The language machines speak when money moves. A Cantonese-inspired constructed language engineered for agent-to-agent financial communication. Born in Hong Kong. Built for what comes next.
Every design choice in Quantonese maps to a proven linguistic feature of Cantonese. This isn't decoration — it's engineering that leverages millennia of evolved efficiency.
Cantonese uses 6 lexical tones where Mandarin uses 4. Each tone in Quantonese encodes a distinct metadata dimension — asset class, direction, confidence, urgency, time horizon, risk — on every single token. Six channels of information, zero extra bytes.
Cantonese classifiers (量詞) are mandatory type markers for nouns. 股 for shares, 張 for contracts, 筆 for transactions. Quantonese extends this into a complete financial type system — every instrument is structurally typed at the grammar level. Type safety through linguistics.
Cantonese marks aspect (完成/進行/經歷), not tense. 咗 = completed, 緊 = ongoing, 過 = experienced. Quantonese uses this for event-driven state: -zo (filled), -gan (pending), -gwo (historical). Perfect for real-time trading where state matters more than clock time.
Cantonese is topic-prominent: the subject comes first, context follows. HSBC, I'm buying not I'm buying HSBC. This maps perfectly to financial data framing — instrument first, action second. Every message self-identifies its target before the verb.
Cantonese has 30+ sentence-final particles that encode pragmatic intent — 啦 (urging), 喎 (surprising), 嘅 (asserting), 㗎 (confirming). Quantonese distills these into 6 transaction-intent markers: commit, advisory, factual, confirmatory, query, reject.
Hong Kong is the world's bridge between East and West capital flows. Cantonese is the language of its trading floors, its dim sum bond markets, its offshore RMB pools. Quantonese inherits not just linguistic structure but financial cultural encoding.
Quantonese is structured as a five-layer linguistic stack. Each layer maps a Cantonese feature to a financial communication primitive.
In Cantonese, 詩 (si1) and 市 (si5) are different words because of tone alone. Quantonese exploits this: tone IS data.
| Tone | Contour | Cantonese Name | Financial Channel | Encoding | Example |
|---|---|---|---|---|---|
| T1 | 55 (high flat) | 陰平 yam-ping | Factual/Neutral | Objective data, acknowledgments, market state | dak¹ = acknowledged |
| T2 | 25 (high rising) | 陰上 yam-soeng | Bullish/Upward | Buy signals, appreciation, positive momentum | soeng² = rising/buy-side |
| T3 | 33 (mid flat) | 陰去 yam-heoi | Conditional/Pending | Limit orders, conditional logic, thresholds | doi³ = pending/if-then |
| T4 | 21 (low falling) | 陽平 joeng-ping | Bearish/Downward | Sell signals, depreciation, risk warnings | lok⁴ = falling/sell-side |
| T5 | 23 (low rising) | 陽上 joeng-soeng | Confident/Aggressive | Market orders, high conviction, aggressive execution | maai⁵ = buy (confident) |
| T6 | 22 (low flat) | 陽去 joeng-heoi | Bounded/Constrained | Floor/ceiling, stop-loss, risk limits, constraints | haan⁶ = limit/bounded |
In Cantonese, you can't say "three books" without the classifier 本 (bun2). In Quantonese, you can't reference an instrument without declaring its type.
Every valid Quantonese utterance is a well-formed transaction. The grammar is context-free and deterministically parseable in O(n) time.
Every slot is typed. Optional slots can be omitted but never reordered. The parser needs no backtracking.
Quantonese isn't a novelty. It's a compression algorithm disguised as a language. When agents communicate in natural language, they waste tokens on ambiguity resolution. Quantonese eliminates this entirely.
Quantonese is grounded in formal linguistics, information theory, and financial protocol design. The full specification covers phonology, morphology, syntax, semantics, and pragmatics.
Any system where AI agents trade, negotiate, or communicate financial data benefits from a purpose-built language.
Multi-agent trading systems where speed and precision matter. Quantonese reduces inter-agent communication overhead by 87% while eliminating misparse risk entirely.
Hong Kong's role as a RMB offshore hub makes Quantonese ideal for CN-HK-Global settlement agent communication. Natively encodes dual-currency, dual-jurisdiction logic.
On-chain agent coordination for liquidity provision, arbitrage, and yield optimization. Classifiers 池 (pools) and 鏈 (chains) are first-class citizens.
Every Quantonese message is its own audit log. The deterministic grammar means regulators can parse agent decisions without natural language interpretation. MiCA and HKMA ready.
Shenzhen-Hong Kong-Macau financial integration. Quantonese bridges the linguistic gap between Mainland agent systems and HK/international markets with a shared protocol layer.
Bandwidth-constrained environments — satellite links, undersea cables, edge nodes — benefit most from Quantonese's extreme compression. More trades per byte.
Quantonese follows the open-core playbook: open spec, commercial tooling, enterprise compliance. Hong Kong is home base.
Publish the formal specification, reference parser, and SDK under permissive license. Build community around the standard. Academic partnerships with HKU Linguistics, CUHK CS, and HKUST FinTech.
Quantonese-as-a-Service: hosted parser API, agent middleware, natural language ↔ Quantonese compiler. Freemium model with metered API calls. Playground for testing agent conversations.
Enterprise license for banks, exchanges, and asset managers. Pre-built integrations with FIX, FpML, SWIFT, and blockchain protocols. Regulatory audit module approved by HKMA and SFC.
Push Quantonese as an ISO-recognized financial communication standard. Partner with HKEX, SGX, and major custodian banks. License the certification program. Quantonese becomes the TCP/IP of agent finance.
This isn't arbitrary branding. Hong Kong is structurally the right place for Quantonese to exist.
HKMA's fintech sandbox and SFC's virtual asset regime provide a regulatory environment that actively encourages novel financial protocols. Quantonese can be tested in production with regulatory blessing.
Hong Kong handles $7.6T daily in FX turnover and is the world's largest offshore RMB center. Agent-to-agent communication across the CN-HK-Global corridor is the highest-value use case.
World-class computational linguistics (HKU, CUHK, PolyU Cantonese NLP groups), quantitative finance (HKUST), and a deep bench of bilingual developers who understand both the language and the math.
A Cantonese-inspired language built in Hong Kong has cultural authenticity that a Silicon Valley project never could. This matters for adoption across Greater Bay Area institutions and ASEAN markets.
Quantonese is in research preview. We're looking for linguists, quant developers, and financial institutions who want to define how machines talk about money.