A Guide to AI-Powered Prototyping for L&D
How (and why) to turn your design idea into a working prototype in minutes
Hey folks!
One of the goals we've long talked about—but often struggled to implement—in instructional design is rapid prototyping. In other industries - from design thinking to product development and programming - building and testing early versions of ideas is a critical part of their work and proven to lead to better outcomes. Working prototypes spark clearer stakeholder feedback than static documents, and early user testing saves costly rework down the line.
Research shows that the same is true in instructional design. Rapid prototyping of courses, modules or even discrete activities is shown to improve instructional product quality, shorten development cycles, and enhance collaboration between designers and stakeholders (Daugherty et al., 2007; Gerber & Carroll, 2012). Compared to traditional linear “waterfall” models, prototyping enables instructional designers to integrate formative evaluation at every step, improving usability, learner engagement and likelihood of impact (Tripp & Bichelmeyer, 1990; Nixon & Lee, 2001).
The bottom line here is that prototyping isn't just a faster way to do instructional design—it's a fundamentally better way to do instructional design. It transforms abstract learning theories into testable experiences, turns stakeholder relationships from transactional to collaborative, and replaces guesswork with evidence.
And yet, despite all of these benefits and despite the availability of wire-framing tools like Balsamiq for almost 20, instructional design has overwhelmingly remained stuck in a waterfall mindset. We write exhaustive analysis documents that stakeholders find hard to picture. We create storyboards that fail to capture the real feel of a learning experience. We wait weeks for development, only to find our ideas didn't translate as intended.
So, the question I've been asking this week is: will AI help us to finally bridge this gap between what we know works and what we actually do? Does AI’ have the capability to enable the kind of rapid, iterative prototyping in instructional design that has transformed other design disciplines?
Here’s what I learned so far from trying AI tools in a real project I’m working on for a consumer healthcare company.
The AI Prototyping Toolkit
Before diving into the prototypes, it's worth understanding the AI tool landscape and why I chose Claude for this experiment. Current AI prototyping tools come in two favours:
1. Chatbots (Claude, ChatGPT 4o, Gemini 2.5): The AI tools you probably already know can also write, explain & “build out” code. These tools are best for getting started and - with the right prompting - can create some simple prototypes pretty quickly and effectively.
2. Cloud Development Environments (Replit, Bolt, v0, Lovable): This is a new generation of AI prototyping tools that can build and run your apps from text inputs. They excel at multi-page experiences with very specific design requirements and backend functionality.
From my experience, for instructional designers new to AI prototyping, Claude offers the most accessible entry point. Here's why it worked perfectly for my FDA training challenge:
Source Document Integration: Prototypes work most effectively in instructional design when you use source docs for the info they contain. Claude does a pretty solid job at analysing uploaded documents and maintaining accuracy—crucial for compliance training where every detail must be traceable to regulatory sources
Immediate Share-ability: Claude's Artifacts system creates instant, shareable links with no setup required—perfect for rapid stakeholder feedback
No Technical Barriers: Unlike cloud development platforms, Claude requires zero configuration or learning curve beyond writing clear prompts
Perfect for Proof-of-Concept: When you need to demonstrate value quickly, Claude delivers functional prototypes in minutes
Once you've proven the concept with Claude, you can enhance prototypes using v0 for beautiful visual design, Bolt for multi-page or more complex experiences, or Replit for complex simulations with data tracking — but for now, try getting started with Claude.
The Challenge: Build Three Real-World Prototypes ~ 1 hr
As part of this project, I was tasked with designing three learning activities for a one-hour online asynchronous training module. The team needed a comprehensive learning experience that would prepare different roles—from customer service representatives to senior executives—for various FDA compliance scenarios.
The challenge was creating activities that would:
Address different skill levels - from 6-month employees to 15-year veterans
Cover multiple learning objectives - from basic procedure recall to complex strategic decision-making
Engage different cognitive processes - visual processing, decision-making under pressure, analytical thinking, and interpersonal skills
Provide realistic practice - scenarios that mirror actual workplace challenges and constraints
Each activity needed to be:
Evidence-based - grounded in learning science and traceable to authoritative sources
Interactive - moving beyond static content to hands-on practice
Immediately valuable - applicable to real work situations
Stakeholder-ready - polished enough for executive review and approval
The three prototypes I'll show you represent different approaches to these learning challenges—each designed to hit specific learning objectives while demonstrating different AI-powered prototyping techniques.
They range from individual decision-making practice to complex strategic analysis, covering the full spectrum of skills needed for FDA compliance roles.
In total, I spent ~60 mins building these.
Prototype 1: Interactive Branching Scenario
What it does: This activity enables medical information specialists to practice navigating reporting decisions with realistic timeline pressure and HIPAA compliance requirements.
How I Built It:
Input Used: FDA Adverse Event Reporting Guidance Document (23 pages) + Company escalation procedures SOP
Prompt Used: Check out the structured prompt I used below. Note how it directs AI to use the documents provided to create a multi-path branching decision tree where learners face realistic regulatory compliance decisions under time pressure. Note also how the prompt specifies exact stakeholder roles, regulatory constraints, and consequence modelling to ensure workplace authenticity.
