A landmark release from OpenLink Software that converges 30 years of enterprise database excellence with a full AI agent runtime — spanning multi-LLM orchestration via OPAL, machine-to-machine payments, passkey-secured identity, and enriched SPARQL capabilities.
The most substantial additions in 08.03.3335 reside within OPAL, which bridges enterprise knowledge graphs with modern Large Language Models (LLMs).
Core administrative functions now exposed as Model Context Protocol (MCP) tools — attach ODBC/JDBC data sources, create RDF Views, and deploy Linked Data entirely via natural language conversation, replacing ISQL and the HTML Conductor for routine tasks.
Built-in support for Gemini native API (file search, batch processing, image generation, Gemma model), GPT-5 attachment capabilities, and Claude function calling. Web Fetch tools tailored for OpenAI and Claude included.
A dedicated Agent class with tracking capabilities alongside OPAL Skills and skill tool rendering — enabling composable, reusable AI agent workflows grounded in the enterprise knowledge graph.
Vector store MCP tools enable embedding management within Virtuoso for hybrid SPARQL + semantic search. New WebMCP tools page and a direct Web Fetch tool for AI model pipelines.
Native support for uploading blobs (up to 10 MB) directly to the AI layer — enabling document and media ingestion into AI processing pipelines without external preprocessing.
Updated security and user management preparing Virtuoso deployments for a passwordless, monetization-ready web and agentic AI economy.
Modern Passkey authentication deployed across both the Virtuoso Authentication Layer (VAL) and OpenLink Data Spaces (ODS) — eliminating passwords and enabling phishing-resistant, device-bound authentication for users and AI agents alike.
A new 402 Payment Required handler over JSON-RPC with callback hooks for access-gated SPARQL endpoints and resources. AI agents can autonomously initiate micropayments for data access — the foundation of a machine-driven data economy.
ODS platform now supports reCAPTCHA verification during registration alongside third-party identity provider binding — improving onboarding security and enabling federated identity for enterprise AI deployments.
Under-the-hood modernization for better data sharing, networking efficiency, and cryptographic integrity in AI-era deployments.
Support for the Delta-Sharing open protocol to simplify secure data sharing with external applications — enabling AI training pipelines at partner organizations to consume live Virtuoso datasets without data duplication.
Replaced older select() patterns with modern poll() support, plus native getaddrinfo() and getnameinfo() — improving network scalability under high-concurrency AI workloads.
A dedicated JSON canonicalization function supporting JSON Web Signatures (JWS) and modern web standards — enabling cryptographically verifiable data payloads for AI agent trust chains and signed knowledge graph attestations.
Heavily optimized query processing and data transformation pipelines with new functional modules for AI knowledge graph reasoning.
Support for the wikibase:mwapi SERVICE handler within the SPARQL query processor — enabling federated queries against Wikidata and custom Wikibase instances, enriching enterprise knowledge graphs with open global data.
Virtuoso can now automatically generate a VoID (Vocabulary of Interlinked Datasets) graph for any specified graph group — enabling AI agents to autonomously discover and understand available datasets without prior configuration.
Enhanced RDF transactional replication now supports multiple publications — enabling distributed knowledge graph deployment and high-availability AI data infrastructure across enterprise environments.
The Sponger data transformation middleware gains multiple task types optimized for LLM batch-processing integration — applying AI models as enrichment steps within the web-to-RDF transformation pipeline itself.
12 key questions about Virtuoso 08.03.3335 and its value proposition in the age of AI.
Key concepts and technologies in the Virtuoso 08.03.3335 release, explained for the AI age.
Seven steps to deploying Virtuoso 08.03.3335 as the data substrate for enterprise AI applications.
Install on your chosen environment: Linux/macOS/Windows on-premise, Docker container, or Cloud Edition on AWS, Azure, or GCP. Before overwriting older binaries, ensure the database is cleanly shut down and the transaction log (virtuoso.trx) is cleared using +checkpoint-only. Full instructions in the Virtuoso Database Upgrade Notes.
Configure the OpenLink AI Layer (OPAL) to connect your preferred LLM providers. Add API keys for Google Gemini (file search, batch processing, image generation, Gemma), OpenAI GPT-5 (attachment capabilities), and Anthropic Claude (function calling). OPAL manages provider routing automatically.
Register Virtuoso's MCP server with your preferred AI agent client. Once connected, attach ODBC and JDBC data sources, create RDF Views, and manage Linked Data deployments using plain English — eliminating ISQL and HTML Conductor for routine administration.
Use OPAL's vector store MCP tools to create embedding stores within Virtuoso. Connect your embedding pipeline to generate vectors for knowledge graph entities. This enables hybrid queries combining SPARQL graph traversal with vector similarity search — delivering both precision and semantic recall for RAG applications.
Enable the MPP/x402 handler to gate access to high-value SPARQL endpoints and data resources. Configure 402 Payment Required responses with callback hooks. AI agents querying gated endpoints will automatically initiate micropayment flows — monetizing enterprise data assets in the machine-driven data economy.
Configure Virtuoso's Delta-Sharing protocol support to publish datasets to partner AI pipelines. External AI training and inference workloads consume live Virtuoso data without requiring full copies, maintaining data sovereignty while enabling federated AI collaboration across organizational boundaries.
Enable the wikibase:mwapi SERVICE handler to federate SPARQL queries against Wikidata. Use VoID graph auto-generation to expose your dataset catalog to AI agents. Configure Sponger LLM Meta Processing tasks to automate web content ingestion and transformation into grounded RDF facts.
text/x-html+tr · CONSTRUCT/DESCRIBE → text/x-html-nice-turtle · Endpoint: linkeddata.uriburner.com/sparqlPREFIX schema: <http://schema.org/>
SELECT DISTINCT ?feature ?name ?desc
FROM <https://linkeddata.uriburner.com/DAV/demos/daas/virtuoso-08-03-3335-ai-release-claude_sonnet_4_6-1.ttl>
WHERE {
?feature a schema:SoftwareApplication ;
schema:name ?name ;
schema:description ?desc .
}
ORDER BY ?name
text/x-html+tr
PREFIX schema: <http://schema.org/>
SELECT ?q ?question ?answer
FROM <https://linkeddata.uriburner.com/DAV/demos/daas/virtuoso-08-03-3335-ai-release-claude_sonnet_4_6-1.ttl>
WHERE {
?q a schema:Question ;
schema:name ?question ;
schema:acceptedAnswer ?ans .
?ans schema:text ?answer .
}
ORDER BY ?question
text/x-html+tr
PREFIX schema: <http://schema.org/>
SELECT ?term ?name ?def
FROM <https://linkeddata.uriburner.com/DAV/demos/daas/virtuoso-08-03-3335-ai-release-claude_sonnet_4_6-1.ttl>
WHERE {
?term a schema:DefinedTerm ;
schema:name ?name ;
schema:description ?def .
}
ORDER BY ?name
text/x-html+tr
PREFIX schema: <http://schema.org/>
PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>
SELECT ?step ?pos ?name ?text
FROM <https://linkeddata.uriburner.com/DAV/demos/daas/virtuoso-08-03-3335-ai-release-claude_sonnet_4_6-1.ttl>
WHERE {
?step a schema:HowToStep ;
schema:position ?pos ;
schema:name ?name ;
schema:text ?text .
}
ORDER BY xsd:integer(?pos)
text/x-html+tr