Introducing Frame

Frame was created to fill a gap in the market: a solution for developer teams that need image search capabilities but want a simple, easy-to-configure deployment without the complexity and overhead of existing enterprise-grade systems. Frame’s design trades features and configurability for simplicity, creating a pragmatic solution that handles the core functionality of an image embedding pipeline (discussed above) without unnecessary complexity.

A comparison of developers’ options when selecting a vector search platform: Frame, Marqo, and Marqo Self-Hosted.

This means Frame isn’t optimized for systems requiring a highly performant enterprise solution or immediate searchability of newly uploaded content (the asynchronous ingestion pipeline means recently uploaded images may take a few moments to become searchable). Instead, Frame excels for teams:

• Implementing search on small-to-moderate image collections or serving limited end users.
• Looking for a simple, low-cost deployment where ultra-low latency is not a concern.
• Prioritizing data sovereignty and AWS integration.
• Building proof-of-concept implementations or internal tools (like reverse image search to detect copyright violations).

Frame is an open-source solution installed directly on the end user’s AWS account via AWS Cloud Development Kit (CDK) and a simple command line workflow. Once deployed, Frame abstracts away the entire embedding and search process behind a simple API. It supports search and recommendation, with minimal setup and a type-safe, developer-friendly Software Development Kit (SDK). With just a few lines of code, users can ingest data and query results, making sophisticated image search more widely accessible. The demonstration videos below show just how easy it is to get started.

Demonstration

Infrastructure Deployment

Frame’s infrastructure can be deployed with just a few commands. This streamlined setup process eliminates complex configuration, getting you from zero to fully deployed in minutes.

Please refer to Frame’s comprehensive deployment documentation for full step-by-step deployment instructions.

SDK Connection

Connecting to your deployed Frame infrastructure is easy with the Frame TypeScript SDK.

With just a few lines of code, establish a secure connection and access six intuitive methods which map directly to Frame’s API endpoints.

Full SDK documentation is available here.

Text-Based Search

Detailed, multi-part text queries receive holistically relevant results because Frame’s embedding technology captures semantic nuances that traditional keyword or manual labeling systems miss. This allows for naturally expressive search that thinks the way humans do.

Image Search In Action

Unlike basic visual matching, our system understands the complete context of images - including objects, relationships, environments, and implicit meanings.

Intelligent Recommendations

Frame can be used to deliver highly relevant recommendations that truly capture the essence of the selected image. The system intelligently surfaces varied yet conceptually similar results—not just visually identical items—creating natural discovery paths that keep users engaged.