👨‍🚀 Who is benefiting from the AI business boom?

Hi AInauts,

Welcome back to your favorite newsletter!

AI makes you rich, AI sets you free, AI makes you invisible. And it is precisely these three aspects that we will be examining today. From an $80 million exit to open models with cool new features to privacy layers, today we'll be covering everything.

Here's what we have in store for you:

• 🤑 The AI Gold Rush - How Solo Founders and Companies Benefit

• 🐙 AI without data octopus: Real innovations from Europe

• 🥷🏻 How to use ChatGPT and similar services anonymously

Let's go!

🤑 The AI Gold Rush - How Solo Founders and Companies Benefit

2025 still feels a bit like the Wild West. Except this time, it's not prospectors panning for gold—today, it’s founders and creators seeing what AI can actually make pay off.

Between the myth of the solo unicorn and real deals, one thing is clear: AI is no longer just a tool, but the playing field itself. And we thought we'd share a few examples with you, without the hype and false promises, if possible.

While others are still debating, the smart ones are already experimenting. They are building solo businesses that would have required entire teams in the past. And in doing so, they are getting 10x or even 100x more done than before. When such power is used in a targeted manner, great things can happen, as the following examples show. Let's dive in!

The fable of the solopreneur who starts a billion-dollar company

In the startup world, there is a lot of talk about how AI makes individuals so productive that a generation of "solo unicorns" could emerge—one-person companies worth more than $1 billion.

That might be a ways off. But it is clear that AI will pave the way—and one person who consistently provides exciting insights on this topic is Greg Isenberg with his podcast guests. Looking to stay on top of things and get some real, practical insights? Check this out.

While a real solo unicorn remains a mythical creature for now, Israeli developer Maor Shlomo has at least offered real proof that things are moving in this direction: the solo founder sold his 6-month-old startup Base44 to Wix for $80 million in cash (interview here)!

Of course, it doesn't always have to be an exit – the 18-year-old founder of the calorie tracker Cal AI makes $1.4 million in profit every month (now with 30 employees). Not too shabby!

… And if Elon gets his way, we’ll be able to simulate entire software companies using AI—in reference to Microsoft, he calls the project "Microhard".

Audos aims to launch 100,000 companies with its system—per year!

Audos also believes that AI can be used to do big business. The founders have nothing less in mind than launching 100,000 (!) companies per year—in a planned, replicable, and systematic manner.

The idea behind this is that traditional entrepreneurs may sense that change is underway, but may not necessarily be able or willing to experiment with AI agents. Audos steps into the breach here and helps them find the right niche customers.

Audos takes a 15% cut of the revenue. In exchange, the founders receive financing of up to $25,000, access to AI-powered tools, and sales support. That may sound sexy at first, but a 15 percent revenue share quickly eats into margins.…

Our take: AI is the enabler for everyone

There are AI YouTubers who earn millions. The demand for support in the business environment is high, and AI automation consultants and agencies are springing up like mushrooms.

And, of course, there are tons of niche SaaS products vying for our attention. Thanks to vibe-coding tools such as Lovable, Replit, and others, nowadays, anyone with a cool idea and a dash of perseverance can build something of their own without having to rely on questionable side hustles.

Bottom line: AI is the perfect launchpad if you’re looking to reinvent yourself. This applies to companies, employees, self-employed people, etc. And those who don't take action today will be left behind tomorrow.

🐙 AI without data octopus: Real innovations from Europe

We've been talking for two years about how ChatGPT, Gemini, Claude & Co. are super handy, but come with a few drawbacks—like sending your data off to the US, high prices, and zero transparency.

This is where it gets exciting: the French Mistral surprises with cool free features, and the Swiss Apertus is releasing a genuine open source model. Let's take a closer look.

Mistral: Free features that make OpenAI nervous

Mistral plays the wild card we didn't expect:

• Memories that remember 10 times more than the competition

• 20+ enterprise-grade MCP connectors (see directory here)

• And all of this is available free of charge in the Free Tier!

