Antimeridian
/

ANTIMERIDIAN
TECHNOLOGY LIMITED

對向子午科技有限公司

Hong Kong · Software development and data processing services

We build the Public-Document Automation System and sell it to institutional clients as an API, billed as an annual service fee plus call volume.

What we are not
We are not a financial services firm — we do not receive client funds, we do not trade on behalf of clients, and we do not give investment opinions.
香港 · 軟件開發及數據處理服務

我們自研「公開資料自動化處理系統」,以應用程式介面(API)方式銷售予機構客戶,按年度服務費及調用量計費。

我們不是什麼
我們不是金融服務機構 —— 不接收客戶資金、不代客戶買賣、不提供投資意見。
01
Problem

Institutions read more public documents than hand-sorting can keep up with.

Research and analysis departments handle a large volume of public documents and data every day — listed-company announcements, periodic reports, industry statistics, public news. The formats are scattered and the volume is large. The work still depends on people reading and sorting by hand: slow, and expensive in headcount.

  • Artificial intelligence has matured enough that this sorting and summarising work can be largely automated, and institutional demand for such tools has risen noticeably.
  • Comparable products from large international vendors are expensive, English-first and built for European and US markets, weakly localised, and slow to implement.
  • Small and mid-sized institutions have limited budgets and are priced out of those vendors. They want the tooling and cannot afford it.
The gap
That segment is our target market: fast delivery, affordable pricing, built around the client's own workflow.
02
System

A system that turns scattered public documents into standardised data fields.

We build the Public-Document Automation System and open it to institutional clients as an application programming interface (API), callable directly from the client's own systems.

  • Automated collection, cleaning and structuring of public information — exchange announcements, listed-company periodic reports, public industry statistics, public news — into standardised data fields.
  • Summarisation across batches of documents.
  • The product prototype is built and can be demonstrated directly. We are not researching it from scratch.
  • Asset-light: no plant, no inventory, delivered remotely, low cost to start.
Boundary
What we sell is software tooling and data processing services. Clients use them to make their own research and information handling more efficient; every business judgement and decision is the client's own and has nothing to do with us.
03
Products

Three product lines.

One primary interface service, plus two delivery and maintenance lines built around it.

Product one
Data-processing API access
Primary
  • The Public-Document Automation System is opened to institutional clients as an application programming interface (API), callable directly from the client's own systems.
  • Interface functions: automated collection, cleaning and structuring of public information into standardised data fields; summarisation across batches of documents.
Billing: annual base service fee including a call quota; calls above quota priced on a tiered per-call basis.
Product two
Custom development and institutional knowledge bases
  • We turn a client's internal historical documents, meeting records and research archives into a searchable knowledge base — data cleaning, labelling, index building, deployment and staff training.
  • We build automation modules around the client's existing workflow and integrate them with the client's existing systems.
Billing: one-off, priced per project.
Product three
Annual software licence and technical support
  • Annual software licence, version updates, defect fixes and online technical support for clients already implemented.
Billing: annual licence fee.
04
Chain

Purchase, process, sell.

Our value chain has three links, with purchases and sales matched one to one. Cost comes mainly from upstream compute and model interfaces; revenue comes from downstream API service and project delivery.

① UPSTREAM · PURCHASE Compute and model interfaces Compute and general-purpose AI model interfaces purchased from cloud providers (Alibaba Cloud, Volcano Engine) as the underlying input. We do not train foundation models and we do not build data centres. ② IN-HOUSE · PROCESSING Our own processing layer Our own data-processing workflow, field definitions and structuring algorithms turn general model capability and public documents into standardised industry data products. These algorithms and the processed output are our own intellectual property. The cloud providers acquire no rights by supplying the underlying resources. ③ DOWNSTREAM · SALE Sold as an API The processing capability is sold to institutional clients as an application programming interface (API), billed as an annual service fee plus call volume; or delivered as a custom system on a project basis. ① UPSTREAM · PURCHASE Compute and model interfaces Compute and general-purpose AI model interfaces purchased from cloud providers (Alibaba Cloud, Volcano Engine) as the underlying input. We do not train foundation models and we do not build data centres. ② IN-HOUSE · PROCESSING Our own processing layer Our own data-processing workflow, field definitions and structuring algorithms turn general model capability and public documents into standardised industry data products. These algorithms and the processed output are our own intellectual property. The cloud providers acquire no rights by supplying the underlying resources. ③ DOWNSTREAM · SALE Sold as an API The processing capability is sold to institutional clients as an application programming interface (API), billed as an annual service fee plus call volume; or delivered as a custom system on a project basis.
Ownership of intellectual property

