Before you begin
Starting point: Comfort with prompting and checking claims. Revisit the Beginner route if either feels unfamiliar.
Your output: A workflow playbook with inputs, review points, and recovery steps.
Read one guide, try its exercise, then explain the idea in your own words. Use fictional data while practicing.
Your reading sequence
- STEP 01
Map the work
Define the trigger, required inputs, expected output, and human approval point.
Read: Reliable AI workflows → - STEP 02
Reduce unnecessary exposure
Remove data that does not affect the task and identify where the remainder travels.
Read: AI data privacy → - STEP 03
Choose the evidence
Explain how the system selects documents and what happens when none answer the question.
Read: RAG explained → - STEP 04
Trace the evidence chain
Locate the first observed break before choosing a repair.
Read: Diagnose RAG failures → - STEP 05
Design a real review gate
Define evidence, authority, rejection, and escalation.
Read: Design meaningful human review → - STEP 06
Test before scaling
Use normal, difficult, and unanswerable cases to inspect the process.
Read: Evaluation & risk →
Put the ideas together
- Choose one recurring task and draw its handoffs.
- Document missing-input, duplicate-request, and conflicting-source behavior.
- Run five fictional cases and record what changed after review.
Keep your first attempt and your revised version. Write a short note explaining what improved and what still needs checking.
Open the practitioner practice modules →The automation that moved the bottleneck
Follow a support-drafting pilot through misleading time savings, a failed handoff, and a bounded redesign.
Work through the Practitioner case →Check your readiness
- I can identify the person who approves an external action.
- I can distinguish a retrieval failure from an answer failure.
- I can hand a colleague instructions for recovering from an error.
Use these statements as reflection prompts, not a formal score. If one feels difficult, repeat the relevant exercise with a different example.
Take the next step
After your core project, use these guides to connect the learning to an operational decision.
Choose your next step
Stay with this route until you can show your reasoning. You can switch levels whenever the work calls for it.
Beginner route → Expert route →