cva-concepts-agent-types

What problem does it solve?

Developing AI agents in regulated domains often leads to over-engineered or under-optimized systems. This Skill provides a clear A/B/C/D taxonomy to classify agents by their data access capabilities, enabling developers to select the most appropriate agent type for a task, thereby optimizing cost, latency, and complexity.

Core Features & Use Cases

• A/B/C/D Taxonomy: Understand agent types from Pure AI (Type A) to maximum context (Type D - AI + Database + Web), with detailed cost and latency profiles.
• Decision Tree: Use a structured approach to select the optimal agent type based on whether external data or tenant-specific database context is required.
• Cost & Performance Optimization: Learn how to reduce costs by up to 67% and improve latency by selecting cheaper, simpler agent types when appropriate.
• Use Case: When building a healthcare content pipeline, use the taxonomy to assign Type A agents for simple data extraction, Type B for personalized SEO, Type C for scientific reference search, and Type D for final, compliance-critical consolidation.

Quick Start

You need an agent to extract entities from text. Consult the decision tree to determine if it needs database or web access. If not, select a Type A (Pure AI) agent for lowest cost and fastest execution.