The algorithms produced by our own data-processing workflow, field definitions and structuring methods, and the processed output they generate, are our own intellectual property. The cloud providers acquire no rights by supplying the underlying resources.

How delivery works
  • Software systems are deployed remotely to the server or cloud environment the client designates.
  • Processed output is delivered over encrypted network transfer to the location the client designates.
  • Technical support and staff training are delivered by online meeting.
  • Our business involves no physical goods: no warehousing, no logistics, no shipping arrangements.
05
Clients

Who our clients are.

Our clients are the research and operations departments of institutions. The purchasing decision is normally made by the client's information technology department or research department; the engagement is technology outsourcing.

Hong Kong · Singapore
Research and operations departments of small and mid-sized asset management firms, investment consultancies and family offices.
Mainland China
Securities research institutions, corporate strategy research departments and industry consultancies.
Starting out
We start with two or three pilot clients to sharpen the product and the delivery process, then replicate the standardised result across comparable clients.
06
Boundaries

What we do not do.

To avoid any misunderstanding, we set out the boundaries of our business.

01
We do not receive, hold or take custody of any client funds or securities assets.
02
We do not execute transactions in securities, futures, foreign exchange or other financial products on behalf of clients.
03
We do not give buy or sell recommendations on any particular security or investment product, and we do not publish investment opinions or recommendations.
04
We do not charge management fees, performance fees or any remuneration linked to investment returns. All of our income is software development fees and technical service fees.
05
We do not carry on any regulated activity under the Securities and Futures Ordinance (Cap. 571) of Hong Kong.
01
痛點

機構要讀的公開資料,已經超出人工整理的承受範圍。

機構的研究與分析部門,日常須處理大量公開文檔與數據 —— 上市公司公告、定期報告、行業統計、公開新聞等。這些資料格式零散、數量龐大,長期依賴人工閱讀與手工整理,耗時且人力成本高。

  • 近年人工智能技術成熟,使這類整理與歸納工作可以大幅自動化,機構對此類工具的採購需求明顯上升。
  • 國際大型廠商的同類產品價格高昂、以英文與歐美市場為主,本地化程度低,實施週期較長。
  • 中小型機構預算有限,往往被上述廠商排除在外,長期處於「想用但用不起」的狀態。
空缺
這一段客戶群體正是本公司的目標市場:交付快、價格可負擔、按客戶流程定製。
02
系統

把零散的公開資料變成標準化數據字段的軟件系統。

本公司自研「公開資料自動化處理系統」,並以應用程式介面(API)方式向機構客戶開放,供客戶系統直接調用。

  • 自動採集、清洗並結構化處理公開資訊 —— 交易所公告、上市公司定期報告、公開行業統計、公開新聞資訊 —— 輸出標準化數據字段。
  • 對批量文檔進行歸納與摘要生成。
  • 產品原型已完成,可直接向客戶演示,不需要從零研發。
  • 輕資產模式:無廠房、無存貨,以遠程方式交付,起步成本低。
邊界
本公司銷售的是軟件工具與數據處理服務。客戶購買後自行使用,用以提升其內部研究與資訊處理效率;客戶的一切業務判斷與決策均由客戶自行作出,與本公司無關。
03
產品