Copy-Paste Prompt Template for Claude:
Build a branching scenario about [SPECIFIC COMPLIANCE/DECISION TOPIC] for [TARGET ROLES].
LEARNER PROFILE:
- Job role: [SPECIFIC ROLES AND EXPERIENCE LEVELS]
- Years of experience: [EXPERIENCE RANGE] in [INDUSTRY/FIELD]
- Common challenges: [TOP 2-3 WORKPLACE CHALLENGES THEY FACE]
- Decision-making authority: [WHAT THEY CAN DECIDE vs. WHAT REQUIRES ESCALATION]
SCENARIO CONTEXT:
- Setting: [SPECIFIC WORKPLACE SITUATION]
- Urgency level: [HIGH/MEDIUM/LOW PRESSURE WITH SPECIFIC TIMELINE CONSTRAINTS]
- Constraints: [REGULATORY/LEGAL/POLICY LIMITATIONS]
LEARNING OBJECTIVES:
After completing this scenario, learners will demonstrate ability to:
1. [SPECIFIC SKILL - e.g., Apply criteria to determine appropriate response]
2. [SPECIFIC SKILL - e.g., Execute proper procedures while maintaining compliance]
3. [SPECIFIC SKILL - e.g., Coordinate appropriate escalation and communication]
DECISION POINTS:
- [NUMBER] major decisions: [LIST KEY DECISION TYPES]
- [NUMBER] realistic alternatives per choice reflecting actual workplace complexity
- Show [TYPES OF CONSEQUENCES - compliance, relationship, business impact]
Create realistic dialogue and consequences based on actual cases from [YOUR INDUSTRY].
The Output:
Prototype 2: Drug Interaction Scenario
What it does: An interactive interface showing how prescription medications interact at the molecular level and in different contexts. This enables the learner to explore the implications of prescription medications in a safe environment.
How I Built This:
Input Used: Clinical Pharmacology Reference Guide (drug interaction mechanisms) + Company product interaction database
What the Prompt Does: Instructs AI to transform dense pharmacological text into interactive visual elements with progressive disclosure. The prompt specifies visual metaphors, clickable pathways, and real-world examples to make complex molecular processes accessible to healthcare professionals.
Copy-Paste Prompt Template for Claude:
Create an interactive visualisation that explains [COMPLEX CONCEPT/PROCESS] for [TARGET AUDIENCE].
LEARNER ANALYSIS:
- Current understanding: [BASELINE KNOWLEDGE LEVEL]
- Professional context: [HOW THEY'LL USE THIS KNOWLEDGE AT WORK]
- Common misconceptions: [TOP 2-3 MISUNDERSTANDINGS ABOUT THIS TOPIC]
- Cognitive demands: [COMPLEXITY LEVEL APPROPRIATE FOR THEIR EXPERTISE]
CONCEPT BREAKDOWN:
- Core concept: [MAIN PROCESS/SYSTEM TO VISUALIsE]
- Key components: [MAJOR ELEMENTS - typically 4-6 main parts]
- Process flow: [HOW THE SYSTEM WORKS STEP-BY-STEP]
LEARNING OBJECTIVES:
Learners will be able to:
1. [IDENTIFY/RECOGNISE SKILL - e.g., Identify the main mechanisms]
2. [EXPLAIN/DESCRIBE SKILL - e.g., Explain how components interact]
3. [PREDICT/APPLY SKILL - e.g., Predict outcomes in different scenarios]
4. [APPLY/USE SKILL - e.g., Apply knowledge to workplace decisions]
VISUALIZATION REQUIREMENTS:
- Visual metaphor: [FAMILIAR ANALOGY - e.g., factory assembly line, highway system]
- Interactive elements: [SPECIFIC INTERACTIONS - clickable pathways, drag-and-drop, sliders]
- Progressive disclosure: Start with [BASIC LEVEL], layer in [DETAILED LEVEL], finish with [ADVANCED APPLICATIONS]
- Include real examples: [2-3 WORKPLACE-RELEVANT EXAMPLES]
Make it accessible with alt-text for visual elements and mobile-compatible for [FIELD STAFF/REMOTE WORKERS].
The Output:
Prototype 3: Interactive Case Study
What it does: In this interactive case study, senior managers navigate a scenario with simulated constraints and pressures like budget constraints, and regulatory timeline requirements.
How I Built This:
Input Used: Company Product Recall SOP (12 pages) + FDA enforcement action case studies + Crisis management procedures
What the Prompt Does: Guides AI to create a complex case study with multiple stakeholder perspectives, budget constraints and time pressures. The prompt structures a guided analysis framework that mirrors actual crisis decision-making while incorporating expert commentary to provide learning moments throughout the experience.
Copy-Paste Prompt Template for Claude:
Build an interactive case study framework for [HIGH-STAKES SITUATION] that develops
[SPECIFIC SKILLS] in [TARGET AUDIENCE].
