Thanks for subscribing to The Blueprint.

If you’re tired of AI hype everywhere and want to know what actually works, you’re in the right place. Every week, I’ll be cutting through the noise, testing the latest repos, and showing you the technical "why" behind the tools that actually help you get things done.

Three months ago, I was an AI skeptic.

Machine learning tools first started coming out when I was in college, way back in 2015. They slowly got better, but I still remember the days of psychedelic conch shell ridden AI generated images and nonsensical text generation. Plus, I was frustrated seeing AI trained on the work of creatives then taking away opportunities from the same people.

In 2023, I quit my career media job and spent the next three years building a CPG business. All the while, AI tools were exploding in advancement and practicality in a way they never had before.

Earlier this year, I realized I was out of the loop. So I hunkered down and decided to become an expert.

I still have a long way to go, but I’ve spent the last few months learning as much as I can about AI. I’ve built myself an evolving open source and free-to-use stack of tools that I’m using to build AI automation to eliminate grunt work.

Here, I’ll be writing my first impressions on tools as I find them, letting you know what is slop and not actually helpful to your workflow, and also sharing any forks or plugins I write that make integrating these tools into your setup easier.

First off, let’s break down the open source stack. This is exactly the set of tools I use every day to build AI automation.

This is the core of the setup. Opencode is an open source AI coding harness, like Claude Code.

Think of your AI model as a car’s engine - there’s a wide range in quality between engines, but they are the thing that actually makes the car ‘go’. (More on different engines later!) Opencode, then, is the rest of the car - chassis, suspension, windshield, etc. This provides a driving experience; otherwise, you’re just driving a go-kart.

Functionally, Opencode is an agent that is able to read and edit files for you out of the box. It can write plans for you, and connect to other software if needed. This logic and these tools are what I think most non-coders are missing out on. Using an AI model in chat is a pretty barebones experience relative to this!

The easiest way to get set up with Opencode is to download their desktop app. Make a new folder and add some relevant documents - your CV, project budgets, kanban boards, etc - and start asking it about your data. Opencode offers some free models through their Zen product so you can get started right away!

Next week, I’ll talk about the different engines you can use; what AI models to actually run your harness on. Zen is okay for starting out but there are definitely better setups out there!

Normally I’m going to be going a bit more in-depth when it comes to new repos on Github, but since we’re talking in-depth about the open source stack I’ll keep these short and just do a couple.

There are a LOT of memory plugins out there for agents, from simple markdown files and built-in tools to tools like this. I wrote myself a quick Opencode plugin for this one (which I will share later - gotta review it first) and it looked super promising. Basically, this ain’t your grandpa’s memory plugin - Akshay Pachaar has a great breakdown on LinkedIn about its multi-layer approach.

Grade: A ✅
Will break this one down more later but it looks like the real deal

I’ve been seeing projects like this pop up a lot - basically it visualizes your AI agents as pixel characters that work in an office. It’s…not for me. I think getting organized and using tools like Trello or Notion is a lot more valuable than draining system resources to watch little characters walk around on a screen 😅 But it’s cool if you’re into that sort of thing!

Grade: D ❌
Probably actually distracting for your workflow

This section will be a super quick look into what’s new in the AI world! Definitely going to be focusing on free and open source tools here as well.

That’s all for edition 1! Thanks for tuning in. As you can see, we’re getting pretty technical here, which is why I’m keeping this at weekly for now. I want to be able to do deep dives instead of just sending out a list of links. Thanks again and see you next week for more tools and the next part of my open source stack!

Don’t get me wrong - it can be super overwhelming thinking about all the AI tools that are out there, which ones are best to use, and what will actually help you automate or optimize your workflow. I’m always open to discuss software or help you build a custom system!

Feel free to reply to this email with any questions or particular work problems that you want to automate :)