三項產品線。

一條主力的接口服務,加上兩條圍繞它的交付與維護業務。

產品一
數據處理 API 服務接入
主力產品
  • 自研「公開資料自動化處理系統」以應用程式介面(API)方式向機構客戶開放,供客戶系統直接調用。
  • 接口功能:自動採集、清洗並結構化處理公開資訊,輸出標準化數據字段;批量文檔的歸納與摘要生成。
計費:年度基礎服務費(含調用配額),超出配額部分按調用量階梯計價。
產品二
定製化開發與機構知識庫建設
  • 為客戶將其內部歷史文檔、會議記錄、研究檔案整理為可檢索的知識庫,包含數據清洗、標註、索引建立、系統部署與人員培訓。
  • 按客戶既有作業流程定製開發自動化處理模塊,並與客戶現有系統對接。
計費:按項目一次性計價。
產品三
年度軟件授權與技術支持
  • 為已實施客戶提供年度軟件授權、版本更新、缺陷修復與線上技術支持。
計費:年度授權費。
04
價值鏈

採購、加工、銷售。

本公司的價值鏈由三個環節構成,採購與銷售一一對應。成本主要來自上游算力與模型介面採購,收入來自下游 API 服務與項目交付。

① 上游 · 採購 雲算力與模型介面 向雲服務商(阿里雲、火山雲/火山 引擎)採購算力與通用人工智能模型 接口,作為底層投入。 本公司不訓練基礎模型, 亦不自建數據中心。 ② 本公司 · 加工 自研數據處理流程 以自研的數據處理流程、字段定義與 結構化算法,將通用模型能力與公開 資料加工為標準化的行業數據產品。 上述算法及加工成果均為本公司自有 知識產權;雲服務商不因提供底層資 源而取得任何權利。 ③ 下游 · 銷售 以 API 方式銷售 以應用程式介面(API)方式,將加工 後的處理能力銷售予機構客戶,按年 度服務費及調用量計費;或以項目方 式交付定製化系統。 ① 上游 · 採購 雲算力與模型介面 向雲服務商(阿里雲、火山雲/火山 引擎)採購算力與通用人工智能模型 接口,作為底層投入。 本公司不訓練基礎模型, 亦不自建數據中心。 ② 本公司 · 加工 自研數據處理流程 以自研的數據處理流程、字段定義與 結構化算法,將通用模型能力與公開 資料加工為標準化的行業數據產品。 上述算法及加工成果均為本公司自有 知識產權;雲服務商不因提供底層資 源而取得任何權利。 ③ 下游 · 銷售 以 API 方式銷售 以應用程式介面(API)方式,將加工 後的處理能力銷售予機構客戶,按年 度服務費及調用量計費;或以項目方 式交付定製化系統。
知識產權歸屬

以自研的數據處理流程、字段定義與結構化算法所產生的算法及加工成果,均為本公司自有知識產權;雲服務商不因提供底層資源而取得任何權利。

交付方式
  • 軟件系統以遠程方式部署至客戶指定的伺服器或雲端環境。
  • 數據處理成果以加密網絡傳輸方式交付至客戶指定接收位置。
  • 技術支持與人員培訓以線上會議方式進行。
  • 本公司業務不涉及任何實物貨品,無倉儲、無物流、無運輸安排。
05
客戶

我們的客戶是誰。

本公司的客戶為機構的研究與營運部門。採購決策通常由客戶的資訊科技部門或研究部門作出,性質屬於技術服務外包。

中國香港 · 新加坡
中小型資產管理公司、投資諮詢公司、家族辦公室的研究與營運部門。
中國內地
證券研究機構、企業戰略研究部門、行業諮詢公司。
起步
起步階段先以 2 至 3 家試點客戶打磨產品與交付流程,形成可複製的標準方案後,再向同類客戶複製推廣。
06
業務邊界

業務邊界與合規聲明。

為免誤解,本公司特此明確說明其業務邊界。

01
本公司不接收、不保管、不代管客戶的任何資金或證券資產。
02
本公司不代客戶執行任何證券、期貨、外匯或其他金融產品的買賣。
03
本公司不就任何特定證券或投資產品提供買賣建議,不發佈投資意見或推薦。
04
本公司不收取管理費、業績報酬或任何與投資回報掛鉤的報酬;全部收入來自軟件開發費與技術服務費。
05
本公司不從事香港《證券及期貨條例》(第 571 章)項下的任何受規管活動。