Roadmap to make useful AI Agents — and levels of makers

Hello world 👋,

This article provides a path that a new graduate or seasonal programmer can use to get started with making useful AI Agents.

It is evident that AI Agent is going to be a multi-trillion dollar industry in the coming years. YC and other tech communities are betting heavily on it

As usual in learning, everyone will start at their own level of knowledge, experience (know-how), interests, etc. So, this roadmap is meant to be followed in a flexible manner that is suitable for your personal journey. Here, relative order is more important than sequence.

Basics

This section covers a ground that one should cover as much as possible before proceeding to the next section. If you are experienced with programming, you have probably already covered this section.

1. Basics of programming using some high-level language — Python, Ruby, etc. You just need to know programming concepts (hands-on) such as loops, conditions, functions, classes, imports, IOs (file and internet calls)
2. Basics of the Internet — how the Internet works (network layers), popular protocols (HTTP/S, TCP/UDP, WebSocket, Email, etc)
3. Basics of API — how typical API request and response works, RESTful APIs, error handling
4. Basics of storage — in-memory storage, SQLite, Relational DBs, Document DBs, cache & indexing.
5. Basics of packages — this depends on your language of choice. For Python, you will learn pip. For Ruby, you will learn gem.

Level G

This is the first level after you cover the basics. It is the hands-on stage where you pretty much try any library, framework, or tools to get the job done. Interestingly enough, at this level itself, you will be making AI Agents. These may not be the best ones to do complex tasks or scale as needed, but they will get the job done!

1. The first AI Agent using simple tools — use Phidata to implement a simple & elegant agent
2. Continue to simple multi-agent implementation
3. Try out other AI Agent Frameworks to create small agentic flows

At this level, as most of your code produced is for learning, I recommend sharing those on GitHub or GitLab as open-source projects. That will serve two important purposes:

• Learn Git and the ecosystem around it
• Start building portfolio

Level C

This is the first level that not only puts you in making real-life useful AI Agents but also to start earning from it potentially.

1. Find common flows that people are automating in some form and then build your agentic flow to do the same. To find, you can scout Reddit, Dev.to, and other forums on the Internet
2. Build with peers. This is an important learning step to start building with other fellow humans. By now you have made AI Agents and are (hopefully) using a few AI Agents to help you build. Extending your collaborators to humans will improve your skills as well.
3. Host your solutions. Launch your project so that non-programmers can also use it. You can host using digital-ocean, free AWS tier, Heroku, etc.
4. Do the freelance work for businesses. Connect with businesses (Reddit, LinkedIn, AIAgentsLive, etc) and offer them your skills. You can charge for your work as you see fit.
5. Join a startup to not only build AI Agents but also to learn many other things about making real-life useful products.

Level J

Welcome to the top level of the AI Agents maker. This level is without the top bar. Here you are capable of and have demonstrated the following:

1. You can build agentic workflows for any arbitrary needs
2. Your work is partly made by other AI Agents (that you developed or using existing ones)
3. You have made money by making AI Agents (could be as a freelancer or a startup engineer)
4. You can study, POC, and evaluate any new AI Agent tool that keeps on coming to market.
5. You have worked in a team of engineers and non-engineers.
6. You have discovered gaps in the market and are likely to start your own initiative/startup to address them.

We could subdivide Level J into three parts.

1. Level J Base — this essentially completes the aforementioned expectations
2. Level J-A — those we completed Level J Base and are already making progress beyond it. Establishing new milestones in the journey.
3. Level J-S — those who are at the leading edge of AI Agent research and development and have published papers to contribute to the field.

Thanks for reading folks. The contents of this article are just rough guidelines and of course, you can forge your own path. I hope it will be helpful for those who are looking for a roadmap 🥷.

Happy coding!