Accelerating Content Creation with AI at Udacity

I led the design of an AI-enhanced content authoring system at Udacity to solve a critical bottleneck: the time-intensive process of creating educational content from scratch. By integrating generative AI into the content creation workflow, we reduced initial content creation time from months to 5-6 weeks while maintaining quality standards.

Digital circuitry background with the Udacity logo and text

My Role & Approach

Udacity's content creation process was slow and resource-intensive. Content creators faced significant blank page paralysis when developing courses, and the solo authoring model created bottlenecks that limited the ability to quickly bring new learning paths to market.

The Challenge

  • Lengthy development cycles: a single author could take a quarter or more to create a Nanodegree

  • Blank page paralysis: difficult to structure courses without a clear plan

  • Solo authoring burden: the entire content creation responsibility fell on individual creators

  • Limited flexibility: the time constraints prevented rapid iteration on learning path design

  • Inconsistent quality: without real-time feedback, quality issues surfaced late in the process

My role

  • Led user research with content creators, Nanodegree alumni, and the technical content team

  • Designed the lesson outline wizard and Studio integration concepts

  • Presented concepts iteratively to the VP of Content, securing H2 roadmap inclusion

  • Partnered with product managers and engineers throughout

Understanding the Creator Experience

The "Golden Questions" Approach

Before designing solutions, I facilitated collaborative workshops using a "golden questions" methodology to align the team on the problem space:

  1. Brainstormed critical questions across multiple areas

  2. Grouped questions by affinity to identify themes and priorities

  3. Consolidated insights to establish key requirements with the VP of Content

This process ensured we addressed the right problems and built organizational alignment from the start.

User Research & Stakeholder Engagement

To ground our solution in real user needs, I conducted research with three key groups:

Content Creators (Primary Users)

  • Interviewed experienced course authors about their workflows

  • Identified pain points in the current content creation process

  • Understood where they experienced the most friction

Nanodegree Alumni (Content Consumers)

  • Gathered insights on content quality expectations

  • Validated that AI-generated content could meet learner needs

Technical Content Management Team

  • Understood technical constraints and opportunities

  • Explored integration points with existing systems

Key Findings:

  • Creators experienced significant anxiety starting from a blank page

  • The most time-consuming phase was initial content structuring and outlining

  • Real-time feedback during creation would improve both speed and quality

  • Generated content needed to align with Udacity's learning philosophy

Key quotes from research

It’s difficult to structure a course without having a clear plan in front of you.
I’d like to see better tools for content creators.
Writing course scripts... it feels like busy work... but it has a lot of value to the end-user.

AI as a Creative Accelerator

From Hackathon to Product Concept

A hackathon prototype created by one of our lead developers using the OpenAI API and Google Sheets demonstrated that AI could generate initial course outlines. I expanded this concept into a comprehensive, wizard-based content creation experience.

Strategic Shift: Nanodegrees to Lessons

Based on research findings and discussions with the VP of Content and Technical Content Directors, we shifted strategy from a nanodegree-focused approach to a lesson-based approach. This would:

  • Improve content development efficiency

  • Ensure a smoother workflow for content creation

  • Enable more flexible learning path assembly

  • Allow lessons to be reused across multiple courses

Lesson Outline Wizard

I designed a multi-step wizard that guided authors through the process of creating AI-assisted lesson outlines.

Step 1: Lesson Overview

  1. Lesson name: Author defines the lesson title

  2. School selection: Constrained by Nanodegree groupings (e.g., Artificial Intelligence)

  3. Topic overview: AI helps define suitable lesson content based on inputs

  4. Lesson duration: Used to inform number of pages within lesson

  5. Learning activities: Selection of quizzes, exercises, or both

Step 2: Student Overview

  1. Learner level of experience: Predefined by Technical Content Directors

  2. Pre-requisite skills: Typeahead search pulling from Udacity's skills taxonomy. AI uses this information to calibrate content difficulty and prerequisite knowledge

Step 3: Learning Objectives

  1. Temperature slider: Authors control AI creativity/variance (higher = more diversity, lower = more predictability)

  2. AI-generated objectives: System drafts 5 learning objectives based on previous inputs

  3. Lock and refresh: Authors can lock objectives they're satisfied with before regenerating others

  4. Drag and drop reordering: Authors prioritize objectives by importance

  5. Edit capabilities: Full editing control to refine suggested objectives

Step 4: Lesson Summary & Page Structure

  • Summary view: All author inputs summarized for review

  • AI-generated page structure: Suggested pages for the lesson with edit, reorder, add, and delete functionality

Studio Integration

To demonstrate the full potential of AI-assisted content creation, I designed future-facing concepts showing how the lesson outline wizard would integrate into Studio, a future vision for Udacity's AI supported content management system.

Envisioned Workflow

  1. AI-generated page outlines provide structure for each page's content

  2. WYSIWYG editor allows authors to edit or create content with formatting tools

  3. First-draft AI content gives authors an accelerated start, including:

    • Video script generation

    • Quiz creation

    • Exercise development

This vision helped stakeholders understand the long-term roadmap and secured buy-in for the broader AI content creation strategy. Rather than replacing human creativity, AI would serve as a collaborative first-draft generator that reduces blank page paralysis and accelerates the initial content creation phase.

Stakeholder Alignment & Roadmap Integration

Iterative Stakeholder Engagement

Rather than a single presentation, I engaged stakeholders through an iterative process as the designs evolved, combining formal presentations with ongoing informal discussions.

Phase 1: Problem Alignment

I presented the "golden questions" workshop findings to the VP of Content and key stakeholders, establishing shared understanding of:

  • The core challenges facing content creators

  • Strategic priorities for the content organization

  • Key requirements for any solution

Phase 2: Lesson Outline Wizard

After designing the wizard, I presented the lesson outline planner concepts, including:

  • Interactive prototypes demonstrating the four-step workflow

  • The strategic shift from nanodegree-first to lesson-first architecture

  • User research validation showing creator pain points

  • How AI-generated outlines addressed blank page paralysis

These discussions led to excitement about the approach and refinement of the design based on stakeholder feedback.

Phase 3: Future Vision

As the lesson outline wizard design solidified, I presented the Studio integration vision to demonstrate the long-term potential:

  • How lesson outlines would flow into full content creation

  • AI-assisted page content generation (scripts, quizzes, exercises)

  • The complete creator workflow from outline to finished lesson

Building Buy-In Through Iteration

This phased approach allowed stakeholders to:

  • Understand the problem deeply before evaluating solutions

  • Provide input as designs evolved rather than reacting to a finished concept

  • See the connection between MVP (lesson outline wizard) and future state (Studio integration)

  • Build confidence in the feasibility and value of AI-assisted content creation

The iterative engagement led to productive discussions within product and content organizations about content creation philosophy, quality standards, and change management strategies.

Impact

The goal was to eliminate blank page paralysis and reduce the time and resource burden of content creation at Udacity, while maintaining the quality standards learners expected.

Measured results

  • Content creation time reduced from several months to 5-6 weeks during early testing

  • Positive validation from existing authors during testing of the lesson outline wizard

  • Secured H2 roadmap placement with committed resources and a phased implementation plan

Strategic outcomes

  • Shifted content strategy from solo authoring to AI-assisted collaboration

  • Established a lesson-first architecture enabling more flexible learning path assembly and content reuse across courses

  • Created a foundation for future AI initiatives across other content workflows at Udacity

  • Demonstrated a clear path from MVP (lesson outline wizard) to full vision (Studio integration)