Repository navigation
Tutorial Fundamentals
The four phases, the one hard gate, the eight loops and the evidence that crosses each hand-off: the model every other track builds on.
9 lessons · about 67 minutes · The lifecycle, phase by phase
Every lesson opens on the site, where the pictures are live and the text is searchable: start this track ↗, or pick a lesson.
| # | Lesson | What it covers | Time |
|---|---|---|---|
| 1 | Evolution of the PDLC: From Waterfall to Agentic Delivery | How the product lifecycle evolved from waterfall to agile, DevOps and agentic delivery, and what genuinely changes when a model does part of the work. | 8 min |
| 2 | Why Agentic AI Projects Fail: 7 Failure Modes and Fixes | Gartner expects over 40% of agentic AI projects to be cancelled by 2027. The seven failure modes behind that, why normal checks miss them, and the fix for each. | 7 min |
| 3 | P0 Frame: How to Decide If an AI Agent Is Worth Building | P0 Frame is the agentic PDLC's first phase: turn a request into a measured pain, test whether it needs AI at all, size the value and set autonomy per action. | 8 min |
| 4 | P1 Design & Spec: Write a Spec an AI Agent Can Build From | P1 Design & Spec turns a framed problem into an eight-field spec with EARS acceptance, a bar per slice and an authority budget an AI agent can build from. | 8 min |
| 5 | The Hard Gate: The Hand-off You Can't Skip in AI Delivery | Only one hand-off in the agentic PDLC halts the build. Which decisions are hard, which run behind a placeholder, and the four questions that tell them apart. | 7 min |
| 6 | P2 Build & Prove: How to Build an AI Agent in Proven Slices | P2 Build & Prove builds an AI agent in bolts of hours or days, gates every merge on a per-slice bar and ends with a shadow run beside the people doing the job. | 8 min |
| 7 | P3 Run & Learn: How to Run an AI Agent in Production Safely | P3 Run & Learn covers cut-over, rollback, drift and cost for AI agents in production, and the two-number report that decides whether the programme survives. | 7 min |
| 8 | The 8 Feedback Loops of AI Delivery and the 3 Teams Forget | Eight loops turn the agentic PDLC from a line into a ring. Five close themselves; three (cost, incident, governance) close only if a named person makes them. | 7 min |
| 9 | The Evidence Pack: What Must Exist Before Each AI Hand-off | The minimum artefacts owed at each hand-off of an agentic AI project, and the one column that separates a real gate from a formality: enforced, or only written? | 7 min |
← Getting started · All tracks · Methods decoded →
Teaching it? Session 1 · Kick-off · Session 2 · Frame and spec · Session 3 · Build, run and the loops in the Cohort Kit turn this track into ninety-minute sessions.
✏️ This page is generated from site/content/learn/curriculum.py: an edit made here is replaced at the next sync.
This wiki is the connective layer. The repository is canonical and versioned; these pages are the cross-surface maps and the notes that change too often to live in a reviewed file. Anyone can edit — see Editing this wiki.
▶ Start here · 55 lessons, on the site ↗
- Getting started · 2
- Agentic PDLC fundamentals · 9
- Methods decoded · 7
- Running delivery · 13
- By role · 9
- Teams and organisation · 4
- Practice · 4
- Interviews and careers · 7
The method
The Agentic PDLC
The Eight Loops
Gates & Governance
The Evidence Pack
The journey · end to end
Product manager
Solution architect
Engineering lead
QA lead
DevOps & platform
The method · by role
Product manager
Solution architect
Engineering lead
QA lead
DevOps and platform
Sponsor
How to
Run an NFR workshop
Design an agent on paper
Build, buy or borrow
Cut sprints into bolts
Review by risk band
Prove the bar
Control the token bill
Hold the security boundary
Run a missing-control postmortem
Reference
Mental Models
Glossary
Formulas & Calculators
Decision Trees
Scenario Library
Exercises & Answers
Anti-Patterns
Sources & Confidence
Where do I find…?
The 16 modules
- Agentic Foundations
- LLM Intuition and the AWS Bridge
- Amazon Bedrock Essentials
- Amazon Bedrock Agents
- Agent Builder, Knowledge Bases and Guardrails
- The Agent Loop: No Framework to Strands
- Strands Foundations: Tools, Memory and MCP
- Multi-Agent Patterns with Strands
- LangChain and LangGraph
- LLM Memory Mechanics
- RAG, OpenSearch and LiteLLM
- Amazon Bedrock AgentCore
- A2A and A2UI: Agent Interoperability
- Agentic QA and Evaluation
- End-to-End Production Pipeline
- Agentic Product Lifecycle
Learning paths · Field guide
Study Plans
Eight sessions
Living notes
Error Index
Community Answers
Field Notes
Model & Region Notes
Cost Log
Scoreboard 🧪
The project
Roadmap
Editing this wiki
Maintainer Runbook
📘 The manual · live
🛫 Ninety Days, the simulator
🧰 The workbench
👔 For leadership
🧠 Mental models
Templates · Prompts
Frameworks & acronyms
The repository · Setup
Pinned discussions
Discussion map — where to post what
Exercise & lab index
Simulator Arena — get a lab graded
Pitch an idea
All discussions →