Aeca

The Bold Ideas to Address the Simulacres of Problems

Aeca is a new kind of OLTP database with high performance in mind for the modern AI-era. It's a multi-purpose database system that provides both key-value and document-oriented models with transaction support. It also offers full-featured, customizable full-text and vector searches.

What makes Aeca unique is that users can use both full-text search and vector search within a single query, which enables hybrid search and better search quality. This hybrid approach is so much more than just a vector search with filters.

Most importantly, users are not required to install Elasticsearch and/or other database systems to get started with Aeca. Aeca is a self-contained, complete database system that provides built-in full-text and vector search support, which enables easier deployments and simpler software stacks.

We are working hard to ship our work as soon as possible; the first release is coming soon.

Feature Highlights

• C++ for high performance and low latency:
  • Run locally on a laptop like MacBook Air, develop, and deploy to cloud servers
  • The entire English Wikipedia can be stored and indexed in Aeca on MacBook Air for full-text and vector searches
• Basic CRUD operations
• Transactions
• Various types of secondary indexes:
  • Unique and non-unique indexes
  • Clustered and non-clustered indexes
  • Partial indexes
  • Full-text search and vector search indexes
  • Supports multiple full-text and vector search indexes on a collection
• Time-to-Live (TTL) support
• Lua and Python scripting support
• User-defined functions
• Change data capture for streaming data to other systems
• PyTorch integration for model serving
• Fast online incremental backups
• Multi-platform support (Linux, macOS, Windows)
• Apache Arrow for faster data transfers between Aeca server and clients

Use Cases

• General-purpose document-oriented database
  • Supports concepts such as a collection, document, and query languages
• General-purpose key-value database
  • Can be used as a cache, session store, or any kind of state management
• Time-series data management
• Full-text search
• Vector search
  • Natural language understanding, image search, audio and video search, and etc.
• Hybrid search
  • Combines full-text search and vector search in a single query
• Content or product recommendation
• Anomaly detection and fraud detection
• Personalization or user profiling
• Clustering and classification
• Question and answering
• External memory for large language models (LLMs)

The Company

We are building a new kind of OLTP database with high performance in mind for the modern era. With Aeca, you can easily build a performant, robust, and scalable applications with built-in full-text and vector searches support.

Our headquarters is located in the San Francisco Bay Area, California, USA. We are a distributed team so that we can hire the best people not limited by location and time zones.

Our Team

The company was started by a group of people who had previously worked together for many years. We are a group of experts with decades of experience in large-scale data management and fast-growing internet services. The team comprises people who know how to solve real-world problems related to modern data management.

Our Mission

We know that terms like big data, data lakes, and web-scale are fancy and attractive, but those are only everyday issues for very few of us. Most companies will never deal with the petabytes scale of the data.

The real problems we face daily are not fancy; we are implementing features for the users and trying to find a way to keep software stacks simple and moving faster, and so on.

Let's be practical and stay on the ground. Most companies just need a simple but powerful database system to solve real problems. We are here to build a product for most companies, not just for unicorns.

Our mission is to solve the common problems often associated with existing database systems and simplify software development by keeping your software stacks as simple as possible.