Outline Engine: An AI-Powered Instructional Design Tool
This project is an AI-assisted drafting tool inspired by a real gap in how instructional design work typically begins. Instructional designers often start from raw, unstructured source material, policy documents, SME interview transcripts, or product specs, and turning that into a usable course outline takes hours before any real design work happens. Outline Engine takes that raw material and generates a structured first draft: Bloom's-taxonomy-aligned learning objectives, a storyboard outline, script text, quiz questions with plausible distractors, and branching points, with every claim traceable back to its source so the designer verifies the draft rather than trusting it blindly.
Audience: Instructional designers and L&D teams
Responsibilities: Instructional design, AI prompt engineering, dashboard UX design, information architecture, front-end development, brand identity design
Tools Used: Claude, Claude Design, API Logics, Google Docs
The Problem - Hours lost before design starts
Every course I have built starts the same way: a policy document, a rough transcript from a subject matter expert, or a product spec with a deadline attached. The material is real and useful, but it is not structured for learning. Before I can start designing, I have to read it closely, decide what matters, and translate it into objectives, a flow, and assessment items. That translation step eats hours that could go toward the parts of the job only a person can do well: pacing, tone, and making sure the practice actually builds the skill.
The Key Decision - A draft, not a finished course
It would have been tempting to make the output feel complete and polished, but that risks a designer accepting weak objectives or shallow quiz questions just because they look done. Instead every field is editable the moment it appears, and the interface treats the AI’s output the way I would treat a colleague’s rough outline: useful, directional, and mine to revise.
Working with AI - The prompt was the hard part
Early versions produced objectives that were really just paraphrased headings from the source text. I had to be explicit that objectives should describe observable learner behavior using Bloom’s taxonomy verbs, and that quiz distractors needed to be plausible rather than obviously wrong. I also learned to ask for structured output rather than freeform text, since that is what let every field become independently editable once it landed in the interface.
Reflection - A first pass, not a replacement
This project changed how I think about AI in instructional design work. It is genuinely good at the unglamorous first pass: sorting raw material into a shape a designer can react to. It is not a substitute for judgment about what learners actually need, and I do not think it should be. The tools that matter most are the ones that get a designer from a blank page to a working draft faster, then step out of the way.