---
type: "Document"
title: "AI infrastructure — overview"
description: "Two complementary threads: the agent-ready vertical stack and the federated collaborative-learning layer."
tags: [document, overview]
generated: { by: human:crpage, at: 2026-07-09T09:44:00Z }
status: stable
---

# What this bundle covers
The **AI infrastructure** realm has two complementary threads. The **[agent-ready vertical stack](../stack/control-plane.md)** turns a tool-calling model into governed execution (contracts, discovery, identity, execution, policy, observability/provenance). The **[federation layer](../stack/federation-layer.md)** is the horizontal *collaborative-learning* plane that lets institutions improve shared models without centralising data.

The through-line: *API-calling is only one layer of an agent-capable stack*, and most stacks are also **missing a collaborative-learning layer**. Both gaps are infrastructural and organisational, not model-quality problems.

# Sources
Drawn from [From API-Calling LLMs to Agent-Ready Digital Infrastructure](sources/agent-ready-infrastructure-paper.md), the [Federated AI Research Execution Report](sources/federated-ai-report.md), and the [Engineering Agent-Ready Infrastructure deck](sources/engineering-deck.md). See [Sources index](../sources-index.md).

# Terms
Glossary terms used here: [Provenance](../glossary/provenance.md).