LEARNER PROFILE:
- Professional role: [SPECIFIC ROLES AND SENIORITY LEVELS]
- Decision-making experience: [YEARS] managing [RELEVANT EXPERIENCE AREA]
- Industry knowledge: [SPECIFIC DOMAIN EXPERTISE REQUIRED]
CASE STUDY DETAILS:
- Industry/Context: [SPECIFIC SETTING]
- Key challenge: [SPECIFIC CRISIS/PROBLEM REQUIRING IMMEDIATE RESPONSE]
- Timeline: [SPECIFIC DEADLINES AND PRESSURE POINTS]
- Resources: [BUDGET/TEAM/TOOLS AVAILABLE]
LEARNING OBJECTIVES:
After analysing this case, learners will be able to:
1. [PROCEDURAL SKILL - e.g., Execute proper procedures including classification and timeline management]
2. [COORDINATION SKILL - e.g., Coordinate cross-functional response involving multiple teams]
3. [STRATEGIC SKILL - e.g., Balance competing requirements and stakeholder needs]
ANALYSIS FRAMEWORK:
- Problem identification: [ANALYSIS METHODS - root cause analysis, scope determination]
- Stakeholder analysis: [KEY STAKEHOLDER TYPES AND THEIR PRIORITIES]
- Option generation: [TYPES OF DECISIONS AND STRATEGY ALTERNATIVES]
- Impact assessment: [CONSEQUENCE CATEGORIES - financial, regulatory, reputational]
- Implementation planning: [EXECUTION TIMELINE AND COMMUNICATION PROTOCOLS]
Include expert commentary from [RELEVANT EXPERT BACKGROUND] with [YEARS] experience.
Reflect the complexity and time pressure of actual [SITUATION TYPE] that [TARGET AUDIENCE] face.
The Output:
Try the prototype for yourself here.
The Reality Check: What AI Does Well & Where It Falls Short
After building these three prototypes, I've learned that AI is not a golden bullet and - of course — needs to be handled with care. By experimenting with Claude, I learned that it handles some instructional design challenges far better than others.
Where Claude Excels:
Text-heavy interactions like branching scenarios and case studies work well because Claude can process complex source documents and create realistic dialogue
Knowledge checks and assessments are natural fits since Claude understands learning objectives and can generate appropriate questions with good distractors
Conceptual explanations benefit from Claude's ability to break down complex ideas into progressive disclosure formats
Where Claude Struggles:
Visual design sophistication - while functional, the aesthetics won't impress stakeholders who expect polished UI
Complex data visualisations - anything requiring dynamic charts or sophisticated graphics needs tools like Replit or v0
Multi-page learning journeys - Claude creates single-page experiences well, but struggles more with navigation and state management across multiple screens
The Bigger Picture: AI Prototyping & the Future Instructional Design
After spending 60 minutes building these prototypes and watching stakeholders interact with them, I've realised that AI prototyping isn't just about faster development—it's about fundamentally changing how we practice instructional design. AI prototyping tools hold some real promise, but we’re at the beginning of a journey that’s both technological and professional.
On the technology: Claude handles some instructional design prototypes really well and others poorly. Text-heavy interactions like branching scenarios and case studies work well because Claude can process complex source documents and create realistic dialogue. Knowledge checks are natural fits since it understands learning objectives and generates appropriate questions. But when it comes to multi-page learning journeys and complex data visualisations, you'll hit walls pretty quickly.
It also becomes evident pretty quickly that - much like prototyping tools of the past like Balsamiq and Figma - AI tools are not built for instructional design. The need for clear and detailed instruction on the how (pedagogy, accessibility etc) is real, and the lack of in-prototype commenting can be frustrating when collaboration is async.
On the professional journey: This isn't "magic AI" that replaces instructional design expertise—it actually amplifies it. The quality of your prototypes depends most on your ability to provide expertly structured prompts with deep knowledge of learning science, pedagogy and accessibility. Claude doesn't inherently know that branching scenarios need meaningful consequences, that progressive disclosure reduces cognitive load, or that feedback should be specific and actionable. You have to teach it these principles.
AI prototyping is another example of how AI leads to the need for deeper ID skills, not automation of our roles.
What excites me most isn't the time savings offered by prototyping—it's the freedom it gives us to focus and professionalise.
The instructional designers who will thrive are those who can combine deep pedagogical knowledge with AI literacy. Not coding in the traditional sense, but prompt engineering informed by learning theory — understanding not just what makes good learning, but how to systematically instruct AI to create it.
I am becoming more and more convinced over time that in the era of AI we (instructional designers) are not becoming obsolete—we're becoming more essential. The strategic thinking, learning science application, and quality judgment remain fundamentally human. But now we have tools that can keep pace with our ideas.
The transformation isn't just technological—it's professional. We can finally be the learning scientists we trained to be, instead of the documentation specialists we've been forced to become.
Happy innovating!
Phil 👋
PS: Want to develop the AI skills needed for this new instructional design landscape? Apply for a place on my AI & Learning Design Bootcamp where I and a cohort of fellow instructional designers / L&D professionals practice and build practical skills for the AI-powered future of learning.