your productivity, efficiency, quality

SKIN CARE

With AI e-commerce Image and Text Generator

Boost:AI-powered image and text generator

Hiring photographers, designers, and copywriters for product images and descriptions in a large online store is costly and time-consuming. To address e-commerce challenges, I'll create a tool that uses AI for text and image generation to improve efficiency.

Methodology

Research Question

How can AI text and image generation boost content creation efficiency and quality in

e-commerce?

Observational survey

In-Depth Interviews

Desktop Research

Case Study

User Journey

Problem statement

Brainstorm

Business Model

Canvas

Information Architectture

Wireframe

UI Design

Prototype

Front-end Development

Back-end Development




Usability Testing

Questionnaire

In-Depth Interviews

Empathize

Define

Ideate

Prototype

Test

>

>

>

>

>

>

>

>

Product Design

I initiated this project from the ground up, defining the problem, conducting user research, analyzing competitors, and developing product strategies. I also handled UI/UX design and full-stack development, exploring how AI text and image generation can boost content creation efficiency in e-commerce.

Type

Overview

June - August 2024 (3 months)

Timeline

Full Stack Designer

Role

Product strategy,User Experience design(UI/UX),

Front/Back end development.

Responsibilities

Although they know AI can enhance efficiency, the learning barrier is too high.

AI technology offers new opportunities for e-commerce growth, but tools like ChatGPT, Midjourney, and Stable Diffusion can be complex to learn and use, with a high entry barrier.

Emphatize

The daily work of e-commerce operators is repetitive and inefficient.

After observing a day in the life of e-commerce operators, we discovered they spend considerable effort on tasks like product photography, design, copywriting, and listing products—activities that are both repetitive and inefficient.

+

=

Business

Viability

Customer

Desirablility

Feasibility

Technology

Product Design Opportunity

Needs&Desires

An AI-Powered Text and Image Generator for E-commerce

Efficient content production methods

User-friendly AI technology

Cost Reduction:Helping e-commerce businesses reduce expenses

Increased Productivity: Enabling companies to generate content faster

Improved Quality:Providing superior product images and descriptions

Requirement

Possibilities

Existed AI technologies:

Text-to-Text

Text-to-Image

Image-to-Image

Opportunity

Define Target User

Target User

💁‍♂️ B2C/C2C business

👨‍💼 Operation and marketing staff

🧑‍🎨 Designers

👩‍💻 Shopkeeper

💁 Photographer


📸

User Interview

In-Depth Interviews

I interviewed retailers, wholesalers, manufacturers, and designers separately, and I found that their pain points are almost all concentrated in the pre-sales stage, and their demands for AI technology are basically concentrated on data analysis capabilities, text-to-image capabilities, image-to image capabilities, and image-to-text capabilities.

Online Interview with retailers(Temu- DIY Bracelet)

Online Interview with wholesalers(Alibaba-Sock)

Online Interview with manufacturers(1688-furniture)

Online Interview with Designers( John&Lewis)

Steve

Por

Painpoints

Needs

1.Product Photos:
Businesses invest heavily in photographing products, including setting up scenes, lighting, and props, or hiring studios, which often have time and style limitations.

2.Title/Description:
Creating unique product titles and descriptions requires extensive keyword research on best-selling items.

3.Customer Service:
Outsourcing to platform teams or hiring customer service staff, with salaries around £4,000 annually.

4.Product Uploading:
Uploading products one by one is time-consuming and labor-intensive.

1. Product Photos:
AI generates realistic, multi-angle scene images from product photos, with customizable size, specs, and style.

2. Title/Description/Keywords:
AI gathers relevant keywords and creates culturally appropriate titles and descriptions, enhanced by generated images.

3. Customer Service:
AI handles customer service and logistics, lowering costs for independent sites.

4. Product Uploading:
AI automates product uploads by organizing photos and descriptions into categories and folders.

Before Travel

During Travel

After Travel

Inspiration

Panning

Depature

Explore

Homeward

Follow-up service

1.Sees a travel post on the social media.

2.Search and compare different destinations on website and app.

3.Edinburgh is a potential travel destination.


1.Booking a trip&tour guide......

2.Prepares for trip(clothes,shoes,others)

3.Books in-trip-activities

1.Confirm order information

2.Check the walking distance

3.Determine foot traffic

1.Visit the historic sites

2.Explore the delicious food

3.Take part in some entertainment

1. Record the travel experience

2. Post the photo and video to social media

3.Return to the hotel

1.Review and Feeback

1.Traveler websites/app

1.Social media

2.Search engines

3.Traveler websites/app

1.Map app

2.Traveler websites/app

1.Historic sites

2.Restaurant

3.Events locations / Shopping Mall

1.Map app

2.Traveler review websites/app

1.Social media(Instagram,Twitter)

2.Traveler review websites/app

search on website and apps

Consult Friends/customer

service

Book on the website/app

Reserve and Pay

Arrive the destination

Review and Feedback

Record the trave experience

Follow-up service

Come back hotel

Visit the historic sites

Explore the delicious food

Explore the events

Shopping

Have an idea for a trip

1.The recommended routes from social media tend to create unnecessary congestion.


