// PROGRAMME_02
Build AI-Native Products. Not AI Gimmicks.
RAG, Graph RAG, knowledge graphs, agentic workflows, memory layers, MCP, A2A. Everyone's adding AI to their products. Most of it is garbage. Learn to build AI systems that actually work.
// THE_PROBLEM
AI Is Everywhere. Good AI Is Rare.
Every startup claims to be "AI-powered" now. But most AI features are just ChatGPT wrappers that hallucinate, cost a fortune, and break constantly.
- [-]Hallucinations destroy user trust
- [-]Costs spiral out of control at scale
- [-]No idea how to evaluate quality
- [-]Agent loops that never terminate
- [-]Security vulnerabilities from prompt injection
// SOLUTION
Building AI That Works
The difference between good AI and garbage AI is architecture. Understanding when to use what technique—and when AI isn't the answer.
- [+]RAG grounds answers in your data
- [+]Agents decompose complex tasks
- [+]Evaluation catches failures before users do
- [+]Guardrails prevent catastrophic outputs
// OUTCOMES
What You'll Build
Not toy demos. Production-ready AI systems with real architecture.
Design AI System Architecture
Know when to use RAG vs fine-tuning, agents vs chains, and how to structure AI-native products.
Build with LLM APIs
Hands-on experience with OpenAI, Anthropic, and open-source models. Understand the trade-offs.
Implement RAG, Graph RAG & Agentic Workflows
Build RAG and Graph RAG systems with knowledge graphs, memory layers, MCP, and A2A protocols.
Deploy AI to Production
Ship real AI features with proper evaluation, monitoring, and safety measures.
// TECH_STACK
The Tools You'll Master
Hands-on experience with the entire modern AI stack.
OpenAI GPT-4
Claude API
Llama/Mistral
LangChain
Pinecone
pgvector
Vercel AI SDK
LangSmith
// CURRICULUM
Two Weeks. From Zero to Production.
Theory in the morning, building in the afternoon. Every day.
AI Fundamentals & LLM APIs
- [+]LLM architecture: How transformers actually work
- [+]API design: OpenAI, Anthropic, and open-source models
- [+]Prompt engineering at scale
- [+]Embeddings and vector databases
- [+]Cost optimization and rate limiting
Production AI Systems
- [+]RAG & Graph RAG: Knowledge graphs, memory layers, embeddings
- [+]Agentic workflows: MCP, A2A, tool use, planning
- [+]Evaluation and testing AI systems
- [+]Safety, guardrails, and failure modes
- [+]Final project: Deploy a production AI feature
Final Project: Ship to Production
You'll build and deploy a real AI feature during the programme. Choose from: RAG-powered search, AI agent workflow, or custom LLM application. Real users. Real feedback. Real learning.
- [+]CTOs adding AI to existing products
- [+]Technical founders building AI-native startups
- [+]Engineers transitioning to AI/ML roles
- [+]Anyone building products with LLMs
- [+]Leaders who need to evaluate AI architecture decisions
- Comfortable writing code (any language)
- Basic understanding of APIs and databases
- Familiarity with Python helpful but not required
- Completed Code Again or equivalent experience
No prior AI/ML experience required. We build up from fundamentals.
AI Systems
2 weeks intensive. Remote cohort. Maximum 12 participants.
- [+]10 days of live instruction
- [+]Hands-on building every day
- [+]Access to all AI APIs during programme
- [+]All recordings and materials
- [+]Private Slack community
- [+]30-day post-programme support
In-person London option: +£2,500
// BUNDLE
Start From the Beginning?
Combine Code Again with AI Systems for the complete technical transformation.
Technical Track Bundle
Code Again + AI Systems. Learn to code with AI, then learn to build AI systems.
Build AI That Actually Works
Next cohort starting soon. Limited to 12 participants. Graduates eligible for paid network deployments.