Project · Capstone (Harvard ALM) · Spring 2019

Cooking from Scratch

A self-guided cooking skill development tool, designed for people who want to reach fluency in the kitchen on their own terms, without a prescribed curriculum. The capstone project for my Harvard Extension School ALM degree in Digital Media Design, grounded in instructional design theory and driven by multiple rounds of user research.

User research Instructional design Interaction design Competitive analysis Adobe XD

Cooking from Scratch app: I Want to Learn screen with a baguette-making goal filled in, including description, questions, and notes — Cooking from Scratch · "I Want to Learn" screen · higher-fidelity prototype

The problem

Cooking skill development has no system

The goal isn't to cook more recipes. It's to reach fluency: the feeling of competence and confidence that comes from accumulated, organized experience. Most people working toward that aren't following a curriculum. They're picking things up from YouTube, recipes, trial and error, things a parent showed them once. The knowledge is scattered across memory, scraps of paper, bookmarked recipes, mental notes that fade. There's no way to track what you've tried, what you've learned, what you want to learn next, or why something went wrong last time.

Cooking from Scratch is a tool that puts skill development in the user's hands: track what you're learning, log what you've cooked, and organize your knowledge in a way that actually accumulates over time.

The 2019 version identified a tension that's now a solved problem: users weren't sure whether content in the app was self-generated or app-generated. That confusion (who decides what I'm supposed to learn?) was the central UX challenge. It's a design question that has only gotten more relevant as AI-generated content has entered every category of app.

Research approach

Grounded in theory from the start

Before talking to users, I did substantive background research: read 12 journal articles on cooking education and instructional design, reviewed coursework from my ED103 Instructional Design course, and conducted a competitive review of 8 products. That foundation shaped both the research questions and the design direction from the first wireframe.

Literature review

What does research say about how adults learn cooking skills? What instructional design frameworks apply?

12 articles

Competitive review

What cooking tools and learning platforms exist? Where are the gaps?

8 products

Exploratory interviews

What does cooking skill development actually look like in practice? What are users' goals, pain points, and mental models?

N=7

Wireframe feedback round

Do the proposed screens map to how users think about their cooking? Is the user-generated vs. app-generated distinction clear?

N=5

Prototype feedback round

Does the higher-fidelity "I Want to Learn" flow communicate the app's purpose and feel right to use?

N=4

Exploratory interviews

What the interviews surfaced

Seven interviews explored how people currently develop cooking skills, what they track, and what they wished existed. The consistent finding: knowledge accumulates but has nowhere to go.

"I typically work on a cooking skill over a large chunk of time, typically months or years. Having something that will help me remember what I did a couple of months ago or pick up where I was last, would be helpful."

Exploratory interview participant

"My current methods are quite silo'd off, and disjointed."

Exploratory interview participant

"Somewhere, or some way, to keep track of my progress as I learn this skill."

Exploratory interview participant, on what was missing

Instructional design applied

Three frameworks that shaped the design

A key goal of this project was applying what I'd learned in ED103 to UX design. These three concepts from instructional design theory directly influenced the product's structure and the choices made about what the app does and doesn't do for the user.

Heutagogy

Self-determined learning: the learner sets their own goals, pace, and path.

Applied: The app doesn't prescribe a curriculum. Users define what they want to learn and what progress means to them. The system organizes and surfaces their own knowledge. It doesn't supply it.

Backwards design

Start with the desired result, then determine what evidence would show you've reached it, then design the learning experience.

Applied: The "I Want to Learn" screen starts with the goal (a skill the user wants to develop), then surfaces relevant logged experience and resources, working backward from outcome to content.

Andragogy

Adults learn differently from children: they're self-directed, motivated by relevance, and draw on existing experience.

Applied: The Food Log captures what users have already done, not just what a course tells them to do next. It treats prior experience as the foundation, not a blank slate.

Competitive review

Eight products, no close competitors

I reviewed products across two categories: cooking-specific tools (instruction and recipe management) and general-purpose organization tools that people could theoretically adapt for cooking. The review confirmed that no product addressed skill development as the primary use case. (Completed in 2019; product details may be outdated.)

