How Blend leverages AWS and generative AI to create eye-catching product visuals for ecommerce merchants

How Blend leverages AWS and generative AI to create eye-catching product visuals for ecommerce merchants

The phrase Eye appeal is the new buy appeal stands true when it comes to describing consumers’ ongoing purchase habits. You will not buy something that looks unpleasant. Whether it is packaging or product visuals, the looks of it play a vital role in helping consumers to make purchase decisions.

However, in the current ecosystem, creating professional product visuals is a time-consuming and expensive process. Sensing an innovation gap in the space, Jamsheed Kamardeen, Vaibhav Prakash and Vishwanath Kollapudi founded Blend in 2021 to help online sellers with high-resolution and eye-catching product visuals.

Jamsheed, Vishwanath along with Khailash Santhakumar, who helms the role of Senior Machine Learning Engineer at Blend take us through their growth journey, and how the collaboration with AWS has helped them build a smart editing tool powered by deep learning and generative AI for ecommerce merchants.

Creating professional product visuals in 3 clicks

Blend is a hyper-fast editing experience that allows ecommerce merchants to create professional product visuals in just three clicks. The startup is reimagining the camera and editing experience by abstracting all the complexity to create visuals that sell more.

We are a deep-learning powered photo and design editing app. Instead of editing pixels, our users edit objects. Our deep learning models ensure the conversion of pixels into objects for contextual edits, says Vishwanath.

If you are an e-commerce seller, all you have to do is upload a product photo and Blend will instantly identify the object in the photo. Further, it also identifies the pose of the object and removes the background. And finally, the app generates thousands of designs which can be readily listed on 16 different ecommerce marketplaces and eight different social media platforms. So a jewellery store owner who is offering 20 percent off on their merchandise of earrings and bangles can create visuals for their eCommerce listings and social promos in under a minute.

For any product-led company, time to value is the key element to the growth journey. Our deep learning and Generative AI models ensure that we provide value to our customers in under a minute, adds Vishwanath.

Prioritising convenience and cost-saving with AWS

Blend uses a lot of managed services and collaborates with AWS for their engineers to build deep tech solutions without worrying about setting up complex infrastructure.

We have been using AWS from Day 1. Amazon ECS on AWS Fargate runs most of our microservices that use Amazon DynamoDB as the primary database. AWS CodePipeline is the backbone of our CI/CD. All the images that users upload and the designs they create are stored on Amazon S3 and are served via Amazon Cloudfront. Most importantly, Amazon Sagemaker helps us deploy and scale our ML models that are fundamental to our solution, shares Jamsheed.

He believes that AWS prioritises saving costs for the customers over racking up a huge bill and has their best interests at heart.

Apart from cutting-edge solutions and AI/ML-powered models, the AWS account management team also helps in setting up training sessions. Action items by the AWS team helped Blend reduce cloud costs by over 25 percent.

Blend’s image classification and pose estimation models are served by Amazon Sagemaker endpoints. It takes care of load balancing and auto-scaling for all our models, while serverless services help keep the costs low, says Khailash.

How AWS is making lean startups a reality

Blend algorithm generates more than 1 million designs every day for users to pick from and thousands of online businesses use the app to generate listing images and social media designs for their websites. These are remarkable numbers considering the startup is only a year old. In another year of its operations, Blend plans to deploy 2x the number of models, and reach 5x the scale.

All this is developed by our lean team of six developers and four ML engineers. This would not have been possible if AWS-managed services were not this intuitive. We see ourselves as a core ML company. AWS plays a huge role in whatever we are doing, concludes Jamsheed.

This content was originally published here.