Permission before capture.
People should understand what is collected, why it is needed, where it goes, and how long it remains.
AN INDEPENDENT BUILDER / CHINA 01—27
Intelligence
shouldn't be a luxury.
I build and explore tools that make powerful technology more accessible — from low-cost intelligent aircraft to creative AI software — while keeping people in control.
ENGINEERING × ACCESS × HUMAN AGENCY
XBBUILD / EXPLORE
TECHNOLOGY / FOR MORE PEOPLE
DESIGNING FOR ACCESS 001
NOT MORE POWER.
MORE POSSIBILITY.
MORE POSSIBILITY.
Abstract artwork · not a product or flight-test image
SCROLL TO DISCOVER2026 / SELECTED WORK
“The real value of technology is measured by who gets to use it.”
— A PERSONAL CONVICTION
01 / THREE PROJECTS
PRACTICE · REFLECTION · MISSION
Three projects.
Three kinds of progress.
One project changed how I think. One shows what I can build today. One points toward the engineering work I want to pursue next. They matter for different reasons.
AIHUMAN AGENCY
PERSONAL CONTEXTPRIVACYCHOICE
A PROJECT THAT CHANGED MY THINKING
THE MOST THOUGHT-PROVOKING PROJECT
Xiaobing Notes
I built a personal-assistant prototype to make useful information and guidance more accessible. Developing it made me question how easily constant assistance can become surveillance, dependence, or influence over a person's decisions. I stopped the intrusive direction — and kept the question.
PRIVACY · PERSONAL AI · HUMAN AGENCY
CREATIVE WORKFLOWCAMERA / COMPOSITION
IDEA · FRAME · CREATEXB AI STUDIO
THE STRONGEST PROOF OF EXECUTION
XB AI Studio
My most developed project: a conversation-centered creative workspace for AI filmmaking, bringing project organization, shot planning, 3D blockout previews, directing decisions, and continuity into one workflow. It includes a real Blender previsualization and an interactive product tour.
CREATIVE TOOLS · 3D PREVIS · DIRECTING
AVIATION × AI ENGINEERING STUDY
CONCEPTUAL SCHEMATIC / NOT A FLIGHT RECORD
THE LONG-TERM ENGINEERING MISSION
Low-Cost Intelligent Aircraft
Exploring lower-cost intelligent flight through modular aircraft systems, flight-control software, simulation, and sensor experiments. The evidence includes ArduPlane SITL simulation and independent phone sensor capture — not a verified autonomous real-world flight.
FLIGHT CONTROL · SIMULATION · HARDWARE
Evidence over promises.
XB AI Studio includes an authentic Blender proxy-scene preview. The aircraft project includes recorded ArduPlane SITL telemetry, a 72-case test campaign with both passes and failures, and an independent Android sensor-capture test. Each project page separates verified evidence, illustrative interaction, and unfinished work.
02 / THE ORIGINPERSONAL HISTORY
Opportunity
changes everything.
I learned early that where you begin can shape which opportunities you can reach.
I grew up in a family with limited resources. I learned early that opportunity is not distributed evenly, and that the options available to a person can be shaped by circumstances they did not choose.
In fourth grade, I noticed that sky lanterns could be bought online at a lower wholesale price than they sold for locally. I saw an opening and acted on it. I started selling them myself and reached my first RMB 10,000 business milestone in roughly one to two weeks. It was a small venture, but it taught me something lasting: recognizing an opportunity matters only when you are willing to act.
Years later, when AI tools became genuinely useful for real work, I recognized another opportunity. I began using them immediately — not only to get answers, but to build products of my own. XB AI Studio, my intelligent-flight-control research, and Xiaobing Notes grew from that decision. This website is part of the same process: I used AI as a tool, while directing the ideas, reviewing the results, and deciding what to improve.
This application is another first. Before this, I had never tried to speak English at this length. I chose to record these videos anyway — not because I felt completely ready, but because I saw an opportunity worth taking. I am learning as I go, and I would rather take a meaningful first step than wait for perfect confidence.
01Recognize the opening
Look beyond the obvious and notice where a useful opportunity exists.
02Act before perfect
Turn an observation into a real attempt instead of waiting for ideal conditions.
03Build with AI
Use emerging tools to turn ideas into practical products and wider access.
03 / THE QUESTION THAT REMAINEDXIAOBING NOTES
A PROJECT CAN STOP. A QUESTION CAN KEEP MOVING.
The more I built,
the more I questioned.
Xiaobing Notes began with a generous ambition: make useful information, context, and thoughtful guidance more accessible to people who have fewer resources or less access to expertise. I wanted AI to help people see more possibilities — not make decisions for them.
Building the prototype changed the question. I began to see that an assistant designed to observe, remember, and advise continuously could also become intrusive. The challenge was no longer just how to make AI more helpful. It was how to make sure that help did not quietly take control.
