benham Neshaber and Harsh Meta. Both of them came from Anthropic, but they're not the only ones working on this. Their two main competitors are going to be recursive superintelligence and recursive intelligence. Both of them are basically doing the identical thing, this kind of self improvement approach. They closed their startup round in late June at a 1 billion doll valuation. So this hundred million dollar compute deal is very aggressive spending for a company that's, you know, brand new. They just closed a seed round, give half the money away. Their bet I think is showing that recursive self improvement is they're trying to move this from just being this kind of research idea to an actual engineering problem. And they think that the real moat is controlling the software, that the orchestration and orchestrating all of this between expensive hardware and it's probably not compute. That's what they're saying. So I'm curious to see where this goes. Google Maps Ask Maps Assistant can now order food, book hotels and buy event tickets directly in the app. I love this. I think that Google obviously has some incredible assets and adding AI in a really thoughtful way makes them much more useful. I've been excited by how they've added AI into a lot of their different platforms and tools. I think they can do better in others. But with Maps in particular, it's, it's kind of cool because I mean you have so many different businesses on Google Maps, you know, you know like having like hotels and restaurants and everything. Google Maps really could turn into like an Airbnb for example, where you're like, you're searching how to get to a certain city and then all of a sudden it's like, well here are all of the hotels in that city and you can go book it there directly. They're already kind of competing with Expedia or you can compete with Expedia in a way. Like I, I've searched and found, you know, depending on if I'm just in a certain geographic location, I'm like, what's the closest hotels around me? I've seen them on Google before. Or sometimes I'll look them up on Expedia and then go look up Google for different reviews or different things. So I think they have an incredible asset there. And now you're actually going to be able to use their assistant, their Ask Maps to go and book all of this stuff. And it's not just hotels, it's also events, it's also tickets and food and all this other stuff, really, really cool. You won't have to leave the actual Google Maps website. It also is able to read your Gmail and your calendar to personalize suggestions, and it can recommend restaurants near your hotel, or it can tell you when your flight landing time is. So when you're actually going to book a hotel, for example, it can put in the correct time or give you a good check in time. Food orders are all going to run through Square Toast and Uber Eats. So it's not like, you know, I heard a lot of people, like, when they first heard the students were like, oh, my gosh, is it going to go kill Uber Eats? I think they do a lot of partnerships now. Those partnerships, of course, are subject to them just deciding one day they don't want to do that particular partnership. But I think we can be pretty confident. For antitrust reasons, that probably won't happen. I remember when Google Flights came out, a lot of people were concerned. And to this day, if you go search Google Flights, you're still going to get recommendations from a ton of different travel sites below. So it didn't completely kill that entire industry. Personal intelligence, which is the Gmail and Calendar feature, is going to ship by default. It's going to avoid a lot of the friction that apps have about sharing your personal data or, you know, going and syncing an MCP or syncing some sort of API or connection to Google or Cal or Gmail or your calendar. It's all built straight in. Ask Maps now remembers all of your past conversations. So if you ask, you know, what activities did you suggest for my Seattle trip? It's going to remember whatever you're talking to about in the past on Gemini, and it's going to pull that up. Maps right now, I think, is really shifting from kind of just a navigation tool that we've all used for the last number of years into a transaction engine. They're trying to actually get you to make transactions on there. And of course, they can take a little piece of those pie. They can charge for ads or they can charge transaction fees. So this is something that makes a lot of sense for them right now. Social media platforms are aggressively using AI to catch spam and other things, and they're also getting real results caught up in this. So some of this wrongful banning, you could say, is getting thousands of innocent users. There's also, like, no way to appeal this before your account gets completely deleted. Discord alone locked out 8,400 accounts over the summer because its AI accidentally mistook chessboards for illegal content, which is a problem. And Reddit's AI tools removed like over 10 year old posts from our Ask Historians. The Reddit thread, our Ask Historians because it said that they were linked to a historical photo site that the system had flagged as spam. So it's like, look, you're linking out to this. Or I don't even think it was all the posts, but sometimes like the comments on it, they're all linking out to a bunch of historical photo sites. They review that as spam and then they go and delete all of the threads. Some of these are 10 years old, right? And they're like, oh no, these are AI generated threads. And of course it's old threads. So it's, it is not the case. Discord's AI was designed to flag content for human review. There was this kind of bug in there, I guess, and it, it let it bypass humans entirely and it just auto banned a ton of users. Some of them were locked out for weeks. Which, if it's a tool that you use a lot, this is obviously not very fun and it's actually interesting. Discord is a platform that my company, AI box, we use kind of like instead of Slack, we use Discord for all of our developers and we have like a community over there and we do a lot of different stuff on Discord. I can only imagine. And like we gotta pay premium, right, to run our company through there, so like we pay for it. But I can only imagine getting banned and locked out for weeks. I mean, I really just switched to another platform. So I think this is quite detrimental for some of those platforms. Reddit reports that their systems now block about 2 million fake votes daily. And their enforcement is ramping up. They say it's on hate content. It's been ramped up over 200%. They use LLMs to spot any spam patterns that they can find. And of course I'm all for blocking spam, so if they're able to get real spam and real AI generated junk cut out, I'm thrilled. But I hate the false positives and I think a lot of people are, you know, leaves bad taste in their mouth if that happens. Naive has just raised $28.5 million to give AI agents the power to basically set up and run an entire company. So they're going to have a startup package. It'll do business formation, it'll do payments, it'll do accounting, it'll do cloud infrastructure. All of that is in a single API that their agent can call so developer can basically give an AI prompt and then it can have this, their tool spin up like an llc, it can make a bank account, it can make emails, it can make phone numbers. I think this is really cool and personally I love this because it takes out a lot of the pain and start in making these startups. And they've been very successful and popular. They have about 30,000 developers that signed up as customers in the first few months after they launched. Their annual revenue 10x over 6 months to the low double digit millions with just 10 full time employees, which is absolutely phenomenal. They have a bunch of customers that are AI agencies, faceless content creators, and they have at least one autonomous rental car business. An example is a TikTok account, uses them to generate and post AI videos of dancing cats and dogs. The company is building infrastructure to cut the cost of running AI agents at scale. So they have sandboxes, they have a model router that picks the cheapest AI model for whatever task you're using. They have a memory layer, they have a serverless runtime. So customers only pay when agents are actually working. There's a lot of really cool features in there and so yeah, no surprise that they're able to raise almost $30 million. I think this is an interesting company. I definitely am going to keep my eye on naive guys. Thank you so much for tuning into the podcast. If you want to try all of the different AI models, test them side by side, compare their results for whatever your latest project is, and also get Claude and any other AI model to integrate with all of those. So like you get image, audio and video generation inside of Claude with our connector, go try AIBox AI. My own startup is $8 and 99 cents to get started. We also have annual plans that are 20% off of that and you can scale up your usage. I'll catch you guys all in the next episode.