2.It is easy to lost in a long walking distance in a new city.

1.I'd be worried about my safety beacause of the dark lighting system.


2.Especially afraid of alcoholics or extreme teenagers.

Elena

Age: 23 years old, Master Students,

Interests: Traveling/Photography/Movie

Personality: Easy-going and enjoys niche exploration

Scene

1.She plans to spend her Christmas holiday in Edinburgh with her friends together.

2.Duration: Maybe 3-5 days

3.She always busy studying and don’t have time to do too many travel plans.

Predict

1.Relaxing, cultural travel on a budget

2.Visit Museum/gallery/natural landscape during the day,View city buildings at night.

3.Prioritize safety and comfort when traveling

Stage

Goal

think

Behavior

Touch points

Emotion

Pain Points

1.It is difficult to make a decision because there are a lot choices.


2. I don't know the history, culture, and localised activities in Edinburgh.

😐

😐

😐

😐

Before Sale

During Sale

After Sale

Data Analysis

Product Selection

Prepare Materials

Product Pricing

After-sales Service

Effect Monitoring

Need to count the advertising data of the previous day and respond tocustomers.

Need to conduct market research and select products, and contact photographer to shoot products.

Need to prepare new materials (scripts and copywriting) and connect with designers to take photo and design.

Price new products, list them on e-commerce platforms, and set discounts with my collegues.

Upload all the materials to the platform, such as ads and videos.

Monitor advertising data from time to time and adjust advertising strategies at any time

1.Websites/app

1.Social media

2.Search engines

3.Websites/app

1.Websites/app

1.Websites/app

1.Websites/app

1.Social media(Instagram,Twitter)

2.Websites/app

Reviewing Competitor Data

Selecting Best-selling Products

Preparing for Product Launch

Contacting Photographers and Copywriters

Preparing Product Copy

Review and Feedback

Customer Pain Point Analysis

Follow-up service

Responding to Customers Selectively


Preparing Photography Requirements


Price and Cost Analysis

Setting Prices and Selecting Product Photos


Product Launch

Data Analysis of Current Products


Title/Content Description:

Spending a considerable amount of time searching for keywords and descriptions of best-selling products on existing platforms to create unique product titles and descriptions.

Customer Service:

Outsourcing to the platform's customer service team or hiring customer service personnel yourself, with each person's salary being approximately £4,000 per year.

Matt

Age: 33 years old, Online merchant,

Interests: Traveling/Photography/Movie

Personality: Easy-going and enjoys niche exploration

Scene

1. He plans to list and sell products online.

2. He is always busy with the operation of the store and does not have more time to take photos, design, and write copy for the products.

Predict

1. Budget-friendly product photography plan

2. Launch 5 to 6 products each week, and provide them with product images

3. When using, prioritize cost, speed, and the authenticity of the implementation effect.


Stage

Goal

think

Behavior

Touch points

Emotion

Pain Points

Product Photos:

Businesses need to spend a significant budget and time to photograph their products, such as setting up scenes, lighting, and props. Alternatively, they can book a photography studio to help with the shooting, but this

Short walking distance travel + Navigation

Highlight the security Level of the travel route

Opportunity

Engage with Harry Potter inspirations to learn culture and history

AI to gather keywords from e-commerce platforms and generate culturally and linguistically appropriate titles, descriptions, and keywords. Enhanced descriptions based on generated images are also desired.

AI to manage customer service and logistics for independent sites to reduce costs.

Opportunity

AI-generated realistic scene pictures in various styles from product photos, showcasing different angles, and allowing adjustments in size, specifications, and style.

😐

😐

😐

😐

Define Problem Priority

E-commerce's biggest challenge is the high cost and time demands of traditional photography, design, and copywriting.

Our research shows that e-commerce operators face challenges in designing product images and writing product pages, tasks that are time-consuming and costly. They hope AI can help boost efficiency and reduce both labor and costs in these areas.

It takes a lot of time and money to build scenes and shoot products

It takes a lot of time and money to design the products image

It takes a lot of time and money for operators to write titles and descriptions

Remove background

Text-to-image

Ideation

Boost offers a competitive edge over existing rivals by integrating features like product image design, creation, and copywriting optimization into a comprehensive, intelligent solution. It enhances e-commerce efficiency while reducing labor input, making it a standout in the field.

Ideation

Comprehensive

text and image solution

Image generation

Product Description

Image Copywriting

Title Generation

Title Optimization

Desktop Research

Competitive product analysis

While each competitor has its unique strengths and niche focuses, they share common features such as user-friendly interfaces, visual content creation capabilities, automation and AI integration, extensive template and asset libraries, and a focus on enhancing efficiency and productivity. These commonalities highlight the essential needs in the visual content creation market.

Flair.ai

Canva

Pebblely

Mokker.ai


Ideation

Prototype

Business Model Canvas

Information Architecture/Wireframing

Prototype

Boost integrates multiple features such as product image design and creation, and copywriting and optimization, providing a more comprehensive and intelligent solution focused on improving e-commerce operational efficiency and effectiveness while reducing labor input.

