Introducing Sigil Logic

I’ve spent my entire career working on systems where the outside world forces you to do better: certification regimes, federal rules, adversaries, and the simple reality that failure means catastrophic outcomes. Those environments demand both rigorous engineering practice and the kind of mathematical reasoning you get from formal methods.

The problem is the world’s systems have gotten so big and complicated that small teams of “unicorns” (i.e., amazing programmers or hardware engineers that know formal methods) can’t keep up. If you’ve only got these small teams of “Avengers,” and yet the pile of work in front of you is a skyscraper, it ends up being hopeless. Meanwhile, the attack surface expands, and adversaries are thrilled. We’re shipping a planet’s worth of software that we barely understand, and then acting surprised when we get the next “we’re sorry, your data was exposed” letter. And now it is 10x worse with the advent of unfettered agentic coding.

So what do we do?

My view is that the only way forward, both for critical systems and for everyday software, is to make rigor normal: to transition rigorous engineering and formal methods into everyday practice, the way testing and modern dev practices became mainstream over time. But we can’t require everyone to become a PhD. I’ve tried :)

Programmers program. Programmers don’t do math. They don’t even write documentation all that much. And you can’t “retrain the world” by force of will.

This is what we want to change with Sigil Logic.

It is evident to me that we finally have a chance to make rigor feel normal. What we’re doing at Sigil is building a platform that wisely combines Machine Learning (ML, mostly in the form of modern LLMs) with Automated Reasoning (AR, mostly in the form of applied formal methods), but in doing so, hides all of the fancy tools and mathematics that sits behind the agents.

These applied formal methods concepts, tools, and technologies permit coding agents to actually “understand” rigor, and not just imitate mainstream software engineering. The goal is to deliver “secret ninja formal methods”: formal assurance, but packaged as tools that feel like normal software and hardware engineering tools, and not like a research project. You can go fast and know where you’re going and have faith in the quality and security of the results. Because fast without understanding is just running into an electric fence.

If you already operate under a certification regime, you likely know where you want to go, your limiting factors are people, time, and budget. We help turn a team of four into something that behaves like a team of a hundred: keeping methodology and artifacts current, helping roll out practices across repos, and putting an AI formal method engineer in every pull request so rigor scales across the organization.

If you’re not in that world, if you’re a “normal” engineering org, your incentives are different, but the pain is real: reliability, security, usability, time to market, and the cost of shipping something fragile. You shouldn’t need to hire 200 formal methods experts to move the bar. Some smart organizations were ahead of the curve and have done that successfully, at great cost. No one should ever have to do that again.

That’s the bet behind Sigil Logic: raise the quality and security bar for all systems by making rigor practical, integrated, and scalable, so we can build faster without building blind.