Restoring incomplete or partially specified automation programs is a growing challenge, as systems become more modular and dynamic. In this paper, we explore program restoration within the context of IEC 61499, a standard for distributed industrial automation. As a first step toward this broader goal, we focus on automatically generating missing logic between two existing function blocks using Large Language Models (LLMs). Our method synthesizes an intermediate function block based on specifications inferred from the context of the initial blocks and user-supplied metadata, then verifies its behavior through formal model checking. By embedding this workflow into the development cycle without requiring engineers to leave their graphical programming environments, we aim to offer a trustworthy AI co-pilot that assists rather than replaces human developers. A case study demonstrates the feasibility of our approach, highlighting both the promise and limitations of current LLMs in generating semantically correct and verifiable automation logic. This paper lays the groundwork for a more comprehensive restoration of IEC 61499 programs and reinforces the importance of combining AI generation with formal verification techniques.