AnyCompany Leader Workshop · Day 1

The AI Hierarchy — From AI to Generative AI

Click the rings to explore each layer and see how AnyCompany Finance uses them

✨

Generative AI

Create new content

The innermost, most specialized layer. Models like Claude and GPT that generate text, code, reports, and assessments. AnyCompany uses this for merchant risk narratives, invoice summaries, and compliance report drafting.

Scope

Narrowest

💡 Click any ring to explore that layer

The Four Layers Explained

Each layer builds on the one before it — from broad intelligent systems to specialized content generation.

🧠

Artificial Intelligence

The broadest level — any system that simulates human decision-making. Includes rule-based systems and modern ML. AnyCompany's compliance engine uses AI rules for multi-jurisdiction regulatory checks across 6 SEA markets.

📊

Machine Learning

Systems that learn from data instead of following explicit rules. Finds patterns and improves over time. AnyCompany uses ML for fraud scoring — velocity detection, geo-anomaly patterns, card testing identification.

🔮

Deep Learning

Complex neural networks processing vast data through multiple layers. Handles image recognition, NLP, sequence modeling. Powers AnyCompany's document processing — extracting data from invoices, receipts, contracts.

✨

Generative AI

The newest layer — creates new content (text, code, reports). Builds on all previous layers. AnyCompany uses GenAI for merchant risk narratives, compliance report drafting, and conversational finance support.

Traditional ML vs Generative AI

Both are valuable — the right choice depends on your task.

Aspect	Traditional ML	Generative AI
Architecture	Task-specific models (XGBoost, Random Forest)	Foundation models (Claude, GPT, Nova)
Training	One model per task, your data	Pre-trained on massive data, adapted to your task
Best for	Structured data, predictions, scoring	Text generation, summarization, reasoning
Speed	Fast inference (milliseconds)	Slower (seconds), more compute
Cost	Low per prediction	Higher (token-based pricing)
AnyCompany example	Fraud scoring: flag suspicious transactions in real-time	Risk narrative: generate GREEN/AMBER/RED merchant assessment

💡 Key insight for leaders: You don't choose one or the other. The best systems combine both — ML for fast, cheap scoring (is this transaction suspicious?) and GenAI for rich, contextual output (write the risk assessment narrative). The forex rates agent from Day 2 demonstrates this: Python script for rate lookup + LLM for narrative summary.