Explore the full Xiaobing Notes project
ORIGINAL NOTE / 04 JUN 2026
04 / WHERE IT BEGAN
AN EARLY THOUGHT — NOT A FINAL POSITION
The first idea was simpler than the question that followed.
On June 4, 2026, near the beginning of development, I wrote a short reflection inspired by a fictional whispering earring — an assistant that could offer guidance throughout a person's life. At that point, I was only beginning to imagine the appeal of an always-available guide. I had not yet understood the risks that would emerge as I explored the product.
That original entry matters because it records where my thinking started, not where it ended.
Read the original Chinese entry and its English translation05 / THE TURNING POINTFROM PRODUCT TO PRINCIPLE
When assistance starts
to take over.
I developed a prototype exploring recording, transcription, personal-context organization, and contextual suggestions across mobile and desktop. As I considered how such a system might become part of everyday life, I began to recognize a deeper risk: constant observation and advice could weaken privacy and independent judgment.
01 / THE INTENTION
Make useful guidance easier to access.
I wanted people with fewer resources, less experience, or limited access to information to have a tool that could help them understand situations, reflect, and make better-informed decisions.
The prototype and its local interface demonstration are not presented as a connected public service. The separate project page explains what was explored and what the demo does not do.
06 / AN OPEN PRIVACY QUESTIONRESEARCH IDEA / NOT IMPLEMENTED
PERSONAL AI · DATA · MODEL ADAPTATION
Could a model learn preferences
without keeping a searchable diary?
I am exploring whether user-approved patterns could be reflected in model parameters or a separate personal adapter, instead of relying on a permanent, searchable record of everyday life. This is a research hypothesis — not an implemented feature and not a claim that weights are automatically private.
COMPARE THREE APPROACHESSELECT A MODEL TO EXPLORE
Personal records→Searchable store→Retrieved context→AI response
WHAT IT DOES WELL
Reliable access to recorded events.
External records can support precise retrieval of specific conversations, dates, and details when the right information is stored and indexed.
THE TRADE-OFF
A persistent personal data store.
It creates an ongoing record that must be protected from over-collection, unauthorized access, breaches, and uses the person did not intend.
User-approved patterns→Personal adapter+Base model→AI response
WHAT IT MIGHT SUPPORT
Learning stable preferences.
A separate adapter could potentially reflect recurring preferences, communication style, or user-approved patterns without retrieving a complete life log for every response.
THE UNRESOLVED RISK
Weights can still reveal information.
Personal parameters may memorize sensitive data, are difficult to inspect or selectively edit, and may be hard to make reliably forget. They are not a dependable database for exact past events.
User-selected context→Short session→AI response→Discard
WHAT IT DOES WELL
Less long-term retention.
Only information deliberately provided for a particular task is used, with a clear intention to discard it when the session ends.
THE TRADE-OFF
Less continuity between sessions.
People may need to provide context again. Temporary processing still requires secure handling, clear consent, and honest deletion boundaries.
Important distinction: model weights are not a safe vault, and learned parameters cannot reliably replace exact-event retrieval. Deleting raw recordings does not prove that sensitive information has been removed from a trained model.
This comparison describes possible design directions, not a system I have implemented. I still do not know which approach — or combination of approaches — would be safe enough.
Explore the deeper privacy research and open risks07 / THE HUMAN BOUNDARYDESIGN PRINCIPLES
The person must remain
in control.
Technology should increase the range of choices a person can make. It should not make the person feel that the system knows better, or that they must follow its advice.
Advice without authority.
AI can surface options, evidence, and uncertainty. The final decision belongs to the person.
The right to step back.
People must be able to inspect, correct, reset, pause, or refuse an assistant without losing control of their own lives.
04 / WHAT'S NEXT
I want to learn
from the world.
And build for it.
Beyond China, I want to meet builders with different experiences, understand new problems, test my assumptions, and learn how to develop tools that people can afford — and actually use.
KEEP BUILDING. KEEP QUESTIONING.
THE MEANING BEHIND THE MARK
A roof above.
A person within.
The upper lines evoke a roof; the lower strokes form a human figure beneath it. Together, they express a simple belief: progress should make room for people — offering shelter, protecting dignity, widening access, and creating more opportunities.
SHELTER · DIGNITY · ACCESS · OPPORTUNITY
09 / THE FIRST COMPANY
FROM AN IDEA TO A REGISTERED COMPANY
My first company
is built around flight.
I have registered Zhengzhou Xiaobing UAV Co., Ltd. as the starting point for a long-term goal: developing and eventually manufacturing practical, affordable unmanned aircraft.
The company is at an early stage. Registration is complete, but manufacturing has not yet begun. I am not presenting a finished factory or a market-ready aircraft. I am putting a formal foundation beneath a direction I intend to pursue.
Why I chose aircraft manufacturingRead the company story and see its registration evidence
REGISTERED IN ZHENGZHOU, CHINA
THE BEGINNING, NOT THE FINISH LINE