If you want to confidently create software with AI, this 3-week live workshop will make you not just AI-ready, but...

Everyone in tech knows we are in a time of more change than ever before, faster than any of us have ever seen.

All the rules of how we build software are being torn down and rebuilt from scratch.

Of course, engineering fundamentals still matter...

But there's another, LLM-flavored layer on top, that is likely to be MORE important in the short, medium and also the long term.

The great news is that by getting ahead of the curve now, you are well positioned for what might be an absolute FLOOD of demand for AI-enabled technology professionals.

Hence, announcing the completely new:

Agentic Engineering Academy

This is a 3-week live workshop, focused on the emerging cutting edge of leveraging AI to write production-grade code.

It is extensively hands-on. You will be using AI to create software applications from day one.

And you’ll get several instructional sessions per week, which are taught live, so you can ask questions during each presentation to get real-time answers.

These sessions are recorded, for you to review at any time you like in the future.

And best of all:

You Do It Together.

With a diverse group of like-minded technology pros.

That’s why we have a dedicated private Slack channel for you to share what you discover, learn from each other, and celebrate as you all progress.

And the instructor (more on him below) will be active in the Slack channel throughout the entire workshop, helping you through the exercises between sessions, and adapting the live sessions based on your experience.

You have certainly used tools like ChatGPT...

And you may have also worked with coding agents like Claude Code, OpenAI Codex, Cursor, or OpenCode.

Or like many, maybe you have read about these tools, but not really used them yet.

To benefit everyone, this workshop starts with the basics of using coding agents, for those who are not familiar.

But after the first day, we quickly advance to...

The Next Level:

• How do you choose which model to use, while managing token costs? There’s so much hype, and the choices are confusing. How do you decide, without making expensive mistakes?
• What are the best strategies and work patterns for using coding agents in different situations?
• How do you use voice input to make your work easier and faster?
• How does it all interact with version control? What are the best practices and different approaches, and their pros and cons?
• You may have heard of “plan mode”. How do you use plan mode vs. build mode or other modes, in practice?
• There are many “stunt” demos of making a program from scratch. But most software development involves working with large, mature codebases. How do you successfully use coding agents there?
• Or put another way: how do you use coding agents to add rich features to complex codebases, while maintaining code quality, maintainability, robustness, and so on?
• How do you control long-running agents? How do you keep one running when you need to turn off your laptop, for example?
• Do you need to read every line of code produced by the coding agent? Because it’s a lot! How do you approach that? Or are there situations where it is safe to not read the generated code at all?
• How do you manage “agent memory”? What does that even mean? Is it just an analogy, or something more?
• What are “skills” about? There plenty of breathless Youtube videos bleating about them, but what are they actually useful for? And where are the “traps” that you want to avoid?

One question you might have: “Which coding agent are we using in this workshop?”

Here’s a big key:

The Coding Agent Doesn’t Matter.

Mainstream coding agents (or harnesses, as some people call them) are largely equivalent in features at this point.

That includes Cursor, Claude Code, Codex, OpenCode, or anything else.

Yes, you may prefer one or the other. Nothing wrong with that.

But every time one coding agent adds a useful new feature, there is nothing to prevent all the others from quickly adding it, too.

We have seen this happen many times already (plan mode, agent skills, etc.)

For this reason, no coding agent is dramatically different in capability than any other. Which is great, because it lets us focus on higher-level strategies for using them.

That all said: in this workshop...

We Use OpenCode.

For several reasons:

• It is the fastest-growing opensource coding agent, by a massive wide margin, with growing momentum. So it is likely to be relevant and widely used for a long time.
• It includes FREE access to surprisingly good models, like Deepseek V4 Flash Free. These are so good, you will be able to fully participate in the entire workshop without paying for ANY model tokens or subscriptions at all.
• It fully supports the important features of other popular coding agents - like “plan” mode, agent skills, memory files, and more.
• It is truly model neutral, unlike other coding agents made by model providers.
• OpenCode is quite similar to tools like Claude Code and OpenAI Codex, so your skills easily transfer to those.
• It is fully cross-platform, working on MacOS, Windows and Linux.
• It has both a text UI interface, and also a well-made desktop app, giving you flexibility and choice in the ways you can use it.
• You can run it as a non-interactive command line program... which opens up many exciting automation possibilities.

Agentic Engineering Academy starts on Tuesday, August 11th, running through Monday, August 31st...

With some fun “pre-workshop” activities for those who enroll early.

