Team Reference Code: 4D
Inspiration
Companies or brands that sell their products online receive a large number of customer reviews and feedback. Customer feedback can be a valuable tool to improve the product and curate effective marketing campaigns that would greatly improve their product sales. However, the huge volume of feedback makes it difficult for humans to go through and draw overall insights about customer's perception of the product, features that are favourable and features that are not. Strategify.ai (tailored for the case of a women's clothing company right now), can do this hefty task for all kinds of businesses.
What it does
Strategify.ai is an SaaS web solution designed for businesses looking to harness the power of customer feedback to enhance their products and marketing strategies. With a focus on delivering actionable insights through advanced analytics and AI-driven tools, Strategify.ai is the ultimate partner for companies aiming to boost their sales and stay ahead in competitive markets. Whether it's improving product quality or crafting compelling marketing campaigns, Strategify.ai equips businesses with the tools they need to succeed.
Core Functionalities
- Sentiment Analysis and text categorization (to obtain feedback analytics) - PaLM 2 Model
- Text content generation (to obtain comprehensive marketing strategy) - PaLM 2 Model
- Image generation (to obtain social media marketing post and product inspiration) - Stable Diffusion V3 and XL Models
How is GenAI useful here
Pretrained Large Language Models are exceptionally good at understanding language (including informal texting language), logical relationships and meaning in text, as well as overall sentiments. Hence, we used instruction fine tuning on Google’s PaLM 2 language model to deduce categorical information from each customer review along 6 key parameters prevalent for our use case: quality, style, fit, delivery time, customer service, and pricing. We use these deduced parameters to obtain advanced analytics for a huge volume of customer reviews to understand the overall sentiment about the product. We use the insights from these analytics to obtain comprehensive AI-generated suggestions for marketing campaigns and product improvement ideas. Moreover, we have performed prompt engineering and instruction fine tuning on Stable Diffusion Text-to-Image Model (V3 and XL) to obtain curated social media campaign posters and product improvement ideas, based on the overall customer feedback perception.
Built With
- Streamlit
- Node.js
- MongoDB
- PaLM 2
- Stable Diffusion V3
- Stable Diffusion XL
Challenges we ran into
- Finding a suitable dataset for our use case
- Instruction fine-tuning and tuning model temperature to produce reliable results
- Integrating with mongoDB cloud database
- Hosting frontend and backend applications
Accomplishments that we're proud of
- Image generation giving useful results
- Generating analytics from raw text reviews along multiple parameters
What we learned
Working with cutting edge GenAI, building and deploying a full-stack application
What's next for Strategify.ai
Expanding to other product categories such as electronics, F&B, beauty and hygiene products, home appliances etc.
Log in or sign up for Devpost to join the conversation.