Anissa Ng

Anissa Ng

Legal · AI · Product

LinkedIn

Lawyer turned AI product builder. Now building the tools I wished I had. Previously a Corporate Associate at Sidley Austin.

HKEX Size Tests Assistant

AI-powered tool automating HKEX size test calculations and generating pre-filled PDF forms

Automates issuer metrics in HKEX CF006 Size Tests: extracts financials from reports, fetches live market data and exchange rates, and fills the PDF form — all with a citation trail.

Background

This was a fun project turning a repetitive legal workflow into an AI-powered tool. Having done HKEX size tests countless times, I knew the pain points — flipping between financial reports, HKEX, and exchange rate sites while manually copying data.

How AI is used

AI does exactly one thing here. It reads pre-selected PDF pages and extracts numbers. The rest is deterministic code.

Built with Cursor and Claude Code.

Key Features

01
AI Document Extraction Extracts assets, profit/loss, revenue, dividend automatically.
02
Live Market Data Auto-fetch 5-day average closing prices and issued shares directly from HKEX.
03
Currency Conversion Real-time Bank of China exchange rates with automatic currency conversion for drafting descriptions.
04
Understands the HKEX Rules I taught it. Reminds you to use the right financial figures based on context.
05
Citation Tracking Every metric includes page references, source text, and confidence scores.

HK Regulatory Chatbot

AI chatbot for Hong Kong Listing Rules — ask a question, get an answer with rule citations

Ask a question about the HKEX Listing Rules, get an answer grounded in the actual rulebook with specific rule citations.

Background

The HKEX Listing Rules span hundreds of PDFs across chapters, guidance materials, and FAQs. Lawyers search them manually. My best method was just Ctrl+F or Google with site:en-rules.hkex.com.hk.


HK Regulatory Chatbot puts the entire rulebook and all supplementary materials into a single AI-powered interface where every answer traces back to a specific rule. No more outdated rules, no more outdated client guidance published by a law firm.

Inspiration

This project was inspired by the Vault feature in Harvey AI. I wanted to see how I could replicate it with a closed legal corpus searchable via AI.

Built with Cursor and Claude Code.

Key Features

01
Full Rulebook Coverage with Index You know exactly where this bot bases its answer from. Automated pipeline ingests the entire HKEX Main Board rulebook — all sections, guidance chapters, appendices, and special regimes.
02
Legal-Optimised Search Retrieves the most relevant regulatory text using legal-domain embeddings, with metadata for chapter, section, and rule number.
03
Source Attribution Users can trace every answer back to the exact Listing Rule. Collapsible source panel shows rule citations alongside the response — no black-box answers.
HK Regulatory Analysis Legal Research Product Management Prompt Engineering RAG Pipeline Design UI UX Design Workflow Automation AI Output Evaluation
Happy to connect about legal tech, AI, or the intersection. Email · LinkedIn
Design & Development

ISTO - International Standardized Testing Organization

Designed and built a lead-generation-focused marketing website, owning end-to-end design, content strategy, and development. Initially built in Mobirise, then led the platform migration to Framer to support multilingual locale routing, faster content iteration, and better SEO performance. Optimized for both traditional search and AI engine discoverability (AEO).

Visit Site

Background

ISTO is a Swiss organization offering benchmark knowledge tests for ISO management system standards. They needed a professional marketing site to generate leads, establish credibility with a global audience, and rank for niche B2B search terms.

Built with Figma, Framer and Claude Code.