Episode summary
Why AI Can Write Code But Can’t Keep It Running (Yet)
The a16z Show
· · 59 min · Hosts: Martin Casado · Guests: Jad Naous
Summary
Jad Naous argues that AI has already changed how small engineering teams build software, but that keeping systems running is a different and much harder problem. He says coding agents can now take work that once took months and compress it into days, especially when a senior engineer knows how to break problems into clear steps, review design docs, and manage multiple agent sessions at once.
The conversation centers on why observability and operations are harder than code generation. Jad says Grepr was built first to cut telemetry costs by finding the signal in massive volumes of logs, then to feed those signals into AI systems that can troubleshoot incidents. He distinguishes between the model predicting an outage and the system detecting that something changed, then using AI to form hypotheses and run experiments.
The host remains skeptical that AI can handle the wild, exception-heavy state space of production systems the way it handles code. Jad counters that the key is not perfect prediction but a workflow that reasons from real signals, benchmarks, and production experiments. Both agree that security is even harder than operations because it adds an adversary, and that the next frontier after code may be constrained domains where the world can be instrumented, modeled, and gradually made more machine-readable.
Key takeaways
- Jad Naous says a 10-person team can now ship some production projects in about 10 days that used to take roughly 6 months.
- Jad argues that good engineers get disproportionately better with AI because they know how to structure work, prompts, and review loops.
- The host says observability budgets often run at 5% to 20% of infrastructure spend because companies are really buying reliability.
- Jad says Grepr’s approach is to detect changes in logs first, then let AI reason through causal hypotheses instead of backtracking from a late alert.
- Jad believes AI is very close to handling query optimization, but not yet able to run a distributed cluster end to end.
- The host remains skeptical that operational systems are as learnable as code because production state spaces are exception-heavy and often computationally irreducible.
Notable quotes
“the more complexity that your brain can handle, the more you can break down the tasks and kind of assign them”
“the better engineers like are much better”
“that took maybe about 10 days. In previous days that would have taken maybe about 6 months”
“the goal is not actually having the uh agent try to predict but have it uh get a signal that something changed”
“I think very close”
How the episode opens
The world that the AI lives in is all digital and there's the actual world and like those things are not necessarily the same. — When we started doing AI for coding, it was autocomplete. Nobody really knew to what extent you can actually build complex software using AI and then it basically broke through the expectations because we were able to model more and more complexity in the systems.
I just don't know if operations is constrained in the same way. Where's even the data to learn how to like build a system that understands like all of the operations? My assumption is that we have all the data.
The question I think is maybe about complexity and like to what extent can you take the things that you've described that are operational and convert them into something that an AI can reason about the dirty secret for the longest time of running like a distributed cluster with just human beings managing it. Do you think we're at a spot where you could like use AI to run one of these services? I think that there's enough to be able to All right, Jad.
thanks for joining me on the podcast. For those of you who don't know, Jaden is the CEO of Greer, but also we uh went to grad school in the same group. And when I first met you, I we we had a new student.
I I walked into the office and there was a motorcycle helmet and a pack of Maros. I'm like, "This guy must be unbelievably cool." So my question is is what happened to all your coolness? Question number one, — Maybe maybe maybe just for a context, maybe give give um a quick um uh overview of your background.
From the episode’s captions: roughly the first two minutes, lightly cleaned. Keep watching on YouTube →
Topics
- ai coding agents
- production observability
- debugging operations
- incident detection
- security for ai systems
Companies and tickers mentioned
- Grepr
- a16z
Watch the episode
Watch “Why AI Can Write Code But Can’t Keep It Running (Yet)” on YouTube
Summary, takeaways and quote selection by Octagon. Quotes are short excerpts; the full conversation is in the episode.
More from The a16z Show
-
How AI Could Bring Halo and Other Classic Games Back to Life -
Mike Volpi on Why AI Infrastructure Will Outlast the Hype Cycle -
Why HubSpot’s Co-Founder Is Betting Against the SaaSpocalypse -
Inside Salesforce’s Acquisition Strategy for the AI Era -
How AI Is Turning Figma’s Finance Team Into Builders -
The New AI Agents That Shop, Plan, and Gossip
The a16z Show, this episode and its artwork belong to Andreessen Horowitz. Octagon is not affiliated with the show; this page is an independent summary.
