You’ve probably been there. You start a side project to learn something new, full of optimism. You write the first few lines of code, hit a wall of incomprehensible documentation, and quietly abandon it for two years.
That’s exactly what happened to one software engineer who wanted to build a BGP (Border Gateway Protocol) speaker from scratch in Rust. BGP is the protocol that literally runs the internet. It’s dense, unforgiving, and typically requires years of networking experience to master. The developer hit a wall at the message parser. The knowledge gap was too vast. The project died.
Until last year, when they resurrected it. But this time, they didn’t face the RFCs alone. They brought AI.
By using AI as a bridge over their knowledge gaps, the developer didn’t just parse the messages—they built a fully functional BGP speaker and ran a live Autonomous System (AS) on the DN42 network. It’s a triumphant story of persistence. But it also exposes a massive, uncomfortable shift happening in engineering right now.
We are moving from an era of ‘learn first, build second’ to an era of ‘build first, understand later.’
Think about the paradox here. BGP is fundamentally about explicit, human-designed rules and deterministic behavior. It is the opposite of a black box. Yet, the tool that finally unlocked this developer’s ability to build it—AI—is the ultimate black box. We are using probabilistic engines to construct deterministic infrastructure.
This is giving rise to a new class of ’emergent engineers.’ These are builders who don’t have the deep, formal expertise of the veterans who built the early internet. They don’t spend months memorizing RFCs. Instead, they use AI to bypass the learning curve and ship working systems at lightning speed.
Is this dangerous? Absolutely. When you build by relying on an AI to explain the parts you don’t understand, you risk missing the nuanced, hard-won wisdom that comes from manual debugging and failing in the dark. When the emergent engineer’s system breaks in a way the AI hasn’t seen, they might not know how to fix it.
Tools don’t replace understanding, but they absolutely destroy the learning curve.
But let’s be real: the gatekeepers of low-level networking can no longer hide behind the excuse that ‘this takes years to learn.’ The barrier to entry has been shattered. If you have the persistence to keep pushing, and the intelligence to ask the right questions, you can build the internet’s backbone from your bedroom.
The future of infrastructure isn’t just being written by the old guard anymore. It’s being generated by a new wave of builders who refuse to let a knowledge gap stand between them and a working system.
FAQ
Q: Isn't building critical network infrastructure with AI dangerous if you don't fully understand it?
A: Yes, it is inherently risky. AI allows you to bypass the deep, manual debugging experience that teaches you how systems fail. You get a working system faster, but you might lack the expertise to fix it when it breaks in unprecedented ways.
Q: What's the practical takeaway for developers today?
A: Stop using lack of formal expertise as an excuse not to build. You no longer need to read a 1000-page RFC cover to cover before writing your first line of code. Use AI to navigate the knowledge gaps and ship functional prototypes.
Q: Does this mean traditional network engineering is dead?
A: Not dead, but disrupted. The barrier to entry is destroyed. The 'emergent engineers' who use AI to build systems quickly will outpace those who insist on doing everything manually. However, the deep expertise of veterans will still be required for the most complex, edge-case debugging.