About Toasty
Toasty is a French startup founded in 2022. It operates as an eco-responsible shopping platform that promotes sustainable brands in lifestyle and wellness sectors.
The app curates products from over 1,000 brands, including vegan sneakers, upcycled swimwear, and solid shampoos, while using gamification to engage users.
The platform has attracted 200,000 users and focuses on expanding its community and brand partnerships.
The Challenge
Toasty regularly imports new products from brands. Each time this occurs, the product description must be manually updated to align with Toasty’s editing style.
Plus, during the important brand products stage, Toasty would also like to import the customer reviews for each product from the brand websites.
The issue is that not every website is the same meaning that Toasty had to adapt its scraping code every time in order to retrieve every customer review.
The Solution
Automat-it was brought in to create a Proof of Concept (PoC) to find a solution to these challenges and improve Toasty’s efficiency and accuracy in the process.
The PoC was based on Amazon Bedrock to seamlessly integrate with Toasty’s infrastructure based on AWS Amplify – a complete suite of tools and services designed to help developers effortlessly build, scale, and deploy mobile and web applications with speed and efficiency.
The solution itself will use Automat-it’s best practices for a secure and stable environment. It is aware of all existing product descriptions created without AI and used them as a reference to ensure the Large Language Model (LLM) maintains a similar style and logic.
Additionally, there are specific instructions per product category – such as the use of emojis and mandatory fields. The LLM determines which set of instructions to apply based on the incoming raw data.
The Results
Fine-tuning and implementing a GenAI solution has allowed Toasty to optimize their product description and product customer review process.
Turning what was a highly manual and time-consuming process into one that is more efficient, comprehensive, and results in a uniform Toasty product.
Standout results included:
- Training and implementing the GenAI solution took less than one month
- Error free sessions increased from 70% to 90% as script errors reduced
- Operational team reduced time for data enrichment by 2x
Boost Efficiency Today
Are you a startup like Toasty looking to achieve greater efficiency and accuracy using GenAI? Automat-it’s AWS and GenAI expertise can make this possible and unlock better data enrichment and more comprehensive results.
