logo

Cosdata Roadmap

Detailed timeline and feature status for our journey from MVP to production-ready enterprise solution

Navigation

Timeline

CompletedFeatures we've finished
In ProgressCurrently working on
TodoComing next

Development Timeline

December 15, 2024

MVP/Alpha: Vector Database

  • Optimized HNSW (dense) and Inverted index (sparse) implementations
  • RESTful API for core operations
  • SIMD optimized major distance metrics and quantization
  • Versioning & "Transaction-as-a-resource"
  • Run basic comparison benchmarks for HNSW & Inverted index (SPLADE)
January 30, 2025

MVP/Alpha: Graph Database

  • Basic graph data structures and operations
  • Simple integration with vector search
  • Rudimentary CosQL for graph queries
March 15, 2025

Beta: Graph Database

  • Advanced graph algorithms and knowledge graph features
  • Enhanced CosQL with graph-specific operations
  • Basic rule evaluation engine
June 15, 2025

RC/GA: Graph Database

  • Full graph database capabilities
  • Seamless integration of graph and vector search
  • Advanced knowledge graph operations and querying
June 15, 2025

Beta: Cloud Services

  • Multi-cloud support and improved resource management
  • Enhanced monitoring and basic serverless functions
  • Development of comprehensive web application
  • Initial integration with major cloud ecosystems
August 30, 2025

RC/GA: Cloud Services

  • Fully automated deployment and scaling
  • Production-ready web application with full feature set
  • Comprehensive management and analytics interface
  • High availability and redundancy features
  • Complete integration with major cloud ecosystems

Feature Status

Indexing and Search

FeatureStatusPhase
HNSW indexing for dense vectors with high dimensionality supportCompletedMVP/ALPHA
Inverted Index for sparse vectors (Splade & BM25), supporting very high dimensionalityCompletedMVP/ALPHA
ANN probabilistic search for Inverted IndexCompletedMVP/ALPHA
Benchmarking Inverted Index against proprietary data type offeringsIn ProgressBETA
Optimized hybrid search algorithmsTodoMVP/ALPHA
Advanced indexing optimizationsTodoBETA
Complete end-to-end comparison benchmarking of HNSW & Inverted IndexTodoBETA
Implement re-ranker integrationTodoRC/GA

Distance Metrics and Quantization

FeatureStatusPhase
Dot productCompletedMVP/ALPHA
Cosine SimilarityCompletedMVP/ALPHA
EuclideanCompletedMVP/ALPHA
HammingTodoMVP/ALPHA
SIMD optimizations for cosine & dot product metricsCompletedMVP/ALPHA
Binary (base 2) quantizationCompletedMVP/ALPHA
Quaternary (base 4) quantizationCompletedMVP/ALPHA
Octal (base 8) quantizationCompletedMVP/ALPHA
U8 (base 256) quantizationCompletedMVP/ALPHA
Sub-Byte Quantization of Inverted IndexIn ProgressBETA
SIMD optimizations for all quantization methodsIn ProgressRC/GA
Implementing auto-configuration for optimal quantization and storage based on statistical samplingIn ProgressBETA

Storage and Performance

FeatureStatusPhase
Buffered IO, equivalent to memory mapped files for efficient cachingCompletedMVP/ALPHA
Custom storage layer with serialization of index and corresponding file formatsCompletedMVP/ALPHA
Lazy Loading of index nodes, fulfilling DiskANN requirements for low memory useCompletedMVP/ALPHA
LRU cache for lazy loaded itemsCompletedMVP/ALPHA
Separation of compute & storage architectureCompletedMVP/ALPHA
Advanced caching strategiesTodoBETA
Distributed storage supportTodoRC/GA
Implement advanced sharding for multi-billion scale datasetsTodoRC/GA
Enhance high availability and redundancy featuresTodoRC/GA

Data Management and Versioning

FeatureStatusPhase
Versioning with transaction-based historical revisions and branchingCompletedMVP/ALPHA
Lazy loadable collections (Set, Map, Vec, Array, EagerLazyLoad, etc)CompletedMVP/ALPHA
Auto creation of indexesCompletedMVP/ALPHA
Advanced versioning features, like branching & related APIsTodoBETA
Improve usability of versioning systemTodoBETA
Multi-modal data supportTodoRC/GA
Add native support for storing documents and multi-modal data typesTodoRC/GA

Query and API

FeatureStatusPhase
RESTful API (upsert, ANN, collection create, create index)CompletedMVP/ALPHA
Developing user-facing RESTful API for Inverted IndexIn ProgressBETA
Integrating HNSW hyperparameters APIIn ProgressBETA
GraphQL API supportTodoRC/GA
Implement metadata filteringTodoBETA

Graph Database and Knowledge Graph

FeatureStatusPhase
Cos Query Language (CosQL) specificationCompletedMVP/ALPHA
Rule, Fact, Schema parser for data definition, manipulation & queryingCompletedMVP/ALPHA
Rule evaluation engine (detailed design document created)CompletedMVP/ALPHA
Enhanced CosQL featuresTodoBETA
Enhance graph database rule evaluation engine and improve performanceTodoBETA
Integrate LLM/model for natural language querying of knowledge graphs and relational dataTodoRC/GA
Implement Agentic Memory capabilitiesTodoRC/GA

Cloud Integration and Web Application

FeatureStatusPhase
Prototype web-based management interfaceTodoMVP/ALPHA
Begin development of comprehensive web applicationTodoBETA
Implement basic serverless functionsTodoBETA
Integrate with major cloud ecosystems (initial)TodoBETA
Release production-ready web applicationTodoRC/GA
Implement advanced serverless functionsTodoRC/GA
Fully integrate with major cloud ecosystemsTodoRC/GA
Develop comprehensive analytics features in web applicationTodoRC/GA

Integration and Ecosystem

FeatureStatusPhase
Integrate with major text and image vectorization modelsTodoRC/GA
Integrate with LangChain, LlamaIndex, and similar frameworksTodoRC/GA
Develop web application and cloud serverless integration with major ecosystemsTodoRC/GA

Security and Access Control

FeatureStatusPhase
Develop authentication and IAM user roles for filtering/joining HNSW and Inverted indexesTodoRC/GA

Ongoing Improvements

FeatureStatusPhase
Ongoing bug fixes and performance improvementsIn ProgressALL PHASES