L1Free with sign-in
017LESSONS
Awaken
Turn Pi into your own Agent.
Start from Pi, run your own Agent, connect a messaging channel, and complete the first real build.
2 hands-on + 5 concept
- Pi → your Agent
- Telegram chat
- AI-first collaboration
✳AI AGENT SCHOOL BY DARKS
Build an agent
you will actually use every day.
Sign in to build your first Agent for free in L1. Expand its memory, tools and practical capabilities through L2–L4, learning to develop independently with AI.
Read L0 concepts without signing in; sign in to build your first Agent for free in L1.
- Start free in L1
- Upgrade from Self-study by paying the then-current plan difference
- Share what you build
WHY BUILD YOUR OWN?
A mainstream Agent is designed for everyone.Your own Agent is designed around your needs.
I started with AI automation, yet spent over a year wondering how to build an Agent. Once I found a workable approach, it was far less mysterious than I had imagined.
I have now spent more than a thousand hours exploring and building an Agent I use every day. If you have been stuck too, I have turned that experience into four stages you can follow.
THE FOUR-STAGE JOURNEY
The whole course, at a glance.
Start with one Agent of your own, learn by building, and grow with it. You do not need to understand all 39 lessons up front — first see where each stage takes you.
Free L0 concepts + the four-stage L1–L4 practical path. L0 foundations · Public access · 7 concept articles.
L2Paid
0211LESSONS
Mind
An Agent that remembers and grows
Memory, personality, a web console, Skills and multiple Agents. Turn your first Agent into a system you can maintain.
11 hands-on
- Long-term memory
- Proactive heartbeat
- Web dashboard
L3Paid
0310LESSONS
Enable
An Agent that gets work done
Delegation, RAG, research collection, images and browser automation. Break down tasks, connect tools and verify real results.
8 hands-on + 2 concept
- Tasks & schedules
- Local RAG
- Browser automation
L4Paid
0411LESSONS
Govern
An Agent in your life, with you in control
Life tools, finance, voice, MCP and safety governance — extend your Agent while learning to develop independently. Practice developing, maintaining and extending your own Agent.
10 hands-on + 1 concept
- Finance & calendar
- API / MCP
- Growth governance
Want every lesson title, publication status, and concrete outcome? Open the full course map.
View the full course mapFROM SELF-TAUGHT LEARNER TO INSTRUCTOR & CONSULTANT
I have been the beginner. And helped thousands take their next step.
- hours exploring Agent development
- 1,000+
- AI automation learners
- 2,000+
I began teaching myself AI automation in 2025, then became an instructor and consultant, helping more than 2,000 learners study AI automation. I know the gap between understanding an idea and making it work.
This site brings together that learning, development and teaching experience: demonstrations, project plans, hands-on practice and verification. Beginners and experienced builders alike can follow a path towards their own Agent.
Explore my earlier n8n learning site ↗REAL NEEDS FROM 200+ PEOPLE
Why do people want to build their own Agent?
In a development-motivation survey of more than 200 people, these three needs came first.
- 0171.3%
Everyday work efficiency
More time for the work that matters
Go beyond another chat window. Let an Agent help organize information and track routine work, so there is less for you to carry.
- 0268.4%
A privacy-focused assistant / second brain
A personal assistant, with clearer data boundaries
Organize your information and methods, understand where data goes, and decide what your Agent may read and do.
- 0361.2%
Practical project-development skills
The ability to build, not just use tools
Practice defining requirements, briefing AI, testing and fixing — and turn your own ideas into usable projects.
Source: AI Agent needs survey, development motivations (multiple choices). Percentages show response shares, not learning outcomes.
NO ADVANCED AI TOOLCHAIN REQUIRED
A fresh Codex can follow the path — no Skills, AGENT.md, MCP, or extra AI tooling required.
When I record the hands-on lessons and validate each development plan, I intentionally use a fresh Codex without preloaded capabilities. The plans are written against that baseline so people who are new to AI development can start from zero.
Already have MCP servers, Skills, AGENT.md, or a toolchain you like? Keep using them. They can make you faster, but they are not prerequisites for this course.
BLANK CODEX / BASELINE
Validated with fresh CodexNo required MCPNo required SkillsNo required AGENT.md
HOW YOU LEARN
Don’t just watch it. Build it yourself.
I build the path and walk it once with you. Learn at your own pace, build only what you need, and understand the architecture as it becomes useful.
Hands-on · 31 lessons
Every hands-on lesson includes a video. When a project can be specified, you also get the project plan I already tested so your AI can build with you instead of rediscovering every dead end.
- 01Watch the build
- 02Get the project plan
- 03Build with your AI
- 04Operate and verify it yourself
Concepts · 8 lessons
Concept lessons use video or written guides depending on the topic. The goal is not memorization; it is understanding the architecture exactly when you need it.
- 01Video or guide
- 02Plain-language explanation
- 03Real use case
- 04Back to building
WHAT YOU CAN BUILD TOWARDS
More than a finished course. An Agent system that fits your life.
A team of AI collaborators with different strengths
Give different roles different tasks, let them discuss and share relevant information, instead of asking one assistant to handle everything.
A second brain that helps you pick up the thread
Organize notes, documents and working methods with indexes and retrieval, so your Agent can find useful context when you need it.
