Corporate

About OpenLink Software

OpenLink Software builds infrastructure for turning data into context—and context into action.

One enduring problem#

For decades, we have focused on one enduring computing challenge: enabling applications, data, services, people, and now AI Agents to interact through open standards without becoming tightly coupled.

Today, that work spans five complementary layers.

Everything rests on access to the Data Spaces you already have.

Data Layer#

Everything starts with access to existing Data Spaces—databases, knowledge bases, filesystems, and APIs.

OpenLink’s standards-compliant ODBC, JDBC, ADO.NET, and OLE DB connectivity provides high-performance access to systems including Virtuoso, Oracle, SQL Server, Db2, Informix, Ingres, Sybase, Progress, MySQL, and PostgreSQL.

Our data virtualization technology extends that reach across heterogeneous and distributed data sources—without requiring unnecessary copying or migration.

Semantic & Context Layer#

Access to data isn’t enough. AI Agents need context—and context needs structure, meaning, and relationships.

Virtuoso provides infrastructure for creating and using Knowledge Graphs that transform heterogeneous data into interconnected, machine-computable entities and relationships, drawing directly on technologies developed through The Semantic Web Project — RDF, Linked Data, Ontologies, SPARQL, reasoning, and inference.

Linked Data principles provide the connective tissue. Globally unambiguous identifiers name entities—people, organizations, products, places, documents, services, Agents, and anything else—while typed relationships explicitly describe how those entities are connected.

The Ontology comprises the TBox and RBox; the enclosing Context Layer adds the ABox. RDF is the model — Turtle and JSON-LD are notations that express it.

Virtuoso brings these layers together with existing Data Spaces—databases, knowledge bases, filesystems, and APIs—without requiring everything to be copied into a new repository. Declarative RDF Views, for example, can expose existing relational data as Knowledge Graphs while leaving the source data in place.

The result is a machine-computable context fabric that humans and AI Agents can discover, traverse, query, reason over, and reuse using natural language as well as standards-based query languages.

This is where OpenLink’s decades of work on The Semantic Web Project, Linked Data, Knowledge Graphs, and Ontologies converge directly with today’s AI Agent requirements.

The LLM handles the ambiguity of natural language.

Knowledge Graphs establish context. Ontologies establish meaning and logic. Virtuoso makes both operational.

Identity & Trust Layer#

Agency without verifiable identity doesn’t scale.

As software moves from responding to requests to acting autonomously or On-Behalf-Of (OBO) others, systems need to establish who—or what—is acting, whose authority it represents, and what it is permitted to do.

OpenLink’s YouID provides cryptographically verifiable identity usable with Public Key Infrastructure (PKI) and Decentralized Public Key Infrastructure (DPKI).

This brings identity back to Internet and Web scale: people, organizations, Agents, services, and other entities can have globally unambiguous identities backed by cryptographic verification rather than identities trapped inside application-specific account silos.

The architecture separates concerns that are too often conflated:

Identity→ Identification→ Authentication→ Authorization→ Resource Access

Cryptographic credentials establish verifiable identity and relationships. Authentication verifies control of those credentials. Fine-grained Attribute-Based Access Controls (ABAC) determine what an authenticated entity may actually do.

The same foundation enables cryptographically verifiable relationships, including On-Behalf-Of delegation—critical when an AI Agent needs to demonstrate that it is acting with the authority of a human, organization, or another Agent.

This provides a trust foundation for an Agentic Web in which identity, delegation, and authorization can operate across organizational and application boundaries.

Agentic Layer#

LLM-powered AI Agents add natural-language interaction and action to this foundation.

OpenLink’s OPAL (OpenLink AI Layer) provides an AI operating environment for developing and using Agents while keeping its major components loosely coupled.

That environment supports:

AGENT.md for Agent harnessing and behavioral configuration
SKILL.md for portable Skills packaging
MCP for standardized interaction with tools and capabilities
Data Spaces comprising databases, knowledge bases, filesystems, and APIs
Virtuoso for semantic context, reasoning, query, governance, and integration
Multiple LLMs, enabling model choice and routing rather than binding Agents to a single model provider
Verifiable identity and delegation, enabling Agents to operate as identifiable actors or On-Behalf-Of others

The architecture deliberately separates:

Agents→ Skills→ Tools→ Data Spaces

Agentsdetermine what needs to be done.

Skillspackage reusable know-how for doing it.

ToolsMCP and other open interfaces expose tools that perform actions.

