Embedded Files
Behavior Change Through a Single Decision Point | AI-Assisted Workflow
ABOUT THIS PROJECT
Description: As the instructional designer for this project, I designed and built this course in Articulate Rise, using Twine to prototype and playtest a scenario activity, with Claude and Gemini Notebook supporting research and content development. I worked with SMEs across the operations and medical teams to validate content.
The course spans three lessons and five scenarios covering four common shelter illnesses found in dogs. New shelter staff learn to recognize common illnesses and follow the correct response protocol. Staff are intentionally not asked to diagnose the animal themselves.
Course Highlight: The scenario activity applies adult learning principles throughout. Scenarios are problem-centered rather than fact-based, feedback is experiential (showing the realistic consequence of a choice rather than just marking it wrong), and the gate allows unlimited, low-stakes attempts, reinforcing correct responses through practice rather than penalty.
Challenge 1 - Predictable Responses: The correct response looks nearly identical across every suspected illness (isolate, notify, escalate), which risks making the activity repetitive.
_________
Solution: I built each incorrect option around an instinct that feels reasonable to a non-medical staff member but is clinically wrong, then used realistic consequences to show why it fails and prompt another attempt.
Result: Rather than five scenarios that all reward the same obvious click, each one forces learners to catch a different plausible-but-wrong instinct, keeping the activity varied while still reinforcing the same underlying protocol through repetition, not restatement.
Challenge 2 - Information Overload: These illnesses are well-documented and staff have access to extensive internal resources. Staff don't need a course that teaches signs, symptoms, and protocols illness-by-illness.
_________
Solution: I distilled the content into a small set of response categories, so learners focus on recognizing which category applies and understanding the nuances of that category's response.
Result: Staff leave the course able to categorize and respond correctly to any presentation, including illnesses not explicitly covered, rather than having memorized a lookup table of symptoms tied to specific diagnoses.
Tools
Articulate Rise 360, AI Assistant
Claude
Gemini Notebook (fka NotebookLM)
Twine
Skills
Instructional Design/Adult Learning
Elearning
Gated Scenarios
AI-assisted Workflows
AI Workflow
Claude: I used Claude for the initial course design, drafting lesson content across the three-lesson structure, building sorting activities and knowledge checks, and producing SME-ready drafts of all five branching scenarios. The biggest gain was volume and consistency across revision cycles. As the scenario structure was revised, Claude helped rework all five scenarios to the new structure and scope quickly and consistently, rather than having to manually re-edit each one by hand.
Rapid Prototyping: Claude + Twine. I used Claude to convert the finished scenarios into a Twine import file, then used Twine to verify the branching logic and export an HTML prototype SMEs could click through before the Rise build. This saved me hours of manual work.
Gemini Notebook: I used Gemini Notebook to augment the initial course design and to check every AI-drafted claim against approved source documentation. Because it can be scoped to a specific, controlled set of source files, Gemini Notebook is well-suited for RAG and catching inaccuracies. I also produced three supporting infographics for the course in Gemini Notebook. Every AI-generated output was reviewed and revised by me before it moved to SME review.
Articulate AI Assistant: I used the built in AI Assistant to generate the images featured in the course.
Page updated
Google Sites
Report abuse