• Agents, projects, image and code generation, voice, and so on have also been included in the free plan for quite some time.

That's bold. And pretty smart too. Give it a try at https://chat.mistral.ai/chat!

And while others want to get their hands on your wallet for memory, here you get full control for free: delete, edit, and import entries (yes, even from ChatGPT).

OpenAI also seems to be following this development. ChatGPT is now making the project feature available to free users as well (see our post from last week).

Swiss AI Apertus: Open, transparent... and still a little shaky

The Swiss also want to know: Apertus from ETH/EPFL is completely open source (details in the paper on Github). 8B and 70B models, documented training, multilingual, auditable. The complete opposite of ChatGPT, so to speak.

If you want to test it, you can try it at https://publicai.co/chat or directly at Huggingface.

The launch itself was a little rocky with unstable access. And tests show many hallucinations with false facts and confused answers, even in simple knowledge tasks.

The reason for this is the smaller knowledge base, as no unsolicited scraped data from the internet was used. It also has no direct internet access, cannot create images or documents, has no voice output, etc.

In the medium term, however, there are plans to add multimodal capabilities and further updates on topics such as health, law, and education. But right now, Apertus is still catching up and lags far behind ChatGPT, Claude, and Perplexity.

Instead, it offers something else: complete transparency and digital sovereignty without dependence on the US or China! We therefore hope that the Swiss will soon upgrade their systems so that the model becomes a real option for practical use.

A comparison with OpenAI, Llama, Mistral, and Deepseek makes it clear that Apertus is not only compliant with the EU AI Act, but is truly a "transparent model."

Who invented it? 🇨🇭

Open Source/Open Weights: Is Llama & Co. really enough?

Let's stay with the topic of open models for a moment. Llama, Mixtral, and Qwen are already 80-90% at the level of the market leaders!

So the question is no longer: Can they keep up?
But rather: Does your team even need frontier models at many times the price?

With RAG databases and MCP integrations to other platforms, you can build internal workflows that often meet your requirements. The big advantage is that the data stays with you—the disadvantage is that you have to deal with questions about architecture and infrastructure yourself.

Our take: You decide who has access to your data!

In many companies, the motto is: Compliance first! So the real debate isn’t just about which model is best—it’s also about who’s actually protecting your data while you use AI?

As you can see, there are options for anyone who doesn't want to blindly feed the data giants, but instead wants more control and transparency.

And yes, of course you can use Microsoft Copilot or ChatGPT in your company without the data being used for training purposes. Trust is good, but …

🥷🏻 How to use ChatGPT and similar services anonymously

There is another development that fits perfectly into the picture for all privacy-conscious end users: DuckDuckGo is positioning itself with Duck.AI as something like a privacy layer for mainstream models!

You can use ChatGPT, Claude, or Llama there, but through the DuckDuckGo protection layer. The company acts as an intermediary and ensures that your conversations do not end up in the training data of the major providers.

DuckDuckGo protects you through Privacy by Design—they don't collect any data that they could share in the first place. And what doesn't exist can't be passed on to authorities. No user IDs, no IP logging, no long-term storage, chats anonymous for a maximum of 30 days—and that's it.

This is exciting because it shows that even if you don't want to switch to Mistral, Apertus, or other open models, there are now realistic hacks for more privacy when using ChatGPT and similar platforms.

Sure, Duck.AI can't (yet) handle memories or voice chats, and you often have to resort to simpler model versions. But: the price is lower ($10/month instead of $20) and the privacy story is worlds better than on the original platforms.

You made it to the end! Don’t worry—we’ll be back soon with even more updates.

Reto & Fabian from the AInauts

P.S.: Follow us on social media - it motivates us to keep going 😁!
X, LinkedIn, Facebook, Insta, YouTube, TikTok

Your feedback means everything to us. We read every single comment and message, so just hit reply and let us know what you think!