Engineering Platforms for the Agentic Era

APIs, AI Agents, and the Foundations of Production-Ready Platforms

🇬🇧 English

Building Platforms for Humans and Agents

AI agents are becoming active users of APIs, infrastructure, and developer platforms. This changes what a production-ready platform needs to provide: agents require useful access to tools and resources, while teams still need security, governance, reliability, and control.

This course explores practical engineering patterns for building platforms that work well for both humans and agents. From Golden Paths and policy as code to scoped permissions, observability, recovery, and human approval, you’ll learn how to turn AI from an experiment into a dependable part of modern software delivery.

From Platform Foundations to Agentic Workflows

Building for AI agents is not about replacing existing platform practices, but about extending them for a new type of user. Agents need clearly defined capabilities, controlled access, reliable infrastructure, and the ability to recover when things go wrong. This course connects these requirements with proven platform engineering practices, showing how APIs, infrastructure, policies, and developer experience can come together to create platforms that are ready for increasingly autonomous software.

Take Your Skills to The Next Level

Article | From Autonomous Agent Prototype to Production | Kylian Cros

Moving AI agents from prototypes into production introduces new operational challenges. This article explores recovery, persistent workspaces, state continuity, browser automation, agent identity, human handoff, isolation, and observability to help teams assess whether an agent is ready to operate as a long-running software workload.

Article | The Golden Path: Platform APIs, and Balancing Standardization with Flexibility | Shubham Thukral

Internal platforms should be treated as products that provide self-service infrastructure without losing control. The article explores platform thinking for application teams, balancing standardization with flexibility, and measuring platform success through developer outcomes.

Article | Is the MCP Debate Asking the Wrong Question? | Chris Carter

MCP adoption is moving quickly, but treating MCP versus APIs as a simple interface decision misses the bigger picture. This article examines agent-specific threat models, authentication, tool scoping, gateway patterns, and enterprise readiness, including what teams can do today as MCP infrastructure evolves.

Session | Designed for agents, useful for humans: Five IaC practices that pay off twice | Wladi Mitzel

AI adoption can increase both delivery throughput and instability. This session explores why platform quality is the differentiator and presents five infrastructure-as-code practices that benefit both agents and humans. Based on external research and Pulumi telemetry across thousands of organizations and nearly 10 million deployments, the session covers non-interactive deployments, policy enforcement, infrastructure access, and practical patterns that apply across Pulumi, Terraform, and OpenTofu.

Session | AI won’t fix your Platform but it can make it exceptional | Jean Burellier

Platform teams are under pressure to adopt AI, but experiments often fail to deliver meaningful results. This session shows how platform engineers can use AI to prototype capabilities, understand real developer behavior, and continuously evolve platforms as products. The focus is on practical ways to turn AI into a lever for speed, insight, and developer impact rather than another technology experiment.

Expert Knowledge for...

Platform engineers building internal platforms for humans and AI agents
Software architects designing secure and reliable agentic systems
DevOps and infrastructure teams working with IaC and autonomous workflows
Security engineers evaluating agent permissions, attack surfaces, and blast radius
Engineering leaders shaping platform strategy and AI adoption