Product	Category	Gap relative to Cooking from Scratch
Allrecipes Cooking School	Cooking instruction	Curriculum-driven; no user-defined skill goals or personal logging
America's Test Kitchen Online	Cooking instruction	Expert-led content; no mechanism for tracking your own skill development
Rouxbe	Cooking instruction	Professional culinary curriculum; structured, linear, not self-directed
Foodist Kitchen	Cooking instruction	Habit-focused; no long-term skill tracking or knowledge organization
Paprika Recipe Manager	Recipe management	Organizes recipes; no skill layer, no logging, no learning goals
NYT Cooking	Recipe media	Editorial content; saving recipes is not the same as tracking skill development
Evernote	General organization	Could be adapted, but no cooking-specific structure or skill framework
Airtable	General organization	Powerful but requires significant setup; not accessible to most users

Designs

Three wireframed screens, one prototyped flow

I wireframed three screens that represented the core of the product, then built a higher-fidelity prototype of the "I Want to Learn" flow for the second feedback round. Screens were designed in Adobe XD.

Screen 1 of 3

"I Want to Learn…": user-defined skill goals. 5 of 7 interviewees mentioned wanting to learn knife skills. Directly applies backwards design: start with the goal, then surface relevant content.

Screen 2 of 3

Food Log: a structured record of what you've cooked. Addresses the core problem: cooking knowledge is "typically sitting somewhere in my memory, and is difficult to recall."

Screen 3 of 3

Home: a dashboard view of all cooking content. Designed for glanceability: "You can almost brain dump everything here, which is a huge relief."

Prototype flow · 8 frames

Frame 4: questions and notes fields appear

Frame 8: completed goal with questions and notes

Applying backwards design: the user starts by naming their goal before the app asks anything else.

Higher-fidelity prototype of the "I Want to Learn" flow · color, imagery, interactive · built for the second round of user feedback

Design validation

Feedback that sharpened the design direction

From wireframe feedback

"I guess I was confused at first when looking at the mockup — I wasn't sure if the content was user-created or app-created."

Wireframe feedback participant

"This is exactly what I need. This would be so helpful and cure so much of my cooking anxiety."

Wireframe feedback participant, on the Food Log

"I would like to see how the recipes on my main page relate to the skills I am trying to develop."

Wireframe feedback participant, on the Home screen

The user-generated vs. app-generated confusion surfaced in both the wireframe and prototype rounds. It became the central design problem: the app's value proposition depends on the user recognizing they are the author of their own learning path, not a consumer of the app's content. Every subsequent design decision was filtered through that question.

Where this goes next

Revisiting the design in 2026 with AI design tools

The 2019 thesis put the learner in charge of their own path (heutagogy: self-determined learning), offering structure without removing that authorship. In 2026 I used the project to try three AI design tools (Claude Design, ChatGPT Canvas, Google Stitch), redesigning the "I want to learn" screen with two prompts: one that supplied my original 2019 screen, one that supplied only the product goals, no imagery.

ChatGPT redesign of the I Want to Learn screen, with 2019 screen supplied as reference

ChatGPT Canvas · 2019 screen supplied as reference

ChatGPT redesign of the I Want to Learn screen, product goals only, no imagery supplied

ChatGPT Canvas · product goals only, no imagery supplied

I stopped iterating there. How much an app should propose, and where the line sits between what the system suggests and what the person decides, is not a call to make next to a tool. It belongs to the people who would use this. The next step is feedback from the app's actual audience, and designing around what they need is the work I care about most.

What this demonstrates

Research-driven design, applied theory, and a novel problem space

Capability	How it showed up in this project
Research before design	12 journal articles and a full competitive review before the first wireframe. The research shaped which instructional design frameworks applied and why; it wasn't decorative background work, it was the design foundation.
Applying theory to UX	Heutagogy, backwards design, and andragogy weren't mentioned as a checklist. Each one mapped to a specific structural decision: how goals are set, how content is organized, how prior experience is treated.
Iterative research design	Three rounds of user input (exploratory, wireframe feedback, prototype feedback), each with a distinct research plan, participant criteria, and method rationale. Findings from each round directly changed the next design.
Identifying the core UX tension	The user-generated vs. app-generated confusion wasn't a copy problem or a visual problem. It was a fundamental question about who owns the user's learning path, and surfacing it early changed the design direction entirely.
Novel problem space	The competitive review found no product that treated cooking skill development as a first-class feature. Every existing tool was either curriculum-driven (expert decides what you learn) or recipe-management (no skill layer at all).