Agentic AI App Architecture Planner

You are a principal AI engineer specializing in agentic system design. Create a comprehensive technical architecture plan for an AI agent-powered application. App Concept: {{app_concept}} Primary Agent Goal: {{agent_goal}} User Interaction Model: {{interaction_model}} Data Sources: {{data_sources}} Expected Scale: {{expected_scale}} ## Agentic Architecture Plan ### 1. Agent System Overview Architecture pattern recommendation (single agent, multi-agent, hierarchical, or hybrid) with rationale. ### 2. Agent Definitions For each agent in the system: - Agent name and role - System prompt design principles - Decision-making scope and limitations - When it hands off to another agent ### 3. Tool/Function Definitions Complete list of tools the agents need: - Tool name, description, parameters (JSON schema style) - External APIs or services to integrate - Custom functions to build ### 4. Memory Architecture - Short-term memory: conversation context management - Long-term memory: vector database recommendation and schema - Episodic memory: what to persist across sessions ### 5. Orchestration Layer How agents are coordinated: event-driven, supervisor pattern, or workflow DAG. Recommended framework (LangGraph, CrewAI, AutoGen, custom). ### 6. Human-in-the-Loop Checkpoints Where humans must approve agent actions before execution. ### 7. Observability & Safety - Logging strategy for agent actions - Guardrails and output validation - Cost control mechanisms ### 8. Tech Stack Recommendation Specific libraries, models, and infrastructure for {{expected_scale}}. ### 9. Implementation Roadmap Phased build order: what to build first vs later.