Out of the 8.1 billion people worldwide, 84% have never used AI.
I came across some stats before, showing that only 16% have tried free AI tools, and the paying user ratio is just 0.3%.
So, the level of AI adoption is akin to the early days of smartphones back in 2007.
If you understand prompts, skills, or even just basic chatbot communication, any one of those makes you ahead of the game.
But with AI products popping up left and right, it’s tough to keep pace with all the updates and iterations.
- The newly launched xBubble from dappOS focuses on having AI use AI, AI learning from AI.
It operates on two layers.
- The frontend is called Bubble Pilot, essentially AI using AI.
For example, if you say "Create a comparison dashboard for BYD vs. Tesla," it recognizes what task that is, digs into its repository for a matching SOP, which is a "Standard Operating Procedure," selects the best model and tool combo, and spits out the result.
If there’s no matching SOP in the repository, it falls back to the general Agent to do it from scratch and logs this attempt.
- The backend is called Bubble Engine. Its core function is AI learning from AI.
It monitors which tasks on the frontend keep falling back, which ones aren’t stable enough, and for those tasks, it has the AI programming the Agent to generate several solutions, testing different model and tool combos to find the highest quality result.
The best solution is solidified into a new SOP added to the repository, and it gets tested again before going live.
The more the Pilot runs, the more the Engine learns, and the thicker the repository gets. It’s a snowball effect.
The biggest difference between xBubble and a general Agent is right here.
The general Agent figures things out on the fly every time, making ten attempts at the same type of task can lead to significant quality fluctuations.
xBubble is more like an "experience-gathering" Agent, using polished SOPs for common tasks, and figuring things out for the novel ones, which get integrated into the repository once they’re refined.
-xBubble has two operating environments.
The cloud-based Bubble Computer runs end-to-end projects, PPTs, research, and comparison web pages all in one go. The local Bubble Personal can access files, browsers, and schedules on your computer but only acts upon your authorization, keeping your system clean.
For instance, that 84% of people who can’t write prompts but can say, "Help me get something done." AI should inherently operate this way.
I came across some stats before, showing that only 16% have tried free AI tools, and the paying user ratio is just 0.3%.
So, the level of AI adoption is akin to the early days of smartphones back in 2007.
If you understand prompts, skills, or even just basic chatbot communication, any one of those makes you ahead of the game.
But with AI products popping up left and right, it’s tough to keep pace with all the updates and iterations.
- The newly launched xBubble from dappOS focuses on having AI use AI, AI learning from AI.
It operates on two layers.
- The frontend is called Bubble Pilot, essentially AI using AI.
For example, if you say "Create a comparison dashboard for BYD vs. Tesla," it recognizes what task that is, digs into its repository for a matching SOP, which is a "Standard Operating Procedure," selects the best model and tool combo, and spits out the result.
If there’s no matching SOP in the repository, it falls back to the general Agent to do it from scratch and logs this attempt.
- The backend is called Bubble Engine. Its core function is AI learning from AI.
It monitors which tasks on the frontend keep falling back, which ones aren’t stable enough, and for those tasks, it has the AI programming the Agent to generate several solutions, testing different model and tool combos to find the highest quality result.
The best solution is solidified into a new SOP added to the repository, and it gets tested again before going live.
The more the Pilot runs, the more the Engine learns, and the thicker the repository gets. It’s a snowball effect.
The biggest difference between xBubble and a general Agent is right here.
The general Agent figures things out on the fly every time, making ten attempts at the same type of task can lead to significant quality fluctuations.
xBubble is more like an "experience-gathering" Agent, using polished SOPs for common tasks, and figuring things out for the novel ones, which get integrated into the repository once they’re refined.
-xBubble has two operating environments.
The cloud-based Bubble Computer runs end-to-end projects, PPTs, research, and comparison web pages all in one go. The local Bubble Personal can access files, browsers, and schedules on your computer but only acts upon your authorization, keeping your system clean.
For instance, that 84% of people who can’t write prompts but can say, "Help me get something done." AI should inherently operate this way.