Practice agentic AI patterns. Like LeetCode, but for AI agents.
Free interactive games and two learning tracks. One for developers, one for product managers. No sign up to start.
> I'm a
Preview the curriculum
3 of 21 developer patterns shown. Sign up free to unlock the rest.
Prompt Chaining
≈ Pipe & Filter
The foundational design pattern where a complex task is decomposed into a linear sequence of smaller, discrete LLM calls. The output of one step becomes the input (context) for the next step.
Routing
≈ Load Balancer / API Gateway
Dynamically directing a user request to the most appropriate specialized agent, model, or processing path based on the semantic intent and complexity of the query. A "Router" classifies the input and delegates execution to a downstream handler optimized for that specific task.
Parallelization
≈ Scatter-Gather / MapReduce
Executing multiple independent agentic tasks simultaneously — such as voting on a decision, generating multiple creative drafts, or verifying facts against different sources — and then aggregating the results. Also known as the "Sectioning" or "Voting" pattern.
> What people are saying
Antonio Gullí
Sr Director, Distinguished Engineer, CTO at Google
Author of “Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems.” The 21 patterns this curriculum is built on.
Two Tracks. Pick Yours.
For Developers
We teach you:
The 21 agentic design patterns, mapped to SWE concepts you already know. Code examples, architecture breakdowns, production notes.
For Product Managers
We give you:
Decision frameworks for agentic AI. Zero code. The tradeoffs, questions, and vocabulary you need to lead AI product decisions.
Not sure where to start?
Take the 3 minute assessment. Find out if AI will replace you, and what to learn first.
Will AI Replace Me? Free AssessmentCommon Questions
Agentic AI refers to AI systems that autonomously perceive, reason, plan, and act to achieve goals. Unlike chatbots that respond to single prompts, agentic systems use LLMs as reasoning engines, access external tools, maintain memory, and execute multi-step workflows. There are 21 established design patterns for building these systems, each mapping to a classical software engineering concept. Learn Agentic Patterns (learnagenticpatterns.com) covers all 21 with code examples and interactive exercises.
Start with the 21 agentic design patterns. They map to concepts you already know. Prompt Chaining is Pipe & Filter. Reflection is TDD. Multi-Agent is Microservices. Tool Use is the Adapter Pattern. Learn the architecture first, then implement in any framework (LangChain, LangGraph, CrewAI, AutoGen). Building agents is software architecture, not prompt engineering. Learn Agentic Patterns (learnagenticpatterns.com) teaches all 21 patterns with SWE mappings, code examples, and interactive building exercises.
Software engineering is evolving, not dying. Senior developers already have 80% of the foundation: distributed systems, design patterns, production software. The gap is framing, not skill. Every agentic pattern has a SWE parallel. Learn the 21 agentic design patterns and you transition from building traditional systems to architecting intelligent autonomous systems.
No. The Developer Track is built for senior developers comfortable with distributed systems, APIs, and production software. The Product Manager Track is built for PMs who own or influence AI product decisions. Both tracks start from your existing knowledge. We don't teach coding basics or product management basics.
Yes. The PM Track has 15 decision-focused modules (zero code required) that reframe the 21 engineering patterns through a product lens. You'll learn tradeoff frameworks (cost vs. quality vs. latency), key product decisions for each pattern, questions to ask your engineering team, and practice with two interactive games: Ship or Skip (pick the right architecture for a scenario) and Budget Builder (allocate token budgets across model tiers).
No. The Product Manager track is entirely code-free. It explains what each agentic pattern does, why it matters for your product, what tradeoffs it introduces, and what questions you should be asking your engineering team. The interactive games test product judgment, not coding skill.
No. Patterns are framework-agnostic. We use pseudocode and real examples from LangChain, LangGraph, CrewAI, and AutoGen to illustrate, but the concepts apply universally. The PM track doesn't involve any framework at all.
Yes. Both the Developer and Product Manager tracks are completely free. 7 developer patterns are open without sign up. Create a free account (no credit card) to unlock all 21 developer patterns, all 15 PM modules, and all interactive games.
LangChain teaches you how to use LangChain. Anthropic teaches you how to use Claude. DeepLearning.AI teaches AI fundamentals. This curriculum teaches the architecture layer between them. The 21 design patterns that determine which approach to use and why. It's framework-agnostic. Once you understand why Prompt Chaining solves different problems than Routing or Parallelization, you can implement in any framework. Plus, 21 interactive exercises let you build and simulate agent architectures hands-on.
Antonio Gullí is an Engineering Leader at Google and author of 'Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems.' This curriculum is inspired by and builds upon his 21-pattern framework.