John Snow Labs 228d ago

Sushant Mehta, Post Training Research at Google DeepMind, presents "Building Effective Agents." Timestamps: 00:00 Intro: Why Post-Training Matters for AI Agents 01:29 Post-Training vs Pre-Training (RLHF, DPO, PPO) 07:01 Reinforcement Learning & Verifiable Rewards 10:44 What Are AI Agents & When to Use Them 15:48 Core Building Blocks: Retrieval, Memory & Tools 17:04 Agent Design Patterns: Sequential, Routing & Evaluator Loops 24:02 Coding Agents & Real-World Use Cases 25:20 Conclusion: Desigions Framework Large language models can now power capable software agents, yet real-world success comes from disciplined engineering rather than flashy frameworks. Most reliable agents are built from simple, composable patterns instead of heavy abstractions. The talk introduces several patterns that add complexity and autonomy only when it pays off: • Augmented LLM (retrieval, tools, memory) as the atomic building block. • Workflow motifs: prompt chaining, routing, parallelization, with concrete criteria and implementation tips. • Autonomous agents that loop through plan-act-observe-reflect cycles to tackle open-ended tasks. Attendees will leave with a practical decision framework for escalating from a single prompt to a multi-step agent, reference implementations they can reproduce in a few lines of code, and robust guardrails for shipping trustworthy, cost-effective agents at scale. Official session recording from the Applied AI Summit 2025 Connect with us: Our website: https://www.johnsnowlabs.com/ LinkedIn: https://www.linkedin.com/company/johnsnowlabs Facebook: https://www.facebook.com/JohnSnowLabsInc X: https://x.com/JohnSnowLabs #BuildingAgents #AgenticAI #AIAgents #ReinforcementLearning #PostTraining #RLHF #DeepMind #AIEngineering