Open for registration
AI Agents Foundations
Build real AI agents in two weeks, with no coding background needed. You bring the judgement and direct the build while an AI assistant writes the code, and you finish with two working agents: one built live with the group, and one that solves a problem from your own work.
Ready to register?
Registration is open. Share your details and Ratio Africa will send you to secure checkout.
Register for this cohortThe outcome
What you'll walk away with.
Two working AI agents you built and can run yourself: one created live with the group, and one that solves a real problem from your own work, each in your own GitHub repo. You also leave with the judgement to know when building beats buying, and a repeatable method for designing, improving, and checking any agent you build after the cohort.
What you'll learn
Syllabus.
01
Technical Setup
Getting your tools, accounts, and environment ready to build You will get your full build environment ready with step-by-step guidance, no technical background assumed. The setup guide reaches you before the cohort begins and our team will be available to answer any questions you may have, so you spend the two weeks building instead of troubleshooting.
02
The Build-or-Buy Decision
Knowing when a ready-made tool is enough and when to build your own You will learn to tell when a ready-made tool is enough and when building your own is worth it. This single judgement decides whether your time and money go to good use, and everything else in the cohort builds on it.
03
Prompt Engineering
Writing the instructions that make AI reliable You will learn to write prompts that produce sharp, honest, useful output instead of vague filler. It is the highest-return skill you can carry into any AI tool, and it keeps paying off long after the cohort ends.
04
Agent Design
Turning one step into a working multi-step agent pipeline. You will learn how a real agent is put together, and how to direct one from a rough idea into something that runs and does useful work on its own. This is where directing a build stops feeling like theory.
05
Context Engineering
Shaping what the model pays attention to for sharper results. You will learn the technique that does more for output quality than anything else, and why most people never realise it is the lever they are missing. It is the highest-leverage skill in the cohort.
06
Agent Evaluation
Measuring quality and proving a real improvement. You will learn how to move from believing your agent is good to proving it, with a repeatable way to measure quality and show a real improvement. It is the habit that separates a builder from someone guessing.
When the cohort runs
Schedule.
Six live sessions of ninety minutes, three evenings a week across two weeks. Tuesdays, Thursdays and Saturdays, 6:00 to 7:30 PM WAT. Every session is recorded and shared afterwards, so a missed session is easy to catch up on.
Completion
What completion looks like.
To complete the cohort and earn your certificate, you take part in at least five of the six live sessions (a recording counts toward one), build and run your own agent through the weekly assignments, and present it on demo day. Cannot make demo day live? A short recorded demo counts instead. Completion here means something concrete: you walk away with a working agent you built, not just notes.
Before you start
Before you join.
No coding background needed. You will need a laptop, a stable internet connection, and a few hours between sessions for the hands-on work. A short setup guide is sent before the cohort begins, and our team is available to support you, so you arrive ready to build from the first session. The cohort uses an AI coding assistant during the build (Claude Pro, ChatGPT Plus, or free Antigravity Models), with free tools for everything else. Curiosity and a real problem you would like to solve are the only other requirements.