Shaped is an AI-native personalization platform whose official website describes it as "the only vector database with a feedback loop," used to power personalized search and recommendations. The platform has been acquired by Whatnot.
Purpose Shaped is aimed at scenarios requiring real-time personalization and agent memory, with the goal of building ranking models and personalized experiences. It is positioned as an end-to-end relevance engine for real-time personalization and agent memory, not just document storage.
Key capabilities
- After connecting data, Shaped handles embeddings, models, and data freshness, and can return relevant results in milliseconds.
- Supports adapting a single query to multiple use cases: retrieval can be done via text, user ID, or item ID.
- Unlike traditional vector stores, Shaped can identify the querying user, personalize results per user, and improve with each interaction.
- Supports personalized hybrid search: fusing semantic search with keyword search, and re-ranking based on that user's actual click behavior.
- Provides a unified, queryable relevance engine oriented toward decisions rather than documents.
- Features an intelligence layer that can be used for real-time personalized content feeds, search and discovery, agent retrieval, large-scale content- and behavior-based similarity, personalized emails, context-aware recommendations for conversational AI, and more.
How to use It can be used via the Python SDK, TypeScript SDK, or MCP; specific usage can be found in the official documentation. The platform provides more than 30 native connectors and can ingest raw event logs directly from data warehouses or real-time streams without ETL.
Use cases
- Personalized content feeds: adapt to user behavior in real time.
- Search and discovery: hybrid search combining keyword and semantic relevance.
- Agent retrieval: provide contextual memory and retrieval capabilities for AI agents.
- Similarity: large-scale content- and behavior-based similarity.
- Personalized emails: dynamically generate personalized email content per recipient.
- Conversational AI: provide context-aware recommendations.
Performance data The official website showcases multiple metrics, including 2.5x engagement, +79% D6 retention, +10% watch time, +13% engagement, +382% diversity, 2.2x conversion, +131% watch time, faster experimentation, and reduced time to first experiment.
Limitations and security Enterprise-grade security: SOC 2 Type II certified, GDPR/HIPAA compliant, and offers a 99.95% uptime SLA. The official website states that it has enterprise-gradecapabilities.
Other information The current materials do not provide specific pricing, plans, platform compatibility details, or complete deployment limitations; these contents are not yet available.
