Notable Quotes — From Pipelines to Prompts

⚠️ Speaker attribution is hedged — the source transcript lacks per-speaker labels.

On the magnitude of the AI shift

• "It's making them a hundred times more dramatic than 12 because cloud was sort of limited to dev teams or engineering teams. What we see now is that the moment developers start using AI very quickly they run out of tasks in the backlog." (Section 2)
• "We are living through a seismic shift. We don't know where we're going to end up. We don't know what the earthquake will do to us because the treasures are still going." (Section 3)

On observability and agents in production

• "They will solve an issue in 30 seconds that could have taken developers hours. Or maybe even days to fully understand." — on agents diagnosing production with elastic logs + ServiceNow + code (Section 3)
• "For products, we really need to. Keep the lights up. So that data that complies super important. These days, including agents fix this 24 hours a day." (Section 3)

On harness engineering & linting

• "For me number one really is linting." (Section 4)
• "If you spend about as much time engineering your harness as you do on top table of the actual functional output, you will get really good quality stuff. Genuinely will." (Section 10)
• "This is your hashimoto harness engineer[ing]. Anything that goes wrong, feed it back into the harness and then never let that class of error happen again." (Section 7)

On human review breaking down

• "You cannot continue this humans for every week... if code development goes so much faster, humans cannot keep up with reviewing the code." (Section 4)
• "That's probably not important. But the trade of decisions cost versus speed. Architectural decisions, which kind of observability tool to connect. Those are absolutely has to be done by humans." (Section 4)

On explainability as the new abstraction

• "If we assume that AI is sort of next level of abstraction, then we need to move away from reviewing code and move to that next level of abstraction to new language that describes the system. I don't think that exists." (Section 5)
• "This is like going into assembly and reviewing every line of post." — on the impracticality of reviewing AI-generated code line-by-line (Section 5)

On hallucination

• "We tend to compare systems to perfection. Like it hallucinates a large problem because we expect it to be perfect... It's like self-driving car, right? It's not about zero accidents. It's about less excellence than what we have with students." (Section 6)
• "I think hallucination is going to work. I don't think hallucination problem is going away... because I think it's inherent." (Section 9)

On AI amplifying existing discipline

• "AI just picks up whatever you already have. If you have a good system that doesn't fail. Then you go faster without failing much. If you have terrible system. It's terrible faster." (Section 7)
• "The most advanced teams they explicitly lower the water, expose the underwater stones and then fix them." (Section 7)

On reflector agents

• "We have like reflectors agents like learn from both the problem and the solution and it gets added to the memory as well... That's a new class of agents for prod." (Section 7)

On the future of software development

• "The direction is that the software development is something that's going to explode. Everyone is going to build software... there will be sort of explosion of small software. On the last mile and there will be massively bigger systems that we could never build [before]. And that will be built by professionals software developers not by [vibe coders]." (Section 9)
• "Software developers plus llms are incredibly powerful... it's the two things working together that really really give the power." (Section 9)
• "4GLs were going to take away [software development]. It doesn't happen. This is no different." (Section 9)

On cognitive load

• "One of our customers said actually every few weeks they need to send a person for a week. Home. Just to relax because the pressure goes so fast." (Section 11)
• "It's addictive because you don't know... working late hours because they want to see what they commit are coming now 10 p.m and 11 p.m." (Section 11)

On co-driving (not self-driving)

• "It's co-driving. Very much co-driving but I think the quality is there." (Section 10)