MLOps.community Tool definitions are the new Prompt Engineering
28 snips
Dec 23, 2025 In this discussion, guests Chiara Caratelli and Alex Salazar dive into the intricacies of AI agent tooling. Chiara shares insights from her work at Prosus, focusing on UX and the challenges of clarity in tool definitions, highlighting how ambiguity can lead to edge cases. Alex contrasts traditional APIs with dynamic tools that capture agent intentions to minimize latency. Together, they explore the balance of tool usage, governance, and the future of multi-agent systems, advocating for a structured approach to ensure efficient production-ready AI agents.
AI Snips
Chapters
Transcript
Episode notes
Building A Conversational Food Agent
- Chiara built a food-recommendation agent for iFood used in-app and on WhatsApp to help indecisive users order faster.
- They added voice, buttons, and a swipe UI to match user expectations and reduce typing when hungry.
Tool Definitions Are Intention Contracts
- Tool definitions matter more than naive API wrappers because agents need intention-focused contracts.
- Standardizing tool names and interfaces cut tokens and greatly improved latency and stability.
Encapsulate Workflows Into Agent Tools
- Build domain-specific tools on top of shared low-level services to make agent actions deterministic and fast.
- Encapsulate workflows like "get brochure" so the agent makes one low-latency call instead of many context-heavy turns.