Claude powered Realtime, Hyper-personalized Marketing Collateral Generation – Solarplexus
Business Problem
About Solarplexus
SolarPlexus is a Sweden based SaaS Marketing solutions provider. Their product helps customers 10X their Marketing outreach conversion by building real-time, hyper-personalized marketing content, leveraging underlying user persona & intelligence.
Solution
With the advent of Generative AI & the capabilities to generate content on the fly, customized to the prompts & data we can generate, GoML’s GenAI consulting team worked with the customer to build a product vision, which allows their clients to build hyper-personalized communication for end customers, leveraging user persona real time, resulting 10X increased conversion. The solution followed a 3-step approach.
Automated Branding for Collateral Standardization (Claude-v2)
Leverage Claude-v2 to extract Branding guidelines from complex brand documents, such as the logo, font, brand colors etc., for standardizing the content & formatting for the collaterals
ML-Powered User Segmentation: Personalization in Real-Time
Leverage ML models for real-time, intelligent user segmentation, based on user parameters, such as demographics, buying behavior and other marketing captured content
Hyper-Personalization with Stable Diffusion
Leverage the user segmented data to power Stable Diffusion to generate real-time, image & text based, marketing collateral, which is hyper-personalized to each targeted user
Architecture
Data Extraction:
- Upload and Store Documents: Users uploads brand guidelines, tone of voice and picture bank which are stored in SuperBase bucket, ensuring centralized data access.
- Structured Data Extraction: Utilizing RAG based approach to fetch the details from the documents uploaded by creating vector embeddings and storing vectors in FAISS Vector DB to extract structured data from unstructured documents.
- Similarity Search: Perform similarity search using stored vector embeddings to identify key document elements like color, font, and tone.
- Information Storage: Extracted data is stored in a dedicated SuperBase table for easy access and reference.
Segmentation:
- User Data Upload: Users upload Excel files containing target audience information, which is stored in SuperBase table for analysis.
- Clustering Analysis: Utilize clustering algorithms to segment user data based on demographics and behavior.
- Data Storage: Store segmented data and corresponding cluster information within SuperBase tables.
- Cluster Naming: Name clusters based on specific segmentation criteria for easy identification and analysis.
- Segment Identification: Utilize clustering results to divide target audience into distinct segments for personalized marketing.
- Accuracy Assurance: Ensure accuracy and efficiency in segmenting user data to facilitate effective marketing strategies.
Asset Creation:
- Asset Data Collection: Gather user-provided data such as brand logos and extracted information for asset generation.
- DALL·E Utilization: Utilize DALL·E to generate marketing assets based on collected data and user preferences.
- Segment Customization: Customize marketing assets for each segment using segment-specific information and user-provided images.
- Brand Element Integration: Incorporate brand elements like color and font into the asset generation process for consistency.
Process Consistency: Maintain consistency in asset generation process across segments to ensure brand coherence. - Personalized Asset Delivery: Deliver personalized marketing assets tailored to each segment’s characteristics for maximum impact and engagement.
