SMT Data launches Grace: Open AI for the Mainframe

Open standards-based AI bringing mainframe expertise into modern developer workflows

We built Grace — a domain-specialized AI designed to make deep IBM mainframe expertise more accessible to developers and enterprises.

The name is a tribute to Rear Admiral Grace Hopper, who championed the idea that programming should work in natural language rather than machine code. Grace carries that vision forward — using natural language to unlock deep mainframe knowledge.

Built on decades of hands-on experience with SMF records, z/OS internals, and Performance & Capacity, Grace represents a new approach to unlocking highly specialized knowledge that has traditionally been difficult to access and transfer.

At the core of Grace is the Model Context Protocol (MCP) — an open standard that enables any AI assistant to connect to domain-specific knowledge sources. By building Grace as an MCP server, SMT Data makes it possible for mainframe expertise to be accessed from widely used tools such as ChatGPT, Claude, GitHub Copilot, or any MCP-compatible platform.

This open approach eliminates vendor lock-in and creates a unified bridge between legacy systems and modern AI-driven development environments.

Addressing the mainframe skills gap

The launch comes at a time when the industry faces a growing shortage of mainframe expertise. Critical knowledge of SMF data structures, system internals, and legacy formats is increasingly concentrated among a shrinking pool of specialists.

Grace addresses this challenge by allowing developers to query complex mainframe topics in natural language and receive accurate, contextualized answers in seconds — rather than relying on fragmented documentation or manual research across extensive technical manuals.

Powered by Retrieval-Augmented Generation (RAG) and grounded in curated knowledge sources — including vendor documentation, internal runbooks, and organization-specific materials — Grace helps ensure accuracy and keeps knowledge current without requiring continuous retraining. As with any AI-assisted workflow, outputs benefit from expert review.

From internal tool to customer solution

Originally developed as an internal AI assistant, Grace is being extended to customers through integration with the ITBI Portal.

This integration embeds AI-driven expertise directly into existing workflows, enabling users to:

• Query SMF records, z/OS performance metrics, and capacity planning topics using natural language
• Accelerate performance analysis and reduce time spent cross-referencing technical documentation

By combining domain expertise, AI, and open standards, SMT Data aims to make mainframe knowledge more accessible, scalable, and future-proof.

A broader vision for domain-specific AI

Grace demonstrates a broader pattern: organizations can transform deep, specialized knowledge into shared, AI-accessible resources.

As industries increasingly look to bridge skills gaps and modernize legacy systems, solutions like Grace highlight the potential of open standards and domain-specific AI to democratize expertise.

Future releases will extend Grace beyond knowledge retrieval, enabling AI-assisted querying and analysis directly against customer data within the ITBI Platform — turning Grace from an expert advisor into an active analytical partner.

SMT Data expects to make Grace available in the ITBI Portal soon, with early access offered to selected customers. Reach out to CPTO Nicklas Laine Overgaard or your daily SMT Data contact to learn more or join the early access list.