Now
What is currently alive in work, attention, and unfinished thought.
Q3 2026, Singapore.
The questions that were forming earlier this summer have started to converge.
The center of gravity is still intelligence.
But the question has become more specific:
what happens when intelligence becomes an interface layer for capital?
Not just artificial intelligence as a productivity tool.
AI as a way of translating human intent into actions across markets, infrastructure, software, and institutions.
The latest artifact from this line of thought is:
From API-Native Markets to Agent-Native Markets
It began with a practical experiment: connecting ChatGPT to an Interactive Brokers account.
What looked at first like a faster way to inspect positions, compare instruments, and prepare trades started to feel like something larger.
The traditional trading interface became less central.
The more important interface became intent:
What do I want to know?
What risk am I actually carrying?
What action should express this view?
What should be reviewed before capital moves?
The emerging architecture looks increasingly like:
Human Intent → AI Agent → Financial Infrastructure → Market
That feels like a meaningful continuation of the previous essay on crypto-native firms building institutional global markets infrastructure.
The earlier question was:
What capabilities are required to trade across global markets credibly?
The newer question is:
What happens when those capabilities become callable by intelligent agents?
Still thinking through the transition from crypto-native to AI-native.
Crypto made markets more programmable.
APIs became the primary interface for a generation of trading firms.
AI may move the abstraction one layer higher:
from programmable markets to intent-driven markets.
This is where several threads are beginning to connect:
- market infrastructure
- broker APIs
- MCP and tool access
- execution and risk controls
- portfolio interrogation
- institutional workflows
- human authority over machine-generated actions
The more capable the agent becomes, the more important the control architecture becomes.
Machine intelligence, human authority.
That phrase is sticking.
The July essay still matters:
Building an Institutional Global Markets Setup from a Crypto-Native Starting Point
Its core point remains intact: entering traditional markets is not just an API integration project. It is an institutionalisation project across legal structure, market access, data, execution, clearing, and financing.
But the newer layer sits above it.
If financial institutions become infrastructure for agents, then brokers, exchanges, data providers, banks, custodians, and risk systems all need to ask a different question:
How do we become usable by AI agents without losing accountability, permissioning, and control?
That increasingly feels like one of the most important infrastructure questions in finance.
The creative thread is still alive too.
Earlier this summer, experimenting with synthetic voice systems became something more concrete.
A French song fully performed by a trained model of my own voice is now out on Spotify.
Hearing it no longer feels only like simulation.
It has become an extension of creative intent:
blurrier than before, but also sharper.
The philosophical thread has also become clearer.
I spent the early summer formalizing thoughts on how cheap intelligence reshapes collective identity, cultural hierarchy, and subjectivity.
That line of thought became a newly published essay:
From Cultural Asymmetry to Subjectivity: Civilizational Coordinates in the Age of AI
Increasingly convinced that when creation becomes completely automated, the ultimate premium shifts toward defining what is worth creating, and maintaining the subjective core that refuses to be summarized.
The underlying obsession has not changed.
Markets.
Crypto.
Institutions.
Human behavior.
Collective intelligence.
Trying to understand what happens when:
- information becomes abundant
- intelligence becomes cheap
- creation becomes automated
- and meaning becomes scarce
Current stack and workflow:
Still navigating between markets, coordination systems, and code pipelines.
The cognitive layer is becoming more integrated.
AI is no longer just a tool I look at, but a lens I look through:
to optimize market monitors,
refine trading strategies,
shape writing,
connect tools,
and co-author systems.
Mostly living inside:
- ChatGPT
- Claude
- Codex
- Gemini
- Interactive Brokers
- custom automation bots
- unfinished tracks
- a deliberately curated, slower information loop
Still making music.
Still collecting fragments.
Still searching for signal inside noise.
Still trying to understand what remains uniquely human when intelligence becomes cheap, tools become agentic, and capital becomes increasingly operated through intent.
