“Every IC engineer is really a front line manager now.” But are they productive?

The Managerial Shift in Engineering

Saying AI has changed the way software engineers work is an understatement, but Cameron Etezadi, CTO of LaunchDarkly and former VP of engineering at IBM, tells The New Stack it goes steps further. He thinks AI has turned all engineers into managers.

“I honestly believe that there is no more IC (Individual Contributor) role anymore. Every IC engineer is really a front-line manager now, even fresh out of school.”

In that regard, he argues the skills needed to be a good engineer today are really just those needed to be a good manager, namely project planning and cross-team coordination - and that companies need to start evaluating engineers’ work the same way they evaluate front-line managers.

His take on the evolving role of the software engineer aligns with new research from Gartner, which predicts that 60% of organizations will have smaller software engineering teams by 2029. How small? Very. Aliyah Camacho, a principal analyst at Gartner, predicts AI will lead to the rise of “tiny teams” comprising as few as two to three engineers.

Productivity vs. Code Volume

Should the software engineering role practically disappear altogether, as Gartner’s research indicates, or morph into another middle management position, then surely AI has dramatically improved engineers’ productivity? But tech leaders aren’t so sure they’re watching the right signals to give an accurate answer. Is more, faster code the same thing as productivity?

AI coding tools are everywhere now, with Big Tech loudly heralding them as the indisputable future. But does helping engineers produce more code faster really translate into greater productivity? Daniel Wang, CTO of Citizen Health and former director of engineering at Uber, tells The New Stack he doesn’t think so:

“Software exists to solve problems. If AI lets us ship 10x as much code but customer outcomes don’t improve, we aren’t more productive.”

Instead, Wang says companies need to shift tracking efforts to decision quality and outcomes, i.e., “everything that happens before and after the code exists.” Specifically, he points to metrics such as:

• Cycle time from idea to production
• Rollback rate
• Escaped defects
• System reliability

At the same time, he encourages leaders to focus on more qualitative questions, like “Did we choose the right solution?” and “Does the code actually solve our customers’ problems?”

What Companies Actually Track

These may seem like obvious signals to watch, yet it’s not where most companies look to assess whether AI is actually improving engineers’ productivity. When asked what, instead, companies are typically tracking, Ameya Kanitkar, founder and CTO of AI measurement platform Larridin and former head of engineering at Coinbase, tells The New Stack it’s largely tangible but irrelevant activity:

“Some version of lines of code, PRs merged, or velocity points, the same metrics teams used before AI showed up.”

Wasted time and effort, apparently, as Kanitkar says these metrics aren’t a real gauge of value, echoing Wang’s sentiment that just because AI can help engineers ship more, faster, it doesn’t mean it’s better. In fact, he argues that living and dying by these metrics could actually steer engineering teams away from truly productive customer outcomes. If teams are rewarded for code output and commit frequency - something AI can easily inflate - then they may end up chasing volume without any real increase in value.

The Exhaustion Paradox

If AI makes engineers more productive, why are they so drained? It can be understandably tempting, however, for engineering teams to focus on outputs, as there’s growing pressure to produce more, more quickly - so much that Kanitkar says some are even turning to AI agents to work for them while they sleep:

“We also see engineers wanting to set agents loose overnight so the work is waiting for them in the morning, which sounds efficient, but it creates this constant background pressure to keep feeding the agents new work so they’re never sitting idle. That pressure is draining.”

His account of AI tools creating both output and exhaustion matches what David Holz, founder of Midjourney, wrote on X:

“my friends are all feeling extremely productive and also extremely drained with the latest coding models. this makes me feel like something is wrong, and also that there might be a big opportunity. does anyone have any strategies they use to make it feel better day-to-day?”

But feeling productive isn’t the same as being more productive - a frustrating disconnect exacerbated by the fact that most companies have few reliable ways to track whether AI is truly helping engineering teams deliver better outcomes.

“It creates this constant background pressure to keep feeding the agents new work so they’re never sitting idle. That pressure is draining.”

When asked for his take on Holz’s observation, Kanitkar tells The New Stack it’s the new managerial aspect of their work that’s causing fatigue, as engineers have to constantly context-switch to manage several agents at once. But if engineering teams shrink, as Gartner predicts, this constant agent-babysitting could become the new norm - though the jury’s still out on whether this new kind of engineering is actually more productive.