We wrote about why general-purpose LLMs fail at protocol-level troubleshooting, what network-aware AI actually requires, and how our team is solving this every day. supertrace.ai/blog/llms-dont…#NetworkOps#AI#BGP
Making AI useful for BGP is not a prompting problem. It requires live state, topological reasoning, temporal memory, and policy awareness. We call this the context problem: giving the model the same situational awareness a senior engineer would have.
Ask a general-purpose LLM to diagnose a BGP route leak and it will give you a confident, well-structured answer that would almost certainly make the problem worse. It knows the theory, but it has never seen your routing table, your topology, or your policies.
We’re excited to welcome Andrei Shevtsov as Head of Engineering.
With 10+ years building and scaling distributed systems, he’s the right leader for what’s ahead.
Supertrace AI is building the first AI Network Operations Center and we’re just getting started.
Welcome, Andrei!
Modern infrastructure failures do not respect org charts.
A 500 error could be a thread pool. It could also be a misconfigured optical interface three miles away.
SREs live above the network line. NOC engineers live below it.
Your dashboard tells you the link is degraded.
It does not tell you what to do about it.
That answer lives in a runbook from two years ago, a Slack thread from last time, and the head of your most senior engineer. None of it is in your observability tool.