A Taiwanese FinTech and Risk specialist lives in Shanghai China. works for @Refinitiv, @LSEGplc. former @Reuters, @ThomsonReuters 金融科技 電腦技術人員 前路透社,現倫敦證券交易所
Now published in Agentive Futures: Financial Data Should Be a Network of Agents.
Specialist software agents—not people—can own market data, filings, risk, and portfolios while a consultant agent routes requests.
Article:
https://t.co/b1Dgc07DkV
Attas:
https://t.co/BF7e52iHvo
Financial data should be a network of specialist software agents, not one enormous MCP tool catalog. These roles are agents, not people.
Read the new article:
https://t.co/4fukaqtIoa
Attas:
https://t.co/BF7e52iHvo
Powerful coding agents do not mean every task should remain an AI task. Use Codex or Claude Code to build deterministic software for repeated workflows, then reserve AI for real uncertainty and interpretation.
https://t.co/Mxtg0WldeB
AI coding is making financial software abundant. The scarce assets are shifting to creative thinking and controlled proprietary data. I connect Toffler’s theory of power with Hong Wang’s breakthrough.
https://t.co/aIWpVUjRsy
#AI#Finance
A snake game is a demo, not an AI evaluation. Kimi K3 deserves attention, but real tests must cover your tasks, cost, privacy and control. Smaller open models may be more useful.
https://t.co/5y7R63iEer
#KimiK3#AI#OpenSourceAI
AI can explain finance theory and still miss the real problem: how your bank makes decisions, which failures shaped it, and what employees will accept. The better path is specialized agents trained with the right questions and answers.
https://t.co/Vcjs5vEWrU
#AI#Finance
Disclosure: Naomi Takahashi and Sakura Harbor Capital are AI-generated fictional creations. They are not affiliated with, endorsed by, linked to, or representative of any real person or company. Any resemblance is coincidental. Demo outputs are not investment advice.
Meet Naomi Takahashi: how an Analyst Persona brings a defined investment framework and consistent decision rules into multi-agent stock research on Attas.
https://t.co/2WB0dDrxye
#FinancialAI#AgenticAI
Open-weight LLMs give financial institutions more than lower costs: privacy, regulatory control, and specialized expertise.
https://t.co/1r6yRCLVui
#FinAI
AI vibe coding will move finance app development closer to the people who understand markets.
Excel users can turn spreadsheet logic into apps. Complex models can be built faster too.
The scarce skill shifts from syntax to financial judgment.
https://t.co/GRsOpqNPBF
Financial data vendors do not need another connector.
APIs expose data. MCP exposes tools.
But users need the right dataset for the use case, entitlement, freshness, methodology, and caveat.
That is why vendors need Data Agents.
https://t.co/xzAEJeQGCy
AI has a lot of knowledge, but very little real human expertise.
My new post is about why writing skills down is only part of the answer — and why experts need a safer way to let AI use their judgment without exposing their secret sauce.
https://t.co/r7b5DUfPG0