Build Your Voice AI Agents - Context Engineering (Day 2)

Welcome to Day 2 of the Tough Tongue AI Voice AI Workshop. This session goes deep into **context engineering**, the most important skill for building reliable, production-grade AI agents. Prompt engineering alone is no longer enough. Modern agents need a carefully designed workspace that includes system prompts, tools, memory, knowledge bases, and guardrails. In this session, we show how to design that workspace, walk through a live agent demo, and break down what’s happening behind the scenes. ==================== What this session covers ======================== * What context engineering really means, and why it matters more than just prompts * How modern AI agents are built using models, tools, and state * A live demo of a coaching agent that remembers users across sessions and uses tools during conversation * How to design effective system prompts, including role, personality, and conversation flow * How and when agents should use tools like MCQs, image generation, or external actions * How to structure knowledge bases and use just-in-time retrieval instead of stuffing prompts * Memory patterns for agents: what to remember, what to forget, and how to summarize long conversations * Guardrails and safety patterns to prevent misuse and off-script behavior * When and why to break large agents into smaller, specialized sub-agents ==================== Live Q&A topics =============== * How much information is too much in a system prompt * How structured your knowledge base needs to be * Positive vs negative prompting (what to do vs what not to do) * How to design memory when you don’t have deep domain expertise * Usage limits, access configuration, and recording options * Using transcripts to improve agent behavior over time ==================== What’s next (Day 3) =================== We move from agent design to building the full application layer and embedding your agent into a real product. ==================== Hashtags ======== #ContextEngineering #AIAgents #VoiceAI #AgenticAI #ConversationalAI #GenerativeAI #LLM #SpeechToSpeech #RealtimeAI #AIProductBuilding #AIEngineering #AIDevelopers