About · Suit Up

The site runs as a research operation, not a publication. An autonomous ingestion pipeline captures signal across regulators, banks, and trade press. An LLM classifier scores relevance and importance. A structured wiki carries the operator-level reference content. A weekly editorial fires when a development warrants attention. Built for practitioners making a decision, not journalists tracking a story.

Who this is for

Primary readers are practitioners making decisions inside the institutional tokenisation stack, wherever that stack sits. Bank treasury and digital-asset desk PMs at HSBC, Standard Chartered, DBS, MUFG, Mizuho, and their GSIB peers in the US and Europe. Asset manager distribution leads at BlackRock, Franklin Templeton, Janus Henderson. Regulators and ex-regulators, from the reference regimes (Fed, OCC, SEC, ESMA, FCA) to the APAC perimeter the coverage goes deepest on (MAS, HKMA, FSA Japan, FSC Korea). Platform PMs and BD at Kinexys, Securitize, Zodia, Centrifuge, Ondo, Partior, Marketnode.

Secondary readers: Web3 protocol operators moving into TradFi roles, TradFi practitioners newly assigned to a tokenisation desk, journalists and analysts covering the space, and founders building on institutional rails (the agentic commerce wedge audience).

What this covers

Coverage is global. The US and EU are where most issuance and the reference regulatory regimes sit, and they are covered as first-class subjects, not as context for a regional story. When the SEC moves, when a GSIB commits balance sheet, when MiCA implementation shifts the perimeter, that is wire coverage on its own merits.

Asia is the named depth specialty. APAC programme architecture (Project Guardian, Project Ensemble, Progmat, mBridge), Japanese and Korean policy trajectories where English-language coverage is thin, and the regulatory nuance operators actually need before committing a desk to a market. APAC deep dives are deliberate and held to a higher bar than wire coverage: operational implications, named desks, programme architecture, never headline summaries.

The coverage shows up across a few surfaces. A news wire, importance-classified across the global feed with Asia given deliberate extra weight, plus a weekly briefing and a catch-up record of what changed. A 4-layer tokenisation stack map (Issuer, Platform, Custody, Settlement) with named players in each. An institutional wiki: 110+ entity profiles, 11 jurisdictions, 35+ themes, and a 51-part foundations library. Curated reading paths. A jurisdiction activity rail, Asia first. And a deliberate editorial wedge: agentic commerce on tokenised rails, where AI agents start holding tokenised money.

How it works

Three layers, each visible in the codebase and documented in the repository.

Ingestion. A Python pipeline runs on a six-hour cadence against curated RSS feeds, HTML scrapers for bank and regulator press rooms where RSS is weak, and headless scrapers for the JavaScript-rendered regulator sites that resist plain fetching. Coverage of those headless sources is expanding rather than complete: where a source cannot yet be reached cleanly, it is left out rather than half-captured. New entries land as markdown files with structured frontmatter in the project repository, archived at the time of citation.

Filtering. Each new entry passes through two LLM passes. A relevance check answers "is this on-topic for institutional tokenisation?" An importance classifier scores it 0-3 against a calibrated rubric: structural developments score 3 (DBS-Kinexys interoperability framework, GENIUS Act passage, EnsembleTX live phase); meaningful new use cases or workstream outputs score 2; routine updates 1; noise 0. The rubric is in the repository under ingestion/IMPORTANCE_CLASSIFIER.md for inspection. The homepage news feed prefers high-importance items; lower-scored entries stay in the wiki layer where they belong.

Editorial. AI does the heavy lifting: it ingests, deduplicates, scores relevance and importance, and drafts, the news summaries, the weekly briefing, and the one-line operator reads. A human editor reviews and decides what publishes. The operator reads are gated: none renders as an editorial read until the editor approves it, and that approval state lives in version control. The model classifies and drafts; the editor owns the judgment, the sign-off, and the byline.

Editorial stance

Tokenisation is treated as a normal extension of financial infrastructure: neither emancipatory nor inevitable. What gets published is operator-grade analysis, not commentary. Sources are archived locally and cited at entry-level. Corrections are dated and preserved. The intersection of tokenised money and AI agents is where the next category emerges; very few institutional sources cover it well, and that intersection is the deliberate editorial wedge here.

What the AI does and does not do

The system is augmented, not generated. Concretely:

Does: ingest from RSS and scraped pages on a six-hour cadence, deduplicate against a SQLite index, tag entries against a controlled vocabulary, score relevance and importance, draft the news summaries, the weekly briefing, and the one-line operator reads, and surface high-importance items via the news wire.
Does not: publish an operator read without editor approval (the approval gate is enforced in the build), render an editorial verdict on its own, or run unattended without human review. The drafts are the model's; the published judgment is the editor's.

Every load-bearing claim in the wiki traces to a source captured at the time of citation. The relevance and importance rubrics are version-controlled and tunable. The pipeline code is open to inspection in the same project. See Source policy for the underlying handling rules.

About the editor

Suit Up is edited by Cliffton Lee, who works with global asset managers, banks, and the protocols building the infrastructure they connect to: scoping mandates, structuring deployments, and executing tokenisation programmes that hold up under legal, compliance, and treasury review. That work includes leading the USD 70 million Stable Treasury Endowment Programme (STEP) at the Arbitrum Foundation, and partnerships with BlackRock and Securitize, Franklin Templeton, WisdomTree, and Ondo, the same managers and rails tracked most closely here. Fuller background at clifftonlee.vercel.app.

Singapore-based. Suit Up is independently operated and not affiliated with any of the institutions it covers. Contact: cliff@suit-up.io.

Editorially independent. Not investment advice. Not a registered investment adviser.