Microsoft announced the preview release of Data Ingestion Building Blocks for .NET, a new modular library designed to help developers build scalable, composable data pipelines for AI and intelligent applications.
With this release, .NET now offers a unified framework for ingesting, processing, enriching, chunking and storing data – whether sourced from files, databases, APIs or cloud services – so it becomes usable for AI workflows such as retrieval-augmented generation (RAG), search, summarization or analytics.
Key capabilities in this preview include:
-
Unified document representation enabling any file type – PDF, Word documents, images, etc. – to be handled consistently for AI-ready processing.
-
Flexible data ingestion from cloud or local sources, with multiple built-in readers for common formats.
-
Built-in AI enrichment such as summarization, sentiment analysis, keyword extraction, classification, and embedding generation, to prepare data for intelligent applications.
-
Customizable chunking strategies – token-based, section-based or semantic-aware – to optimize document segmentation for downstream retrieval or processing.
-
Storage support for processed data including vector stores and document stores, enabling enterprise-grade persistence and scalable retrieval use cases.
-
Composer API that allows chaining readers, processors, chunkers and writers into end-to-end workflows, reducing boilerplate and simplifying pipeline orchestration.
Also Read: OfficeSpace Unveils AI Canvas for AI Space Planning and More Agentic Capabilities Transforming the Human Experience of Work
The Data Ingestion Building Blocks are built on established .NET components – including – ensuring interoperability with existing AI workflows and extensibility for custom connectors and storage back ends.
This release drastically simplifies the effort required to build AI-powered data pipelines within .NET – turning disparate, unstructured sources into structured, enriched data streams ready for analysis, search, or AI-driven tasks.
SOURCE: Microsoft























