English cohorts aren't open yet. These programs currently run live in Spanish for Latin America. Join the waitlist and you’ll be the first to know when an English cohort opens.
AI Agents for Developers: From Low-Code to Code-First
Build and deploy your first real agent, and learn when n8n is enough and when you need code with LangGraph.
Next cohort: join the waitlist
Key facts
- Format
- 100% live online
- sessions per week
- 2
- Each session
- 1.5 h
- sessions in total
- 8
- live hours
- 12 h
- Weeks
- 4
- Seats
- 25
Next cohort: join the waitlist
Objective
In 4 weeks, build an end-to-end agent: first low-code to learn the pattern, then in code with tools, memory and RAG, and finally with evals, security and deployment.
Who it's for
Developers with 1+ year of Python or JavaScript/TypeScript experience who have worked with APIs.
Not for
People without programming experience (see AI at Work).
What you leave with
- A low-code n8n agent connected to real APIs
- A code-first Python agent with LangGraph and LiteLLM
- RAG with a vector store over your own documents
- Evals, observability and prompt-injection protection
- A deployed project presented at demo day
What's included
- 8 live-coding sessions of 1.5 h
- Recordings for 6 months, with the code from each session
- GitHub starter repo: LangGraph, LiteLLM, RAG and deployment
- Written feedback on your repository
- Demo day: present your deployed agent
Session by session
Two evening sessions per week (7:00 pm, Latin America time) · Every session is recorded
-
01
Anatomy of an agent
- LLM, tools, memory and the loop
- Tool calling and structured outputs
- When low-code is enough
DELIVERABLE Setup and first tool call
-
02
First low-code agent in n8n
- Webhooks, APIs and the agent node
- Error handling
- Limits of low-code
DELIVERABLE Working n8n agent
-
03
Code-first with LangGraph
- Graphs, state, nodes and routing
- Retries and flow control
DELIVERABLE Base code-first agent
-
04
LiteLLM, tools and MCP
- Multiple models with LiteLLM
- Expose and use tools with MCP
DELIVERABLE Agent with 2+ tools
-
05
RAG with a vector store
- Supabase pgvector
- Chunking and retrieval quality
DELIVERABLE RAG over your documents
-
06
Memory and multi-agent
- Short- and long-term memory
- Supervisor pattern
DELIVERABLE Agent with memory
-
07
Evals, observability and security
- Evals with Langfuse or LangSmith
- Cost and latency
- Prompt injection and tool permissions
DELIVERABLE Eval suite
-
08
Deployment and demo day
- Containerized API and human-in-the-loop
- Project presentations
DELIVERABLE Deployed agent
FAQ
Are sessions live?
Yes. Every session is live online, with time for questions and work on your real cases.
What if I miss a session?
Every session is recorded, so you can watch it anytime during your access period.
When does the next cohort start?
Dates are announced first to the waitlist. Join and we'll let you know.
Can I enroll my team?
Yes. Team packages and private cohorts are available. Mention it in the form.
What do I need installed?
Python 3.11+, Docker, a GitHub account and an API key from a model provider. You'll get a setup guide before the first session.
About the instructor
Lorgio Trinidad
Founder & CEO of Braincoders and Orvium
20+ years in technology: engineering at Globant, three startups as CTO and, today, founder & CEO of Braincoders and Orvium from Dubai. I have led 70+ engineers and taken teams to ship 3× faster. What these programs teach is what I actually use — no borrowed theory.
Join the English waitlist
No dates yet. Cohorts are announced to the waitlist first — leave your email and you'll hear from us when the next one opens.