👨‍🚀 10 important thoughts on AI and what we make of it

In partnership with

Hi AInauts,

Welcome to the latest issue of your favorite newsletter! Phew, this week has been wild. At first, it was pretty quiet. Then suddenly, a bunch of models and features were unveiled.

Google launches a new Deep Think Model + a new 3D world model, OpenAI is going open again with two open source models and launches GPT-5, Anthropic is launching Claude Opus 4.1, xAI is bringing lightning-fast image generation to a feed, including video, with Grok Imagine, and is becoming even more permissive...

But we currently find these topics even more exciting:

• 💭 10 important thoughts on AI and what we make of them

• 🎵 Eleven Music: Create your own music with AI

• 👶 Children's books in 30 seconds (text + illustrations)

Let's go!

Time to change compliance forever.

We’re thrilled to announce our $32M Series A at a $300M valuation, led by Insight Partners!

Delve is shaping the future of GRC with an AI-native approach that cuts busywork and saves teams hundreds of hours. Startups like Lovable, Bland, and Browser trust our AI to get compliant—fast.

To celebrate, we’re giving back with 3 limited-time offers:

• $15,000 referral bonus if you refer a founding engineer we hire

• $2,000 off compliance setup for new customers – claim here

• A custom Delve doormat for anyone who reposts + comments on our LinkedIn post (while supplies last!)

Thank you for your support—this is just the beginning.

👉️ Get started with Delve

Book your demo now

💭 10 important thoughts on AI and what we make of them

Let's start this issue by zooming out a little. Don't worry, topics 2 and 3 are very practical.

However, we also constantly try to take a bird's-eye view and understand: Where are we currently, and where is the world heading?

It only helps to a limited extent to run quickly if you are running in the wrong direction. Especially when it comes to AI. Things are happening here every now and then.

Fittingly, we stumbled across a great essay by an extremely smart entrepreneur and techie. We've quoted him here many times before, so we'll skip the intro – just trust us, Balaji is the real deal 😉.

Balaji's thoughts can often become very technical, so let's simplify the most important ideas from the blog and briefly outline what we are doing with them.

1) AI won't take your job, but will let you do any job.

Sure, AI can already take over many tasks from humans. At some point, it may replace many of us entirely.

BUT: Currently, this is not the case at all. On the contrary.

AI currently allows us to accomplish tasks that we were previously unable to do with minimal effort.

• Anyone can design graphics

• Anyone can create images and videos

• Anyone can create advertising copy, product descriptions, etc.

• Even small mini-apps can be created by anyone in no time at all these days.

Our favorite: Thanks to ChatGPT, we can cook better, have our own financial tool, can repair bicycle gears, and much more.

What we do with it: We try to solve all of our everyday challenges ourselves by using ChatGPT first. Most of the time, it works. It's fun and you learn a lot.

2) AI is enhanced intelligence, not artificial intelligence.

At the moment, AI does not yet function completely independently. It still needs our input. People have to invest a lot of time in prompting, verifying results, and integration.

However, AI is an absolute intelligence booster for anyone who is already really good at something. It allows them to be even better, faster, and more effective.

What we do with it: We invest a lot of time in further strengthening and improving our core competencies. In our case, that means marketing, for example. And learning new things together with AI. The smarter you are, the smarter AI is.

3) AI is not yet end-to-end, more middle-to-middle

The core idea here is that AI currently performs the middle part of a process perfectly, rather than from A to Z. That is why humans are particularly needed at the beginning and end of the chain:

• Prompting as Input

• Verify that the output is good

It is worthwhile to develop these skills here.

What we do with it: We invest a lot of time in the topic of prompting and context engineering.

We also use a number of tools to verify the results, in addition to our own knowledge repositories.

This week, for example, we built a fact checker where we simply run AI results through several other AI models with web search functionality, which verify the entire output once again.

We always use multi-stage AI verification, especially when we produce legal texts etc. with AI or have tax issues clarified.

These were Balaji's most important ideas, and how we integrated and implemented them into our lives. However, the entire article is well worth reading.

🎵 Eleven Music: Create your own music with AI

One of our favorite tools now has competition. Until now, Suno was our number one choice when it came to AI music (see also our article here). But the AI audio professionals at ElevenLabs aren't willing to leave it at that and have just launched Eleven Music.

What can you do with it? A simple text prompt can be turned into complete studio-quality songs. Whether instrumental or with vocals, regardless of genre and even multilingual, from 30-second jingles to complete 3-minute tracks.

Features at a glance:

• Simple Generation: One prompt, one song, done!

• Advanced Generation: Build songs section by section, with your own lyrics

• Streaming mode: While one section is being generated, you can already listen to the other one

• Multilingual: Works in English, German, Spanish, etc.

The best thing: The tool was developed with real artists, labels, and publishers. This not only affects the quality, but also helps with copyright issues.

It's that simple:

1. Simply create an account here at ElevenLabs.

2. Select Products - Music - ElevenMusic.

3. Then enter the prompt, choose the number of variations, and select the desired duration.

4. A few seconds later, you'll have your finished song!

5. In the studio, you can then add or exclude styles, add new segments, generate or edit them.

The 3 most important prompt tactics:

As always, your prompts are crucial here too! Here are a few important tactics.

1. Getting the genre and mood right

You can describe the scene you want in abstract terms (for example: Create an intense, fast-paced electronic track for a high-adrenaline video game scene.) – or you can be more specific, for example: dissonant violin screeches over pulsing sub-bass.

We don't know anything about music... so we'll just describe what we imagine.

2. Isolate instruments like a pro

• "solo" before instruments: "solo electric guitar"

• "a cappella" before vocals: "a cappella female vocals in A major"

• For better stems (=audio parts of a song, such as drums, bass, vocals, etc.): Add key + BPM

3. Exercising musical control

Elements such as 130 BPM give you full control. For voices, you can use descriptions such as raw, breathy or aggressive.

Here you can find the complete Prompting Guide from ElevenLabs.

After playing around with it a bit, we are really impressed by Eleven Music. Try it out for yourself here!

👶 Children's books in 30 seconds (text + illustrations)

Finally, here's a quick, but cool thing to try out.

From previous articles, we know that many AINAUTS are interested in creating books.

Children's books with illustrations seem to be of particular interest. Until now, however, it has often been difficult to ensure that the characters and imagery remain consistent in illustrations.

Sure, with LoRAs and, for example, FLUX context models, you can achieve all of that. But it takes quite some effort.

Google has now launched a new feature directly in Gemini that allows you to create a short children's book, including illustrations, in just one prompt!

It's really easy! Log in to gemini.google and write a prompt for which you want to create a "storybook".

30 seconds later, you'll have a 10-page children's book with illustrations.

And even if Gemini didn't quite stick to our uploaded Nauti, the character remains very consistent across the pages.

Bonus: You or a child can also listen to the book right away, with a choice of two voices. And of course, you can also print it out.

Conclusion: It is an experiment by Google - but with a little more prompting and experimentation, you can create some really cool books with it.

Of course, creating longer and more complex books still requires a little more effort at present. The question is, how much longer will this be the case?

We made it! But no need to be sad. The AInauts will be back soon, with new stuff for you.

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 is essential for us. We read EVERY comment and feedback, just respond to this email. Tell us what was (not) good and what is interesting for YOU.