APIs

Overview

APIs (Application Programming Interfaces) are programmable interfaces that let software systems—and increasingly AI agents—access capabilities, data, and services in a structured way. In the context of AI products, APIs are one of the core mechanisms that allow agents to take actions on behalf of users, such as retrieving information, triggering workflows, updating records, or completing end-to-end tasks across products.

For AI Product Managers, APIs matter because they shift product value from just user-facing screens to machine-accessible functionality. As agents become a more common interface layer, products that expose reliable, well-designed APIs are better positioned to be invoked by assistants, copilots, and autonomous workflows. In the newsletter, APIs are framed not merely as developer tooling, but as foundational infrastructure for the agent era.

Key Developments

• 2026-02-22 — Peter Yang argues that in the AI agent era, product teams should aim to reduce direct user time spent in their product by enabling agents to complete tasks seamlessly through APIs, skills, and MCPs.
• 2026-02-25 — Peter Yang explains APIs, skills, and MCPs using a professional kitchen analogy, positioning APIs as the kitchen itself: the underlying environment where capabilities exist and work gets done.

Relevance to AI PMs

• Design products for agent access, not just human navigation. AI PMs should evaluate which product capabilities can be exposed programmatically so agents can complete common user jobs without relying on manual UI flows.
• Prioritize API reliability, permissions, and task completeness. An API that is technically available but incomplete, inconsistent, or hard to authenticate against will limit agent usefulness. PMs should define high-value action paths and ensure the API supports them end to end.
• Use APIs as a growth and distribution layer. In an agent-driven ecosystem, products may be discovered and used through external assistants rather than direct app engagement. PMs should treat APIs as a strategic surface for adoption, retention, and ecosystem integration.

Related

• Skills — Skills often package or orchestrate specific actions that can be executed by an agent, frequently relying on underlying APIs to actually perform the work.
• MCPs / MCP — Model Context Protocol-related tooling helps agents connect to external tools and systems; APIs are often the service layer those connections ultimately call.
• Peter Yang — Peter Yang is the source of the newsletter references that frame APIs as a key building block for agentic products and explain them through the kitchen analogy.