Workshop Schedule

Week 1:
Coding Agent Foundations

• Fast but thorough intro to working with OpenCode, and how to use its most important features and tools
• Voice control of agents
• Agent modes and session management
• Context wrangling: what context (and context windows) really are, and how to manage them
• Understand tokens, models and providers, and how to pick which LLM to use
• Work strategies for coding agents, from “vibe coding” to methodical engineering processes; and when to use each in different situations
• Working with agent memory files
• When you can safely NOT look at the code; when it is necessary to review every line; and how to decide safely what you should do
• Developing a nutrition-data web application (several times, with different agent strategies, to identify the pros and cons of each)

Week 2:
More Advanced Agentic Strategies

• The idea of Spec-Driven Development... Its weaknesses and its strengths... and how the rules for SDD have changed with agentic coding AI
• Agent skills: how to use them, their traps and limitations, and how to win the skill game
• How to avoid harmful agent skills (because there’s a lot of them on the internet)
• Working with subagents
• Using tmux for persistent session management
• Creating a private VPS to let agents run continuously
• Sandboxing agents inside of VMs
• Dimensions of evaluation of the quality of agentically generated code

Week 3:
Agentic Development On Mature Codebases

• Extending the test suite on a system monitoring tool
• Defining new complex validation types for a command-line interface library
• Adding a challenging but in-demand feature to a popular and widely-used webapp framework
• And more. In all, over 30 projects that you can start during the week, and continue after - adding REAL features to REAL codebases.

Plus extra topics based on the interests of workshop participants, and as time allows.

What You Get

• Full attendance to the entire 3-week workshop, including:
• Attending all live-taught sessions, where you can get your questions answered in real time, as the material is presented
• Full recordings of all sessions, for you to review any time you like, or in case you cannot attend a session live
• All hands-on exercises from every session
• The private student Slack channel, where you can talk with other students and ask the instructor questions between sessions

Please note this workshop is limited to 30 attendees, so that there is plenty of time for individual attention with the instructor.

The Investment

To my knowledge, there is NO workshop or training available which is comparable to the depth, altitude, and hands-on guided experience of Agentic Engineer Academy.

Not One.

There are courses which cover parts of its curriculum. For example, I recently saw someone offer a weekend workshop on Claude Code, with 2 live sessions over the weekend, at a cost of $497.

While I believe it was a good workshop, its short timeframe means there just isn’t enough time to develop in-depth knowledge and broad skills.

Plus, it focused on a single proprietary tool. Almost an advertisement for that vendor’s product, arguably filling in gaps in its user documentation.

Mastering the coding agent’s features is of course an important first step, which is why we cover it for OpenCode in the first session.

But I think you can see how Agentic Engineering Academy...

Goes Far Beyond The Scope

Of any specific vendor, harness, coding agent, provider or model.

And it goes beyond teaching you how to go through the motions, performing some memorized menu of tasks...

To understanding the key principles behind the rapidly-changing best practices.

So you can jump ahead of that curve, instead of breathlessly trying to keep up.

On top of all this: did you notice there is NO hands-on training available for many of the topics you learn in Agentic Engineering Academy?

Like Voice Control Of Coding Agents...

Or sandboxing your agent inside a virtual machine...

Or setting up a cheap private server to allow your agent to run continuously and persistently, even when your laptop is turned off, and how to manage that successfully.

Or Specific And Clear Strategies For:

• Context management,
• Avoiding harmful agent skills,
• Objective criteria for evaluating generated code quality,
• And how to pick the right model to work with...
• (Not just today, but in the future as new models come out...)

Plus tackling the big question facing the entire software development profession right now:

• How do we handle the bottleneck of manually reviewing all this generated code?
• Are there situations where it is safe to NOT read every line?
• And if so, how do we know?

In short:

Do We REALLY Need To Read All The Code?

And another thing that frankly annoys me about every single training on agentic coding I have seen so far:

They ALL, without exception to my knowledge, focus on building applications from scratch...

Which Is The EASIEST Situation You Will Ever Face In Software Development!

In reality, nearly all the coding you are paid to do in your job is modifying existing software applications, often large codebases with long histories.

That is why Agentic Engineering Academy invests a whole week on how to use coding agents with mature codebases...

Adding realistically rich (not simple) features, substantial refactorings to improve code organization, covering giant gaps in the test suite, and more. In short:

Actual Software Engineering!

Seriously: Are ANY of these topics covered in ANY other training right now?

I am sure some of them are, here and there...

But not in a complete and cohesive package, like you learn in three weeks here.

Given all this... what is this training worth?

At a minimum, I would say:

At least $10,000 USD.

Especially since what you learn is increasingly in demand by employers.

Just think about what you’ll be able to demonstrate in interviews, that no competing applicant will be able to show.

And also because of the long-term focus of this workshop, teaching you key principles that will be relevant for the foreseeable future.

(In contrast to a quick workshop that teaches you how to use a single tool as it exists today, which may well be obsolete in just a few months.)

Again, I am limiting this to just 30 attendees.

I don’t want more than that, so I can give enough individual attention to everyone.