Work that can continue after a clear brief
Set up steps, tools and checkpoints. Let your Agent act within the permissions you choose, while keeping consequential decisions in your hands.
A long-term partner you do not need to reintroduce yourself to
Carry preferences, habits and important context into future conversations. Shape memory and personality around the way you work.
MAKE THE PATH YOUR OWN
Close your eyes for a moment.
Think about using an Agent today. What frustrates you? Which steps feel awkward? Is there a feature you wish you could change without waiting for someone else?
Now think about everyday work you would like help with: organizing incoming information, remembering repeated preferences, tracking tasks, or taking a draft further.
Pick one small thing you would love to change. This time, try building it yourself.
Try L1-01 for free →ONE PAYMENT · COMPLETE COURSE
Build your Agent with the complete L2–L4 course.
Both paid plans unlock the complete course, with future L0–L4 project-plan revisions and content updates included.
YOUR OWN PACE
Self-study
$219 USD
One-time paymentAll L2–L4 lessons, videos and project plans, with future L0–L4 project-plan revisions and content updates included.
Start with Self-studyLEARN TOGETHER
Community
$299 USD
One-time paymentEverything in Self-study, plus the dedicated Discord learning space and the AI Agent Real-World Demand Survey Report.
Choose Community143 paid learners have joined AI Agent School. At list price, one payment unlocks every L2–L4 lesson; future L0–L4 project-plan revisions and content updates are included.
YOUR PATH TO INDEPENDENCE
Build an Agent — then learn to develop independently with AI.
Learn to define requirements, brief AI, test, fix and extend your projects. Take away a repeatable method, not just a copied example.
L2
An Agent that remembers and grows
Memory, personality, a web console, Skills and multiple Agents.
Turn your first Agent into a system you can maintain.
See this stageL3
An Agent that gets work done
Delegation, RAG, research collection, images and browser automation.
Break down tasks, connect tools and verify real results.
See this stageL4
An Agent in your life, with you in control
Life tools, finance, voice, MCP and safety governance — extend your Agent while learning to develop independently.
Practice developing, maintaining and extending your own Agent.
See this stageBoth paid plans unlock every level. Learning outcomes are goals; individual progress varies.
PLEASE CONFIRM BEFORE PURCHASE
Try L0 and L1 first. Buy only after you are sure the course fits you. Digital access is generally non-refundable after purchase.
If you are not sure yet, use the free L0 and L1 material first and decide later.
YOUR GUIDE
I turned the path I figured out myself into a course you can follow.
I started from planning and operations, not software engineering. I learned AI automation by building from scratch, grew an audience of 23,000+, helped more than 1,000 entrepreneurs apply AI automation, and spent 1,000+ hours this year building an Agent I use every day. This course is the path I wish I had at the beginning.
MOVE AT YOUR OWN PACE
Build your own Agent for free first. Then decide whether to keep expanding.
L1 lets you finish the first stage for free and build a real Agent. Continue into L2–L4 only if you want the architecture and capability expansions that take you toward independent development.
FAQ
Before you start
Can I start without knowing how to code?
Yes. Stage 1 is designed to get a complete beginner to a first success. You still operate the steps, show errors to AI, and verify results yourself.
What is the difference between hands-on and concept lessons?
All 31 hands-on lessons include operational video and aim at a concrete capability. The 8 concept lessons use video or written guides depending on the topic, then send you back into the next build.
What is free and what is paid?
L1 is free and gets you through the first stage with your own Agent. L2–L4 are paid architecture and capability expansions designed to teach you by building until you can continue independently.
What does a purchase include? Will updates cost extra?
Both paid plans give you access to this product line from L0 through L4. If new material or updates are added to the three paid stages, L2, L3 and L4, they are included at no extra charge. Further additions beyond the announced curriculum are being considered, but are not guaranteed and have no promised frequency or delivery date.
Is this just learning another tool?
No. The goal is to walk the whole path once, from the first step to the last, so the next time you know how to walk it yourself.
Can I learn this just by watching the lessons and videos?
It is difficult. The best result comes from building it yourself once, which is why the course is designed around hands-on work rather than passive watching.
If I can ask AI anything, why buy this?
This is not a single lesson. It is the path I already worked through and organized so you can follow it. You could ask AI to rediscover the path and teach you afterward, but the repeated trial-and-error and high-end model token cost may end up costing more than the course.
Do I need specific MCP servers, Skills, or AGENT.md before I can follow the course?
No. My hands-on lessons and development plans are validated against a fresh Codex with no preinstalled Skills, no AGENT.md, no MCP, and no private AI toolchain. I first make sure that this blank baseline can complete the plan before it becomes course material. That means beginners without an existing AI development setup can start directly. If you already prefer your own MCP servers, Skills, AGENT.md, or other development tools, you are free to use them — they are accelerators, not course requirements. Individual lessons may still require the basic runtime, service, or account needed for the feature being built.
Can I get a refund after purchasing?
L0 is public without sign-in, and L1 lets you complete the first stage for free after signing in. Please watch the lessons, try the workflow yourself, and review the L2–L4 roadmap before you buy. If you are not sure this teaching style or roadmap is right for you, please do not purchase yet. Paid access is a digital product delivered immediately after purchase and cannot be taken back like a physical item, so general refunds are not offered.