
LESSONS FROM THE ROAD | Trip Reflections
Last month, I spent time travelling for Kry10, visiting clients, partners, investors, staff, and attending conferences across Japan, the US and UK. With that much distance covered, there were plenty of conversations. Here’s three key lessons that I took away from them:
1. ONE YEAR MAKES A BIG DIFFERENCE
Conversations felt very different this year across all Kry10’s connections. Here’s why:
a) The acceptance of the “The Big Problem”
Cyber vulnerability is genuinely starting to worry people. The ‘Mythos Moment’ has bought it into focus. Those I spoke to knew and understood the problem. They also understood that the solution to AI finding increased vulnerabilities is not to “throw more AI at it”, it’s something deeper than that. Formal methods is suddenly part of every conversation.
b) The shift in knowledge about Formal Methods
It felt like formal methods was all the rage. The majority of those I spoke with had heard of it, many in the context of being able to solve The Big Problem. My colleague Ihor Kuz reflected on the evolution of the formal methods conversation at the High Confidence Software and Systems (HCSS) Conference in Annapolis, MD this way,
When AI and formal methods were spoken about together in 2024, they were just crazy ideas. Last year there were experiments with AI and formal methods. This year, we’re seeing real, applicable improvements, beyond the proof of concept stage. There’s a genuine buzz that this could actually happen.
Suddenly the level of understanding has shifted, enabling us to jump straight to talking about how Kry10 can help solve that Big Problem.
2. GOVERNMENTS WANT GUARANTEES AND PEACE OF MIND
Many people asked me “what's the minimum amount of formal methods I need to ‘sprinkle on my code’ to be safe?” That’s not how it works! The resulting discussions had me describing formal methods integration into existing systems as “lifting up the whole house that's built on sand, pulling out the old foundations and replacing them with new ones”. People got that it’s doable, just not necessarily simple or straight-forward. With this clarified, we were then able to progress to discussing how Kry10 and KOS can be applied in their specific circumstances.
3. AIxFM: COMMUNITIES LEARNING TO WORK TOGETHER
At the HCSS Conference, it was fascinating to see the different dynamics between the AI and formal methods communities. AI teams are driven by urgency, trying to get things out as quickly as they can. In contrast, the formal methods community moves slower, in line with their academic and government grounding. The AI and FM communities need each other. FM can resolve AI’s safety issue. AI resolves FM’s time and cost issue.
I’ll be watching with interest as priorities align between the AI and FM communities, and Kry10 plans to be actively engaged in supporting this alignment.
Are these the same conversations you’re hearing as well? If there's anything that Boyd has mentioned above that makes you curious let us know.