Stage 3 — Master AI, Agents & Automation
Go under the bonnet. Learn the principles behind modern AI, then build and evaluate connected tools, knowledge systems and agents for practical use at home and in business.
24-LESSON SYLLABUS PLANNED PREMIUM COURSE IN DEVELOPMENTProposed price: £15. Not available to purchase or enrol in yet.
Who is this for?
For technically curious learners, professionals and business builders. Stage 2 is not required. Some projects involve code, APIs and approved accounts, with guided options for different experience levels.
What you’ll be able to do
- Explain the modern AI stack beyond marketing slogans.
- Understand how models, data, retrieval, APIs and agents connect.
- Build a small technical prototype with appropriate tools and safeguards.
- Evaluate risks, cost, quality and repeatability before deployment.
- Design a practical, reviewable agent workflow for life or business.
Try a complete Stage 3 sample lesson
Understand an agent loop, run a safe Python example with fictional tasks and test three failure cases before designing a real integration.
Try the agent lessonA real result, not just a lecture
Every course is being designed around a result you can keep: a tested companion blueprint or a safe technical project with documented evidence and checks.
Strengthen your foundationsAgent Builder: 12 free engineering templates
Open twelve practical, reusable prompt templates covering the core skills of this course. Test them using fictional or approved data, keep the versions that work and improve them as you learn.
Explore all 12 templatesDesign a safe agent before you give it tools.
The Agent Workflow Studio creates a reusable architecture brief from your goal, data sources, permissions, approvals and failure-handling choices. It does not run or connect a live agent.
Open Agent Workflow StudioSix substantial modules. Twenty-four lessons.
This is the detailed learning plan, not a claim that all 24 lessons are already available. Full tutorials, practical projects, assessment materials and teaching examples are being developed.
01 · How AI Actually Works
- From training to inference: the AI system explained
- Tokens, context windows, transformers and attention
- Models, reasoning, limitations and evaluation
- Choosing between cloud, local and hybrid AI
Each lesson is planned to include an explanation, worked example, hands-on exercise and knowledge check.
02 · Data and Knowledge Systems
- Working with JSON, tables, documents and databases
- Embeddings, vectors and semantic search
- Build a retrieval-augmented knowledge assistant (RAG)
- Check quality, permissions, freshness and source grounding
Each lesson is planned to include an explanation, worked example, hands-on exercise and knowledge check.
03 · Programming, APIs and Tools
- Python foundations for practical AI building
- HTTP, APIs, authentication and secret handling
- Tool calling, structured outputs and validation
- Integration patterns, MCP concepts and error recovery
Each lesson is planned to include an explanation, worked example, hands-on exercise and knowledge check.
04 · Build and Coordinate AI Agents
- What distinguishes an agent from an ordinary chatbot
- Design a bounded, single-agent task with tools
- Coordinate specialists and manage handovers
- State, scheduled jobs, interruptions and recovery
Each lesson is planned to include an explanation, worked example, hands-on exercise and knowledge check.
05 · Deploy Agents in Life and Business
- Choose valuable tasks and calculate time/cost trade-offs
- Automate safe household or personal workflows
- Design a business process with real permissions and approvals
- Monitor results, evaluate failures and improve performance
Each lesson is planned to include an explanation, worked example, hands-on exercise and knowledge check.
06 · Security, Reliability and Your Final System
- Prompt injection, least privilege and sensitive information
- Human approval, audit logs, testing and observability
- Deployment, maintenance, change control and provider choice
- Build and present a working agent-assisted workflow
Each lesson is planned to include an explanation, worked example, hands-on exercise and knowledge check.
Make something you can use
A functioning, bounded AI-assisted workflow using permitted sample or business data, with documented architecture, tested tools, human-approval gates, evaluation results, security review and a maintenance plan.