AI Fluency & Building · AI Systems Analyst
Go deeper. Real-world tasks that need more research, planning and a polished result.
Task 1 of 3
Find, rate and fix 15 ways a support bot can fail.
What to do
List 15 ways an AI customer-support chatbot for a business you choose could fail (wrong prices, rude tone, made-up policies, missing hand-off). Rate each by impact × likelihood and propose a fix for the top 5.
You’ll make
Done looks like
A table of 15 failure modes with scores and fixes for the top 5.
Task 2 of 3
Mark where people must stay in an AI workflow.
What to do
Map an AI-assisted business workflow (for example, AI drafting social posts or replying to reviews) and mark where a human must check or approve, and why.
You’ll make
Done looks like
A workflow map with human checkpoints marked and a reason for each.
Task 3 of 3
Recommend the right AI tool with evidence.
What to do
For a business you choose, compare 3 AI tools for one job (chat support, content writing or image creation) on quality, cost, ease of use and data handling. Write a 2-page recommendation.
You’ll make
Done looks like
A 2-page comparison with a clear recommendation and reasons.
Learn before you build
Free course
Learn the standard notation for drawing business processes.
Free tool
A searchable database of AI risks, sorted by cause and domain.
Free tool
Thousands of real AI failures to study, cite and learn from.
Free tool
Free diagramming tool for process maps and system diagrams.
Article guide
The most widely used framework for mapping, measuring and managing AI risk.
Article guide
The classic essay on where to intervene in a system.