I created this launch video for Tony Bianco using both CGI and AI. This is how I did it. The background was generated with ChatGPT image gen 2, denoised with Gemini. Then, because AI video can reduce quality quite a bit and we wanted to have the people in the store moving, I animated the video with Seedance but mask cropped it so that the only visible part of the video is the store and the people inside moving. For the banner animation, this was done completely with CGI, as well as the taxi. The lighting was done well enough that everything in the scene feels real, but nothing was shot in real life!

Really clean work

🚨 MOST PRODUCTIVITY APPS DON’T HAVE A RETENTION PROBLEM.

THEY HAVE A VALUE-COMPOUNDING PROBLEM.

They help users capture more: 📝 More notes 🔗 More links 📸 More screenshots 🎙️ More voice memos

But a larger archive does not automatically become a more valuable product

The real test starts later:

❓ WHAT HAPPENS AFTER THE USER HAS ALREADY CAPTURED 300 THOUGHTS?

That is where most products quietly fail. Because the archive grows. But the user’s understanding does not. A voice note from Monday. A screenshot from Thursday. An article saved two weeks later. A sentence typed at 1 AM.

Individually, they look unrelated. But sometimes they are the same unresolved question appearing in different forms.

And this is where AI becomes genuinely interesting. Not when it summarizes another note. Not when it rewrites another paragraph. Not when it gives users another chat box.

But when it recognizes: 🧠 YOU HAVE BEEN CIRCLING THE SAME PROBLEM FOR 12 DAYS.

For this concept, the core mechanic was built around one distinction: ⚡ FREQUENCY IS NOT REPETITION. REPETITION IS EVIDENCE. A thought mentioned once may be noise. A thought mentioned four times across different contexts may be a signal. And a product that can detect that signal can create value a normal archive cannot.

01 - CAPTURE

🎙️ The thought should enter the system before it disappears.

No folders. No tagging. No forced structure.

02 - UNDERSTAND

AI extracts the underlying meaning.

“Maybe onboarding should get people to value faster…”

becomes:

💡 REDUCE TIME TO FIRST VALUE.

That is already more useful than transcription. But it is still not the interesting part.

03 - CONNECT

The system sees another thought about activation.

Then an article about onboarding. Then a screenshot. Then another voice note. Now the product has enough context to say:

🔁 YOU KEEP COMING BACK TO THIS.

4 mentions. 12 days.

At that moment, the system is no longer helping the user remember information.

It is helping the user recognize importance.

04 - TURN CONTEXT INTO VALUE

This is where the growth logic changes. Most retention loops ask users to come back and repeat the same action:

📝 Create another note 📊 Track another habit 🍽️ Log another meal ✅ Complete another task

A stronger loop can work differently:

📈 EVERY NEW INPUT INCREASES THE VALUE OF PREVIOUS INPUTS.

The product becomes better not simply because the user used it more often, but because accumulated context creates new insight. That creates compounding value.

Day 1: The product understands what you said.

Day 10: It understands what keeps repeating.

Day 30: It may understand what deserves action.

That is a much more defensible form of retention. Because the user’s history is no longer just stored data. It becomes product intelligence. ✨ AI SHOULD NOT JUST HELP USERS PRODUCE MORE.

IT SHOULD HELP THEM NOTICE WHAT THEY ARE ALREADY REVEALING THROUGH THEIR OWN BEHAVIOR. That is where a note-taking product stops being an archive. And starts becoming a thinking system.

👇 FULL CASE STUDY BELOW

👉 BOOK A MOBILE APP SESSION

Check it out

Wow Amazing Project😍