OxygenVision – AI E-commerce Design Agent by JD Retail 186

OxygenVision Introduction

OxygenVision is an AI e-commerce design agent built by JD.com based on its self-developed Oxygen e-commerce architecture and the Joyai large model. Its predecessor was "JingDianDian" (京点点). It is primarily aimed at e-commerce operations staff and positioned as the first design agent in the retail industry. JD merchants can use it completely for free.

JD itself has a huge volume of products, and the demand for SKU creative assets has always been very high. This platform can generate tens of millions of assets per day, serving millions of merchants. It helps operations staff quickly handle 90% of retail image and basic product video production with AI, especially in those situations where "assets are needed urgently and there's no designer available."

OxygenVision Core Features

• E-commerce Asset Generation: Supports product main images, scene images, detail page images, AI model images, store decoration images, marketing posters, and short video scripts. Covers common scenarios such as New Year, home, outdoor, etc. The generated assets have a relatively consistent style and can basically be used directly.

• Real-Person Livestream Video Generation: Not those obviously fake virtual humans, but AI-generated livestream videos that look close to real people. You can customize the model's gender, facial features, age, body shape, and outfit, with quite natural skin texture and lighting effects. No real shooting required — a version can be produced in just a few minutes.

• Cross-Border Product Image Sets: Aimed at cross-border merchants, generates product image sets that comply with target market platform specifications and aesthetic preferences, adapting to the style requirements of different regions.

• AB Testing and Data Optimization: Can generate multiple versions of assets simultaneously and serve them to different audiences for testing. The system automatically collects click-through and conversion rate data to help select the best-performing version for continued use. Official data shows ROI can be improved by about 40%.

• Copy and Script Generation: Enter product selling points and the target platform, and it can directly generate titles, recommendation copy, and short video scripts.

OxygenVision Application Scenarios

• Daily Product Listing: New product listings require main images, detail pages, and video scripts. Originally this required renting a studio, finding models, shooting, post-production, etc. — now it can be done in two steps.

• Major Promotions: During 618 and Double 11, large volumes of promotional assets are needed and can be generated in bulk, without rushing to get them done at the last minute.

• Asset Iteration and Optimization: Some old assets may have low click-through rates. They can be used for AB testing, and underperforming versions can be replaced after comparison and optimization.

• Cross-Border E-commerce: Generates product images that meet the requirements of platforms like Amazon, Joybuy, and TikTok Shop, adapted to the aesthetics of different markets.

• Store Decoration: From product listing to overall store visual identity, brand asset production can be completed in one stop.

OxygenVision Advantages

• Significant Efficiency Improvement: Asset generation efficiency can be increased by over 300%. Traditional single livestream video costs around 500–2,000 yuan; with this tool, there's essentially no shooting cost. Brands like Nike, Adidas, and HLA (Hai Lan Home) have seen product listing speed improve by over 90% and conversion rates increase by 29.03%.

• Large Production Capacity: The platform serves over 3 million merchants, providing over 30 million business decision-making services daily, with overall operational efficiency improved more than 10 times.

• Data-Driven: Built-in AB testing and data feedback mean asset optimization is backed by data, not by gut feeling.

• Low Barrier to Entry: No design background required — ordinary people can handle most e-commerce image needs. The interaction is quite natural, and beginners can get started in about 5 minutes.

• Compliance Assurance: Integrated with the HongDun (Red Shield) system to filter out non-compliant content, so usage within the JD ecosystem carries no copyright issues.

How to Use OxygenVision

1. Log In: JD merchants can log in directly with their merchant account on the official website (ai.jd.com/oxygenVision/) to access the workspace. Individual users or non-JD merchants need to contact customer service to apply for an invitation code.

2. Generate Image Assets: Upload a white-background product photo or a real-shot photo. The system will automatically cut out the subject, which can also be manually adjusted. Then select a template scene (such as New Year, living room, outdoor) or enter your own description. The system will generate several high-definition images. Finally, you can add marketing tags (like "gift with purchase" or "new arrival"), and after downloading, they can be uploaded directly to the JD merchant backend.

3. Generate Copy or Scripts: Fill in the product selling points and target platform, select whether you want titles, recommendation scripts, or short video scripts, and the system will output content directly that can be used with minor edits.

4. AB Testing: Select 2–3 versions of assets, set the target audience, duration, and optimization goal (click-through rate or conversion rate). The system will automatically serve them and collect data. Based on the test report, choose the best-performing version for large-scale use.

Why Choose OxygenVision

OxygenVision's core value lies in its integration with JD's data — the generated assets aren't just good-looking, but their conversion performance is validated by real data. From generation to testing to optimization, it forms a complete loop, and it's currently completely free. For JD merchants, it's a relatively direct choice for improving asset production efficiency and conversion rates.

Competitor Comparison

OxygenVision FAQ

Q1: Can both self-operated and POP merchants use it? Yes, both can, with no category restrictions.

Q2: Is it paid? Currently free, with no paid plans.

Q3: Will the generated materials violate any rules? It integrates the Red Shield system for content filtering, ensuring safe use within the JD ecosystem.

Q4: Can people without design experience operate it? Yes, templates and prompts are ready-made, and you can get started in minutes.

Q5: What formats and sizes are supported? Images support PNG, videos support MP4, with 5 common aspect ratios built in.

Q6: Can it be used on mobile? Currently only supports desktop.

Q7: How long are materials saved? They are automatically saved to personal cloud storage, with no expiration limit, and can be reused or modified repeatedly.