Software Tree has launched ORM_Skyway, an automation tool that converts relational databases into AI-ready business object models
ORM_Skyway automates the creation of a governed path from enterprise databases to AI agents through the Model Context Protocol (MCP), with verified support for nine major database platforms
A governance-by-construction model ensures only data explicitly mapped by developers is exposed to AI agents, with an option to enforce read-only access
Software Tree, a Campbell, California-based provider of object-relational mapping technology, announced on August 20, 2026 the launch of ORM_Skyway, a new automation product designed to make existing relational databases accessible to AI agents. The tool reverse-engineers a database schema and automatically builds a governed path to AI applications, rather than requiring enterprises to expose raw database structures directly.
From Tables and Joins to Business Objects
Enterprise data typically sits in relational databases organized around tables, columns, keys, and joins, while AI systems work more naturally with business concepts such as customers, orders, and products. ORM_Skyway is built to close that gap. Starting from a database already running on-premises or in the cloud, it generates a business object model and its underlying object-relational mapping, then packages the result as a RESTful microservice before configuring the connection for AI applications through the Model Context Protocol (MCP).
The approach works with the enterprise's existing database in place, without requiring data migration, data duplication, or adoption of a new database platform.
The system draws on three existing Software Tree technologies: JDX for the object-relational mapping layer, Gilhari for exposing business objects as JSON through a Docker-native microservice, and ORMCP for MCP connectivity that lets agents interact with those objects semantically. Once ORM_Skyway has generated and configured the architecture, the company describes its role as complete — it functions as design-time automation rather than an ongoing runtime dependency.
ORM_Skyway architecture: from an existing relational database, through governed business object and REST layers, to an AI agent via ORMCP.
The tool has been verified with PostgreSQL, MySQL, Oracle, SQLite, Microsoft SQL Server, Snowflake, IBM DB2, CockroachDB, and SAP HANA, and works across any JDBC-compliant database, with additional platform support planned.
"For decades, our focus at Software Tree has been to free application developers from the complexity of relational database structures by letting them work naturally with business objects. ORM_Skyway extends that philosophy all the way to AI."
— Damodar Periwal, Founder and CEO, Software Tree
Governance as the Central Pitch
The company frames governance as the core problem ORM_Skyway is built to solve: giving AI systems useful access to operational data without exposing an entire production schema. Developers curate the generated object model to decide which tables, attributes, and relationships are available, and can apply business-friendly naming. Anything left unmapped remains invisible to the AI agent, and ORMCP can additionally enforce read-only access across a deployment.
That design choice reflects a broader anxiety running through enterprise AI deployment right now. Gartner has forecast that up to 40% of enterprise applications will include integrated task-specific agents by 2026, up from less than 5% today, a shift the firm's analysts expect to reshape how software vendors and internal IT teams think about data access and control. Vendors positioning themselves between databases and agents — rather than requiring new data platforms altogether — are betting that enterprises will favor incremental, controllable paths to agentic AI over wholesale data migrations.
Software Tree is pitching ORM_Skyway and ORMCP beyond its own direct customers, describing them as building blocks for AI platform vendors, database providers, integrators, and systems integrators building agentic AI offerings for their own clients. ORM_Skyway is being made available without a license fee under its published terms, while the JDX, Gilhari, and ORMCP components that run the generated architecture carry 30-day trial or beta evaluation licenses.
ORM_Skyway arrives as Software Tree extends decades of object-relational mapping work, with past licensees including British Telecom, Los Alamos National Laboratory, and Darden Business School, into the AI infrastructure market. As enterprises evaluate different approaches to connecting AI agents with operational data, ORM_Skyway offers a governance-by-construction alternative that works with existing relational database infrastructure.