In the early stages of design, Information architecture is used to confirm the structure of the Boost, and sketches are used to outline the UX components, allowing for quick adjustments and modifications.

I quickly built the overall process and key feature experience related to the business using Figma's prototype feature.Including the official website, Image generation feature, text generation feature.

REVENUE STREAMS

The revenue you generate from each customer segments

COST STRUCTURE

The costs incurred to operate a business model

KEY RESOURCES

CUSTOMER RELATIONSHIPS

The types of relationships you

establish with specific customer

segment

CHANNELS

How you communicate or deliver

value proposition to your target

customers

The most important assets required

to make the business work

KEY PARTNERS

The network of suppliers and

partners that make the business

model work

CUSTOMER SEGMENTS

The different groups of perople or

organizations you aim to reach

KEY ACTIVITIES

The most important activities a

company must do to make its

business work

VALUE PROPOSITION

Product, service, features and

aspects that create value for a

specific customer segment

1.AI technology providers

2.E-commerce platforms (e.g., Amazon, Shopify, Temu, TikTok)

3.Design and content creation tools (e.g., Canva)

4.Marketing agencies

Logistics and customer service providers

1.Developing AI-driven tools for image design and copywriting

2.Integrating with e-commerce platforms for seamless operation

3.Continuously improving AI algorithms for better performance

4.Marketing and promotion of the platform

5.Providing customer support and training

1.AI and machine learning expertise

2.Software development team

3.Partnerships with e-commerce platforms

4.Content and design template libraries

5.Customer support team

1.Research and development

2.Marketing and sales

3.Customer support and training

4.Platform maintenance and updates

5.Partnerships and integrations

1.Subscription fees (monthly/annual plans)

2.Pay-per-use fees for specific features

3.Premium services and features

4.Affiliate marketing and partnerships

5.Advertising and promotional services

1.Comprehensive solution for product image design and creation

2.Advanced copywriting and optimization tools

3.Improved operational efficiency and effectiveness for e-commerce

4.Reduction in labor input and costs

5.Increased click-through rates and conversion rates

1.Personalized customer support

2.Online tutorials and training programs

3.Community forums and user groups

4.Regular updates and improvements based on user feedback

1.E-commerce businesses

2.Small and medium-sized enterprises (SMEs)

3.Individual online sellers

4.Marketing agencies

5.Entrepreneurs and startups

1.Direct sales through the website

2.Partnerships with e-commerce platforms

3.Social media marketing

4.Online advertising

5.Email marketing campaigns

Test

Usability Testing

To validate the user experience and the effectiveness of image and text generation, shop owners and target consumers were invited to participate in usability testing. They were asked to use Boost to generate product images and record their questions and feelings during the experience

Tool's interface has navigation difficulties,

lacking functional demonstrations and guidance.

The home page and generation history are hard

to tell apart due to their similar appearance.

The final version of the design optimised the Logo, official website, and the interface of the image and text generator based on the user feedback, and introduced a brand identity system.

Iteration

Logo & UX & UI

Brand System

A brand Identity system was created to guide all relevant designs, including Logo, colour, typography and

so on, providing the brand guideline for the future marketing campaign design.

Technical Implementation

Front-end Development

Back-end Development

The development framework used is Next.js, the frontend styling is designed with Tailwind and Headless UI, local storage is handled by Local Forage, and object storage is provided by Vercel Blob. The background removal model uses the RMBG-1.4 model, and the background generation model is based on Stable Diffusion's SDXL model. The project's ComfyUI workflow and backend API are deployed on comfy.icu. Text generation references the Moonshot AI open platform API.

First, determine the image size and zoom in. Identify the main subject, remove the background,

and apply a mask after keying. Create a line drawing from the keyed subject for ControlNet to guide image generation, allowing for the addition of related backgrounds, coloring, and rendering. Then, introduce positive and negative cue words and incorporate the SDXL model for Stable Diffusion. Use random seeds to diversify results, then composite the image with VAE Encode to add a background. Finally, render again to enhance the quality of the product image.

Result

Reflection

Positive Impact

Limitations in Innovation and Authenticity

Boost reduced e-commerce product content creation time from 3 days to just 2 minutes. Additionally, survey results indicate that most respondents found AI-generated images visually appealing, with values ranging from 4 to 5.

Thanks for reading!

Please click this button to try out the website:

Let’s Go

The Length of Image and Text Content Creation

The Length of Image and Text Content Creation

Photographer + Designer + Shopkeeper

Before

1 + 1 + 1

3 Days

2 minutes

After

1

Photographer or Designer or Shopkeeper

AI image generation boosts visual appeal but lacks innovation and authenticity.:

• In creative design scenarios, it may fail to deliver unique, personalized results.

• Some consumers doubt the authenticity of these images, questioning their accuracy in representing

products. This scepticism can influence purchasing decisions, highlighting the need for transparency

to build trust.