A More Powerful, Code-First Knowledge Base Experience on the DigitalOcean Gradient™ AI Platform

Building production-ready retrieval-augmented generation (RAG) systems can be complex, time-consuming, and often requires months of engineering effort. Developers and enterprises struggle to ingest diverse data sources, structure content for semantic search, and maintain accurate, verifiable answers.

Enhancements to DigitalOcean Gradient™ AI Knowledge Bases, now in public preview, are designed to solve this problem. Its code-first feature lets developers create, manage, and query knowledge bases entirely from code, giving full control over ingestion, chunking, embedding, and retrieval without having to worry about the underlying infrastructure.

Flexible, production-ready toolkit

Many existing solutions let developers create a basic knowledge base, but they often struggle to scale, customize, or integrate it into production workflows. The improvements address this by providing a code-first, developer-focused toolkit that handles the full knowledge base lifecycle. Developers can ingest data from files, Dropbox, web crawlers, control chunking and embedding strategies, and run natural language queries that return citation-backed answers with metadata filters. With well-documented APIs and SDKs, these integrations are seamless, letting developers manage everything entirely in code.

What’s new in the public preview

The public preview highlights the essential tools developers need to build and manage knowledge bases effectively:

• Direct API Access: Query knowledge bases directly without needing an agent, giving full control for integration into apps or RAG pipelines.
• Customizable ingestion: Ingest content from supported sources such as files, web crawlers, and Dropbox datasets. Supports structured data, sitemap crawling, and accurate parsing of complex PDFs.
• Flexible chunking and embedding: Choose the chunking strategy that fits your content and select from high-performance embedding models (including a multi-lingual embedding model). Intelligent defaults allow for quick setup. The latest update allows you to modify and update chunking strategies for both new and existing Knowledge Bases via the data source tab on the DO control Panel or via the API providing more flexibility and improved accuracy for existing content.
• Advanced retrieval and citations: Run queries with exact-page citations, metadata filters, and hybrid search.
• Developer-first tooling: Fully code-driven SDK and API functions makes creation and integration seamless.

Get started with the improved Knowledge Base experience today

The improvements are available in public preview. Start building smarter AI applications faster by managing your knowledge bases entirely in code. Explore the API documentation to start experimenting, and see how quickly you can turn your data into actionable, context-rich answers.

To explore the Knowledge Base improvements, enable the public preview on your Feature Preview page in the DigitalOcean Cloud Console. Once you’ve opted in, access will be granted within approximately 10–15 minutes.