AI Menu Builder

Grubhub

I led the design of an AI-powered menu builder that can translate plain-text menus into Grubhub’s menu platform, significantly speeding up the merchant onboarding process.

ROLE
Lead Product Designer

TEAM
Myself, (1) Product Manager, (4) Frontend Engineers, (2) Backend Engineers, (1) Copy Partner, (3) Sales Partners

SKILLS
User Research, Prototyping, Wireframing, Usability Testing, Interaction Design

Problem

At Grubhub, efficiently onboarding new merchant partners is essential for expanding the platform’s offerings. Many restaurants provide plain-text menus in formats like spreadsheets, PDFs, or even handwritten notes, requiring Grubhub’s menu-building team to manually convert them into the platform’s menu system. This labor-intensive process was prone to errors and often delayed the time it took for restaurants to go live and start receiving orders.

This led to:

  • Delays in restaurant activation (sometimes weeks).

  • Inconsistent menu quality, with missing descriptions or low-quality images.

  • High operational costs due to manual data entry.

HMW streamline the menu-building process by leveraging AI?

Solution

We built an AI-driven system that could parse and structure menus from spreadsheets, PDFs, and text submissions, generate descriptions based on item name, cuisine type, and common ingredients, create or enhance food images using AI, ensuring a visually appealing menu, and provide an intuitive review interface for restaurants to make final edits.

Research

We conducted interviews with restaurant owners, onboarding specialists, and operations teams to understand pain points. We learned that:

  • Speed matters—restaurants wanted to go live as fast as possible.

  • Consistency is key—poor descriptions and missing images led to lower sales.

  • Control is necessary—restaurants wanted the ability to edit AI-generated content.

90%

Couldn’t complete the flow because of the time it took to locate required information.

20-30%

Required assistance from a sales representative due to a language barrier.

80%

Struggled to understand the value of our offerings, particularly our marketing plans.

User flows

I designed user flows that ensured a smooth experience:

  • Step 1: Restaurant uploads a menu file (CSV, PDF, or text input).

  • Step 2: AI structures the menu into categories, items, and modifiers.

  • Step 3: AI generates item descriptions and images.

  • Step 4: The restaurant reviews, edits, and approves the final menu.

Wireframing

  • I created low-fidelity wireframes to validate the flow with stakeholders.

  • Designed a clean, minimal UI with a focus on speed and ease of use.

  • Used AI previews to let users see auto-generated descriptions and images side-by-side with their original input.

Testing & Iteration

  • Conducted usability tests with merchant partners to refine the AI outputs.

  • Improved description customization after feedback that some AI-generated descriptions felt generic.

  • Enhanced image generation settings, allowing merchants to adjust style and realism.

Results

150% increase in conversion

Conversion more than doubled. 20% of users could complete the flow on their own.

1,400 more signups

Within one month after launch.

Reduced time-to-live by 10 days

Resulting in an increase of $1.84M in yearly revenue for merchants on Grubhub.

Key Learnings

  • AI is a starting point, not a replacement. Giving users control over AI-generated content builds trust.

  • Clarity in UI is critical. Users needed clear guidance on reviewing AI outputs.

  • Speed matters. The biggest value to restaurants was how quickly they could launch.