Tobias Ludwig

About the Speaker

Dr. Tobias Ludwig is CEO and co-founder of LUBIS EDA, where he leads the company’s formal verification methodology and the development of the FormalOS infrastructure.
Tobias has dedicated his career to making formal verification more practical, predictable, and scalable for semiconductor engineering teams by transforming expert knowledge into structured, repeatable engineering practice. Under his technical leadership, LUBIS EDA has become one of the industry’s leading independent formal verification specialists, helping organizations worldwide reduce verification risk, improve sign-off confidence, and achieve predictable tapeouts.
Tobias holds a Ph.D. in Electronic Design Automation from RPTU Kaiserslautern-Landau and continues to advance the practice of formal verification through academic engagement and industry collaboration.

From Craft to System: A Platform Approach to Structured and Automated Formal Verification

Overview

Demand for new chips is outpacing the industry’s capacity to verify them. Formal verification is a powerful guard against costly bug escapes, yet it remains a major effort sink, and projects still slip unpredictably toward sign-off. A closer look reveals why: alongside the expert work only domain specialists can do, sits a persistent layer of orchestration overhead, from extracting design information to updating verification plans and maintaining tool-specific scripts. This overhead is well-scoped and low-risk, precisely where AI can deliver value safely. But naive adoption fails, because formal verification demands the determinism and consistency that unguarded AI agents undermine.
The answer is a platform approach. This talk presents an AI-first platform built at LUBIS EDA that grounds automation within well-defined guardrails, making process discipline the default rather than a deliberate investment. The platform acts as a central cockpit for AI- and tool-based automations. It loads the design into the formal environment, extracts key I/Os and behaviors with generated descriptions, populates the verification plan, and maintains the scripts formal checks depend on, all feeding an integrated project view with no manual reporting overhead. The result is a clean separation of concerns: engineers focus on the work only they can do, while the platform owns everything else.

Key Points

• The expert judgment only specialists can provide is distinct from the orchestration overhead around it, and that overhead is where AI belongs.
• Unbounded agents break the determinism formal verification depends on. A well-defined process with discipline enforces the consistency that raw automation can’t.
• A platform approach makes discipline the default. By owning the plumbing from a central cockpit, it frees engineers to focus on the work only they can do, and scales verification without scaling headcount.