Architectures for Building Agentic AI

2025-12-10

AI Summary

To address the insufficient reliability of proxy-based and generative AI agents, this paper proposes an architecture-centric methodology for constructing highly reliable autonomous agents. Our approach introduces a novel agent architecture paradigm grounded in principled modular decomposition, constrained interface contracts, and explicit assurance loops. We define seven core modules—goal management, planning, tool routing, execution, memory, verification, and security monitoring—and integrate mechanisms including schema-enforced interfaces, least-privilege tool invocation, transactional execution semantics, memory provenance tracking and cleanup, budget/termination governance, and simulation-based pre-execution safeguards. The methodology systematically addresses tool usage, memory management, and runtime control. Furthermore, it explicitly characterizes failure boundaries and prescribes hardening strategies for five prevalent agent paradigms: tool-augmented, memory-enhanced, self-improving planner, multi-agent, and embodied/web agents. Experimental evaluation demonstrates substantial improvements in system reliability and behavioral predictability.

Technology Category

• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Multiagent Systems: Agent/AI Theories and Architectures
• Philosophy and Ethics of AI: Safety, Robustness & Trustworthiness

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

This chapter argues that the reliability of agentic and generative AI is chiefly an architectural property. We define agentic systems as goal-directed, tool-using decision makers operating in closed loops, and show how reliability emerges from principled componentisation (goal manager, planner, tool-router, executor, memory, verifiers, safety monitor, telemetry), disciplined interfaces (schema-constrained, validated, least-privilege tool calls), and explicit control and assurance loops. Building on classical foundations, we propose a practical taxonomy-tool-using agents, memory-augmented agents, planning and self-improvement agents, multi-agent systems, and embodied or web agents - and analyse how each pattern reshapes the reliability envelope and failure modes. We distil design guidance on typed schemas, idempotency, permissioning, transactional semantics, memory provenance and hygiene, runtime governance (budgets, termination conditions), and simulate-before-actuate safeguards.

Problem

• Architectures ensure reliability in agentic AI systems
• Componentisation and interfaces enhance goal-directed tool-using agents
• Design patterns reshape reliability and failure modes analysis

Innovation

• Componentised architecture with goal managers and planners
• Disciplined interfaces using schema-constrained tool calls
• Runtime governance with simulate-before-actuate safeguards