(I heard that other workshop had over 200 signups. I’m happy for their bank account, but there’s no way everyone got all the attention they needed from the instructor over that weekend.)

So I certainly think charging $10k is justified. Maybe even $20k, all considered.

Nevertheless, I have decided to offer this at the probably too-low price of just $897 USD.

I could say more. But if you have read everything up to this point, then you already know if this is for you.

Sign Up Now

Questions And Answers

If you have a question not answered here, send an email to service@powerfulpython.com and we will respond promptly.

Who is this for?

For technology professionals who are already writing code, and are ready to embrace the new age of AI and agentic engineering.

If you are brand new to coding, this is not for you, and I suggest you look elsewhere.

When does the workshop start?

It officially starts on Tuesday, August 11th, and concluding on Monday, August 31st. But as soon as you enroll, you get access to the private Slack channel and the introductory course material, so you can start learning right away.

I recommend you sign up ASAP, so that you can start preparing and get the most from the full workshop.

How much will I have to spend on tokens? Or will I need a subscription for models?

You will have to spend ZERO! One reason I chose OpenCode is that it integrates free access to surprisingly good models - yes, at no cost. You can do the entire workshop with just those models.

Optionally, you can use a paid subscription plan, which will let you work faster and do better software development than the free models allow. You can upgrade by using OpenCode’s “Go” provider subscription (for $10/month); or if you have an OpenAI subscription, you can use that. (Anthropic currently blocks use of their subscriptions with agents other than Claude Code, unfortunately.)

You can also pay for tokens directly. However, for safety reasons, I do not recommend this during the workshop. Because if you purchase blocks of tokens, it is possible to accidentally spend more money than you intended. The subscriptions all have built-in limits that make this impossible; you can be confident you won’t spend more per month than the subscription fee.

When you are using what you learn on the job, you can have your employer buy those tokens. That way they take on the financial risk, and not you! And you will not run into the limits of the subscription plans.

What programming language is used?

We’re using Python for the examples in this workshop. But that’s a minor detail. Nearly 100% of what we are doing is completely independent of programming language.

How much time does this take?

I recommend you set aside at least 7 hours per week. You will benefit if you can invest more time, up to about 12-15 hours per week during the 3 weeks of the workshop.

What operating system do I need?

This workshop supports all major operation systems: macOS, Windows and Linux. But if you are using Windows, see the next question.

I’m using Windows. Anything I need to know?

For the best experience, I strongly recommend you use Windows Subsystem for Linux (WSL2).

This is not strictly necessary; you can do most of the workshop just working in the main Windows context. But most coding agents (not just OpenCode) are built to work in “unixy” environments like macOS and Linux. Otherwise you have more friction, janky behavior, broken features, and so on. It’s similar to the situation with Docker, in many ways.

I’ll provide detailed instructions for how to set this up and work with it, if you are not already familiar with WSL2. This workshop is the perfect opportunity for you to learn!

As soon as you enroll, the material for WSL2 will be available to you. If you are new to WSL2, I recommend you sign up early, so you can be fully ramped up by the time the workshop starts.

I prefer a different coding agent instead of OpenCode. Can I use that instead?

Here’s what I recommend. Every activity in the workshop that uses a coding agent, do it first using OpenCode, since you can follow along with exactly how I show you.

Then, do the same activity AGAIN, using your preferred coding agent. It will go much quicker the second time, and you’ll be broadening your experience with different tools.

Alternatively, you can skip OpenCode and just do every exercise in your preferred agent. There is a good chance this will work well; after the first day, nothing we do in this workshop is unique to OpenCode. But for logistical reasons, I won’t be able to support every different coding agent during the live class; we are just covering too much material. So you may be on your own if you encounter issues specific to that coding agent.

When are the live sessions?

The exact schedule is still being set, but there will be 2-3 live sessions each week, during the morning and/or afternoon in North American time zones. Recordings will be posted ASAP, typically within 2-3 hours. Since everyone has different schedules, I’m expecting not everyone will be able to attend live, and that will work out fine.

How many people will be in the workshop?

It is limited to 30 spaces, so that there is plenty of time for individual attention with the instructor.

I have another question.

Email us at service@powerfulpython.com and we will reply as quickly as possible.

Sign Up Now

About The Instructor

Aaron Maxwell is a software engineer, researcher and author. After working in the early and founding engineering teams of several Silicon Valley startups - including two Unicorns, SnapLogic and Lyft - he fell in love with teaching, partnering with O’Reilly Media and running his own independent trainings for over 10,000 technology professionals worldwide... in just about every country, engineering domain and culture.

He has approached AI-assisted software development as a skeptical hard engineer, insisting on cutting through the hype to find what really works. He has hand-written several books (including Powerful Python, published by O’Reilly), and never uses AI when writing for humans, ever.