Hud CEO Roee Adler: Runtime Intelligence Is the Next Stage of Software Operations

AI has sharply accelerated software development: AI coding agents can now generate large volumes of production-ready code in a matter of minutes. But according to Roee Adler, CEO of the startup Hud, writing code faster doesn't solve the central engineering problem: making sure that software actually works correctly once it's in production.

Why the old monitoring tools aren't enough

For decades, observability platforms have helped teams watch their infrastructure through logs, metrics, and traces, the three classic sources of data about a system's health. This approach assumed that humans write code deliberately and relatively slowly, which gives a team time to understand the system's architecture and tailor monitoring to its specifics.

• AI agents already generate large volumes of production-ready code in minutes
• Code-writing speed has gone up, but correct behavior in production remains an unsolved problem
• Classic observability tools rely on logs, metrics, and traces
• The position comes from Roee Adler, CEO of the startup Hud

What changes as AI-driven development grows

When an AI agent, rather than an engineer, writes a large share of the code, new code lands in the system faster than teams can figure out how it will behave under real-world conditions. That opens a gap between how quickly code reaches production and how deeply the team understands its behavior under load. In Adler's view, closing this gap is the job of a new class of tools that work not only with logs and metrics but also with a deeper understanding of what happens while the code runs.

What this means

As AI agents take over more and more code writing, the industry will likely have to rethink its approach to production monitoring, shifting from passively collecting logs and metrics to actively understanding system behavior in real time. That is what Adler means by runtime intelligence.