AI Letters #06 - Give Your Agent Hands: Tools, APIs, and the Model Context Protocol
· 3 min read
A team demo'd an agent that "connected to the database." The model described what it would do with the data - in detail, confidently. The database call was never wired up.
Tool descriptions are not documentation. They are the interface between language and execution. Get them wrong and the agent calls the wrong tool, at the wrong time, with the wrong arguments.
Part 2 of 5: tool design rules, a production tool registry, safe execution patterns, and the Model Context Protocol - the standard that is quietly becoming the default way agents talk to the world.
This letter was first published on AI Letters →
Continue reading on Substack →
Interactive Data Stories
Interactive Chart
The Evolution of Tool Use in AI Systems →
From 1980s expert systems to MCP (2025) - click each milestone to understand what changed and why it mattered.
Interactive Chart
4 Tool Integration Patterns →
Inline tools vs tool registry vs tool chains vs MCP server - click each to see execution flow, code pattern, and when to use it.
Interactive Chart
Tool Calling Evidence Dashboard →
Accuracy by model, description quality impact (41% → 94%), and MCP adoption trajectory - fully interactive.