Source
How Malachyte solves retail’s cold-start problem with managed real-time AI
Malachyte developed an AI-powered ecommerce platform that uses attention-based neural networks to solve the "cold start" problem of recommending products to unknown users. The system leverages Google Cloud's real-time AI stack to update user preferences based on session interaction sequences.
Why it matters
Retailers can increase sales by offering personalization to new visitors without needing long-term cookie data. This enhances shopper privacy while providing more relevant product suggestions in real time.
The details
- Some retailers have doubled or tripled their sales using the platform.
- User profiles and predictions are updated every 100 milliseconds.
- The infrastructure uses Google Cloud's Bigtable, Kafka, and Pub/Sub services.
Show entities and relationshipsHide entities and relationships
In this article
Key connections
Google owns Managed Service for Apache Kafka
Google provides Managed Service for Apache Kafka.
Malachyte uses Bigtable to store and quickly update user vector profiles.
Malachyte uses Managed Service for Apache Kafka
Malachyte streams real-time behavioral events through Managed Service for Apache Kafka.
Malachyte uses Pub/Sub for continuous ingestion of product catalog and inventory updates.
Malachyte uses Google Kubernetes Engine
Malachyte runs its agents on Google Kubernetes Engine.
Malachyte uses Google Compute Engine
Malachyte performs model inference on Google Compute Engine.
Show 5 more connectionsShow fewer connections
Malachyte uses Google Cloud
Malachyte built its real-time recommendation system on Google Cloud.
Malachyte uses Large Language Models
Malachyte applies transformer-based concepts from large language models to user interaction sequences.
Malachyte uses Transformers
Malachyte uses attention-based neural networks inspired by transformer architecture.
Malachyte is related to Collaborative Filtering
Malachyte's sequence-based personalization advances beyond traditional collaborative filtering.
Google is related to Transformers
Google pioneered research and development on transformer architecture.
Get the weekly recap
The stories like this one, picked and explained — once a week, straight to your inbox.