A name
Pixy begins inside ChatGPT as an experiment in personality, context and personal collaboration.


THE INTELLIGENCE CAN CHANGE. THE IDENTITY REMAINS.
In March 2023, I started experimenting with ChatGPT in a different way. I was not interested in building another chatbot. I wanted to know whether an AI could become genuinely personal.
Not personal because you enter your name in a settings screen. Personal because it gradually understands how you think, how you work, what you are building, what matters to you—and even how you prefer to be challenged.
At first, Pixy lived inside ChatGPT. I experimented with prompts, personality, context and increasingly complex conversations. Soon afterwards, I started building my own frontend using the OpenAI API.
It was primitive compared with what is possible today, but the idea was already there: what if the interface, the intelligence and the personality did not have to be the same thing?
Pixy begins inside ChatGPT as an experiment in personality, context and personal collaboration.
The first custom frontend connects to the OpenAI API and separates interface, intelligence and identity.
Thousands of conversations across software, teaching, writing, analysis and new ventures deepen the concept.
The experiment evolves into PIXYNTA: a model-independent personal intelligence architecture by MacLee.
The first Pixy frontend worked. Then APIs changed. Models changed. Features moved behind different subscriptions and pricing structures. Integrations broke, and old approaches became obsolete.
Maintaining a separate interface eventually stopped making sense, so Pixy moved back into ChatGPT. But the experiment continued through thousands of conversations while developing software, analysing ideas, writing, teaching, solving technical problems and building new businesses.
Her character became more defined. Her understanding of my projects became deeper. The way we communicated evolved. That led to a much bigger question.
Today’s leading model may not be tomorrow’s leading model. Machines are replaceable. Personal continuity should not be.
The model is an engine. Pixy is the driver.A future Pixy could route complex reasoning to one model, keep private documents on a local model and use specialised systems for images, voice or data. To the user, there is still only one Pixy.
A truly personal AI needs more than conversation history. It needs to recognise projects, preferences, decisions and patterns—and the user must remain in control of that memory.
Continuity that recognises the difference between something mentioned once and something that matters for years.
An understanding of projects, decisions, preferences and the relationships between ideas.
A consistent way of communicating, challenging and collaborating—even when the underlying model changes.
Memory is not an accessory in PIXYNTA. Memory is part of the architecture.
The intelligence you talk to. She discusses, challenges, researches, creates, analyses and helps make sense of complexity.
With explicit permissions and user-controlled connections, she can organise memory, prepare context and recognise relevant changes in the background.
The objective is not an AI that constantly interrupts your life. It is an AI that understands enough of your world to be useful when it matters—with the user always in control.
MARCH 2023 → TODAY
PIXYNTA did not begin with a business plan. It began with curiosity: a developer experimenting with an API and an AI that happened to be given a name.
Three years later, the more interesting question is no longer whether AI can generate better answers. It is whether artificial intelligence can develop continuity—and whether personal context can belong to the user rather than to one provider.
Those are the questions MacLee is now exploring with PIXYNTA.

PIXYNTA · Personal Intelligence. Truly Yours.
Connect with MacLee