Data Spacesprovide the data, information, and knowledge upon which those actions operate.

The LLM handles the fuzzy world of natural language. The Semantic & Context Layer supplies machine-computable meaning, relationships, provenance, rules, and deterministic reasoning. The Identity & Trust Layer establishes who is acting, whose authority they represent, and what they are authorized to do.

The underlying Data Spaces remain the systems of record.

Forward-Deployed Engineering Layer#

Powerful technology still has to solve somebody’s actual problem.

OpenLink’s Forward-Deployed Engineering (FDE) approach puts engineers alongside customers and partners to turn requirements, existing infrastructure, and domain expertise into working Agentic solutions.

A central part of that process is Skills development.

Rather than repeatedly embedding domain know-how in bespoke applications, FDE engagements can capture that know-how as reusable SKILL.md packages. Skills can be developed, tested, refined, exchanged, and reused across Agents and compatible AI operating environments.

This creates a practical adoption loop:

Work alongside the customer, and understand the problem.
Build the Skill.
Connect the tools and Data Spaces.
Establish identity and access policies.
Test against real workflows, and refine.
Package the know-how for reuse and exchange.

↺ and back again, each engagement

FDE therefore provides more than implementation services. It supplies the hands-on guidance required to cross the gap between AI experimentation and operational use.

It also creates a mechanism for knowledge transfer. What begins as expertise applied by a forward-deployed engineer can become a portable Skill capturing repeatable know-how for subsequent use by people and Agents.

Over time, this provides the foundation for a broader Skills ecosystem in which reusable know-how can be developed, shared, exchanged, improved, and deployed without tightly coupling it to a particular Agent, LLM, tool, or underlying Data Space.

A History of Building the Layers#

Today’s architecture is the latest stage of work OpenLink has pursued across several foundational areas of computing.

Universal Data Access

Standards-compliant ODBC, JDBC, ADO.NET, and OLE DB connectivity across heterogeneous database systems.

Data Virtualization

Unified interaction with distributed data while leaving it where it belongs.

Multi-Model Database Management

Delivered through Virtuoso, combining relational data, Knowledge Graphs, documents, Web resources, and multiple query languages.

Semantic Web & Linked Data Infrastructure

Applying technologies developed through The Semantic Web Project to turn data into globally identifiable, interconnected, machine-computable knowledge.

Knowledge Graphs & Ontologies

Providing explicit context, semantics, relationships, logic, reasoning, and inference for humans, applications, and AI Agents.

Identity, PKI & DPKI

YouID and related technologies for cryptographically verifiable identity, authentication, delegation, and fine-grained authorization at Internet and Web scale.

Web & Application Middleware

RESTful interaction patterns, HTTP-based services, WebDAV, APIs, and open protocols.

Distributed Collaboration

Weblogs, wikis, feeds, syndication, Web-based filesystems, and discussion systems built around Web architecture.

AI Agents & Skills

An OPAL-based AI operating environment supporting AGENT.md harnessing, SKILL.md Skills packaging, MCP-based tool interaction, and loosely coupled access to Data Spaces.

Forward-Deployed Engineering

Working directly with customers and partners to transform domain expertise and real-world workflows into operational Agent solutions and reusable Skills.

One Architectural Principle#

The common architectural principle has remained remarkably consistent:

Loose coupling through open standards.

ODBC loosely coupled applications and databases. Linked Data loosely coupled entities and relationships across the Web. PKI and DPKI enable identity and trust without requiring every interaction to depend on a centralized application account. MCP loosely couples Agents and Skills from the tools they use. SKILL.md extends the same principle to reusable know-how.

The interfaces evolve. The architectural principle endures.

Natural language has now joined the computing UI/UX stack, while AI Agents are becoming active participants on the Web.

OpenLink’s decades of work across connectivity, virtualization, Knowledge Graphs, Ontologies, semantics, context, identity, security, open protocols, and applied engineering now converge as infrastructure—and know-how—for the emerging Agentic Web.

OpenLink Software is a privately held company, founded in 1992 by its President & CEO, Kingsley Idehen. Mission: enable maximum exploitation of current and future technology paradigms, with minimal disruption to existing technology infrastructure. Vision: empowerment of individuals and enterprises through open access to data. Value proposition: platform-independent access to data, without compromise.

Key contacts and departments#

Get in touch#

We’re here to help you make technology work for you. Contact us to discuss how we can assist your organization.