CocoIndex V1 Released: AI agent data updates only the changed parts, pip install runs directly

CryptoWorld News: CocoIndex open-source project releases V1, featuring a complete rewrite of the engine.
The core selling point of this version is incremental processing, where AI agents can convert data sources such as code repositories, documents, meeting notes, PDFs, and more into usable structured context.
When documents are modified, only the affected parts are recalculated, and when source files are deleted, the corresponding database records are automatically cleaned up.
V1 also removes the project’s own configuration language; data processing workflows can be written using plain Python functions, supporting direct breakpoint debugging.
The engine automatically tracks each function’s input and code changes in the background, only re-executing the parts that have truly changed.
Additionally, the forced dependency on PostgreSQL has been removed, allowing users to run the system immediately after installation without setting up a database.
The underlying engine is still implemented in Rust and follows the Apache 2.0 license.
Data can be exported to over ten targets, including PostgreSQL, LanceDB, Neo4j, SQLite, Kafka, S3, and more.
The GitHub repository provides over 20 ready-made examples covering scenarios such as text vectorization, PDF conversion, code indexing, and transforming conversations into knowledge graphs.