1 of 2,009
Why execution matters less in the age of AI
Inbox
| The Deep View <newsletter@thedeepview.co> Unsubscribe | 1:35 PM (6 minutes ago) | ||
to me | |||
September 19, 2026 | Read online  Hello, friends. With AI advancing at a rapid rate, the days of experimentation have given way to scale. As engineers turn from executors of tasks to orchestrators of agents, skills such as good judgement and strategy are more important than ever. That’s the topic of this special weekend edition, presented in partnership with Rokt. Let us know what you think! —Nat Rubio-Licht Though some enterprises are still figuring out their AI strategies, at Rokt, the tech is moving beyond the experimentation phase. Sam Dozor, CTO of ecommerce technology company Rokt, told The Deep View that the days of cautiously using AI as an experimental copilot are long gone, especially for engineers. If engineers aren’t creating and orchestrating teams of agents, they are “no longer somebody who can keep up with the pace of development,” he said. This shift has rocked a lot of traditional builders and engineers who “got by because they were amazing executors,” said Dozor, capable of writing code quickly or debugging live code flawlessly. But AI has decreased the value of those kinds of skills, said Dozor. However, for those who embrace AI tools to their fullest potential, there’s never been a more exciting time to be an engineer, he said. At Rokt, Dozor’s focus has been to help foster the skills that can’t be supplanted: Good judgement, strategy and client empathy. These are the things that represent the biggest bottlenecks at organizations, rather than technical execution. For instance, because AI can handle so many tedious tasks, “we can go directly to training people on systems design and overall product strategy,” he said. “So much of what was very hard and valuable before, things that forced them to crawl through glass, is no longer needed,” Dozor said. And even beyond engineering roles, AI may give organizations the ability to collapse the career ladder into fewer and fewer rungs, he said, democratizing the ability to do anything within an organization while allowing people to be differentiated by the things that they are interested in working on, rather than the technical skills that they’ve been trained on. Dozor has facilitated the shrinking of the career ladder himself: At Rokt, everyone on the engineering team is a “builder,” not differentiated product managers, engineers, or designers. “Some people fell in love with the manual mechanics of building and testing software,” said Dozor. “It’s those with high-level product and commercial aspirations and interests that ultimately will be the winners.”Still, one of the biggest roadblocks to get over has been simply getting people on board. Every company, Rokt included, has its fair share of AI skeptics, said Dozor, feeling the broader societal anxiety around the impacts of AI on the workforce and economy. Going from experimentation to scale required “creating a culture of excitement around it rather than fear,” said Dozor. Additionally, getting beyond the fear required Rokt to accept a “J-curve of productivity,” in which the first few weeks or months of learning how to properly work with AI created “temporary productivity hits across the organization,” but resulted in a sharp uptick in productivity and output once the employees were properly trained. And because AI is still nascent, with true gains in productivity as a result of increased model capability only truly ramping up over the past year, those who were skeptical of the technology are now being faced with the power it puts in their hands, he said. “Getting over that hump of viewing this as something that threatens you to something that actually will accelerate their career is about getting them to realize that, whatever gap they had before — like how to build complex engineering systems, customer understanding or product strategy — AI lets you fill those gaps more than you’ve ever been able to,” said Dozor.Learn more about Rokt |
canisgallicus 