📲 3 ideas for your own AI app (and how to build it)

Have a wonderful day,

Welcome to the latest issue of your favorite newsletter! Today, we have these exciting practical topics for you:

• 📲 3 ideas for your own AI app (and how to build it)

• 🔎 AI web search now also with DeepSeek

• 👯‍♂️ New video model creates perfect deepfakes

Let's go!

📲 3 ideas for your own AI app (and how to build it)

The founder of the AI coding platform Replit shared this video on Tuesday …

We've written about AI coding assistants before. In other words, AI helps people to build apps - even if they have zero knowledge of programming.

In addition to Replit, there are also Bolt, Cursor, Lovable and many more.

All well and good.

But why we think the video above is so important:

"Let AI program an app for you, conveniently from your phone."

All simply via the Replit app.

In other words: you're sitting on the sofa and have a flash of inspiration. Then you simply describe or speak the idea conveniently on your cell phone and AI builds your app right away…

The first steps are actually super easy.

• You describe your app.

• The Replit agent analyzes the idea and creates a plan.

• You confirm this and your app is built, designed, and you can also publish it directly via Replit.

As is so often the case when using AI, the devil is in the detail.

It can quickly lead to a long back and forth. Sometimes you get stuck, frustration is inevitable.

But that doesn't mean you shouldn't give it a try! On the contrary, it has actually become incredibly easy to build and publish mobile apps without any technical knowledge.

Technically, there is hardly anything standing in the way of you and your app millions. 🤑

3 ideas for apps that could be successful

Now that the technology has been clarified, all you need is a brilliant idea for your AI app success.

So-called AI wrapper apps are an exciting approach that works well.

The idea behind this is that you use the existing large AI models such as GPT, FLUX, Minimax etc. to equip an app with cool capabilities.

Your app is just a better or more specific user interface, so to speak, which uses the capabilities of the AI models in the background via an API interface.

There are already many successful examples of this, with the creators generating between 5,000 and 500,000 dollars in revenue per month with their apps. Here is a polarizing video example:

But now for a few ideas that we find exciting - and that are relatively simple to implement:

1) AI image creator app (without subscription)

There are now countless image models and even more AI image apps. The problem is that the apps usually don't have all image models to choose from and often require a subscription.

But we don't want subscriptions and still want to be able to test our prompt with all the important image models.

Function: Based on a prompt, the app creates one image on each of the large image models at the same time. You can then conveniently choose the best one.

2) Senior assistant

A simple chatbot designed for older people. In other words: large buttons, voice input, clear features.

Function: The bot asks for a few important things and then reminds you of important things such as medication to take, doctor's appointments, birthdays, etc.

3) "Healthy alternative" app

You take pictures of food and groceries and get suggestions for healthier alternatives that still taste good. Can also be combined with recipes.

Bonus: Social Media Calendar Assistant

We might even build this app ourselves, unless you're faster 😁..

Basically, you store your company data (such as website, social media, online store, ...) in the app.

The app then always reminds you of certain holidays, festivals, etc. and also suggests a post or mailing tailored to you and your company that you could publish.

In the spirit of: Today is World Ant Day - Get 20% off with the code: Ant.

So, those were just a few ideas, and of course we have even more up our sleeves. 🙂 Just keep reading.

🔥 🔎 AI web search now also with DeepSeek

If you haven't found anything for you in the ideas above, you can also try different AIs to search for ideas.

The latest update of the AI search engine Perplexity is suitable for this purpose.

You now have the option of using two reasoning models for your query - the R1 from DeepSeek and the o3-mini from OpenAI.

The advantage should be that the models work out a better plan for the search, what they are looking for, how they search, which intermediate results are processed, etc. This should make the end result significantly better. As a result, the end result should be significantly better.

Here is an illustration of the entire reasoning process when asked for ideas for AI wrapper apps:

The question is: Is it worth switching to Perplexity with Reasoning, or is ChatGPT with o3 Reasoning sufficient if you already have a ChatGPT subscription?

The good thing is that you can test it yourself - 5 advanced queries per day are free of charge.

We have also compared the whole thing to a simple ChatGPT search query.

Here is the answer from Perplexity first:

Of the five ideas, we thought two were okay. All the others were very general.

Here now ChatGPT with o3-mini Reasoning:

Result: That's about it. Also pretty unspectacular. Except the story starter app is a really cool idea.

We also asked ourselves whether we always need reasoning models for search queries, or whether normal queries are sufficient.

Here is a comparison of the result of a search via ChatGPT-4o:

Also, super generic answers. Very boring and much less research in terms of sources than Perplexity with DeepSeek.

Our take: ChatGPT for simple queries, Perplexity for more details

We have tried the whole thing with several more searches. ChatGPT is sufficient for simple things.

The more specialized the topic is, and the more facets need to be looked at, the better Perplexity performs with DeepSeek Reasoning.

Perplexity is therefore still our number 1 search AI in everyday life, and with DeepSeek Reasoning it has become even better.

But, of course, there is also Google Gemini's Deep Research (for a fee), and the brand-new OpenAI's Deep Research (for Pro users only). These two engines are in a class of their own and put everything else in the shade!

👯‍♂️ New video model creates perfect deepfakes

We don't really like to write about news and topics that you can't try out straight away.

In this case, however, there's already a lot to look at. And it's scary good!

A team led by Chinese company ByteDance (owned by TikTok) has unveiled a new video model that produces incredibly realistic deepfake videos (see also our article from Monday).

The model is called OmniHuman1 - and here's a brief summary of what makes it so incredible:

• It creates super realistic deepfake videos of humans.

• All based on just one reference image or sound snippet.

• The videos are not restricted in length or style!

Crazy, isn't it?

It's not public yet. But it is 100% certain that such a system will come. Our society will soon be faced with the gigantic task of dealing with what is real and what is not ...

You can find more impressive demos here.

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 - that motivates us to keep going 😁!
Twitter, 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.