Regulated Intelligence
The contradictions inside AI governance in regulated markets,
named, decomposed, and resolved.
Most AI programs in regulated financial services are not slow because the technology is immature. They are slow because the contradictions inside them are misnamed. Speed versus quality. Innovation versus compliance. Autonomy versus auditability. Each one looks like a trade-off. None of them is.
This publication takes those contradictions apart, one at a time, using TRIZ — the Theory of Inventive Problem Solving, and rebuilds them as structural design choices. The frame is named on purpose: TRIZ for AI. The discipline is engineering, not opinion. The output is a method you can apply on Monday.
What you’ll find here
Essays (free, ~2 per month). Each essay names a single contradiction, decomposes it with a TRIZ inventive principle, and traces the implication into a specific regulatory frame — MAS in Singapore, IMDA Agentic AI v1.5, ISO 42001, NIST AI RMF, the broader trajectory of Asian regulatory guidance. No theory without an artefact. No artefact without a forcing question.
Prompt Kits (free, paired with each essay). A structured prompt template you can run against your own program. Not a “ChatGPT productivity hack.” A diagnostic instrument. Kit 01 — Tiered Governance Diagnostic. Kit 02 — AI CoE Governance Architecture.
Deep dives (paid tier, launching when reader base supports it). Long-form architecture work — reference frameworks, vendor comparisons, regulatory cross-walks, incident taxonomies. Currently configured, not yet activated. Free essays remain the spine.
Who this is for
Heads of AI, CIOs, CTOs, Chief Risk Officers, GRC leaders, and the AI Centres of Excellence sitting between them — in financial services, insurance, wealth management, and adjacent regulated sectors across Singapore, Hong Kong, Dubai, Bermuda, Ireland, and the broader MAS / IMDA / DIFC / BMA / CBI orbit.
If you have ever sat in a steering committee where “we need to move faster on AI” and “we cannot relax governance” were said in the same fifteen minutes, this publication is written for you.
If you build, sell, or regulate AI for these markets but do not sit in the seat — you are still welcome. The frame travels.
Who writes it
JL CREPPY — senior IT leader, Head of Technology Innovation at a high-net-worth insurance group operating across six jurisdictions (Singapore, Bermuda, USA, Ireland, Dubai, Hong Kong). Currently runs the group AI Centre of Excellence and leads agentic AI, generative AI, and automation deployment under MAS and PDPA frameworks.
This publication is independent. It is not endorsed by, written on behalf of, or representative of any employer. All views are personal. All examples are sanitised, generalized, or composite — never sourced from confidential forums or specific institutions.
Cadence
- Essays: ~2 per month (free)
- Prompt Kits: paired with each essay (free)
- Deep dives: activated when reader base supports the tier (paid, configured)
- Editorial discipline: every piece passes a three-test filter before publish — *named contradiction · regulator-checkable claim · artefact you can use*
What this publication is not
- It is not vendor analysis dressed as thought leadership.
- It is not a personal brand exercise.
- It is not “AI explained for executives” — the assumed reader is already running a program.
- It is not anonymous. The author’s identity, role, and jurisdictional context are stated above so the reader can calibrate the lens.
Subscribe
Free subscribers receive every essay and every Prompt Kit by email at publication.
Paid tier — when activated — unlocks deep dives, reference architectures, and the full Prompt Kit archive. Founding Member tier will be available at the same time.
> One question to leave you with:
Which contradiction inside your AI program has been costing you the most time, and who in your organization has the standing to name it?
Regulated Intelligence · TRIZ × AI · Regulated Markets
www.regulated-intelligence.com
