As I finished reporting two recent AI stories, I noticed a common theme staring me in the face. The pure speed enabled by AI-powered engineering is having a much more nuanced impact than simply allowing individuals and startups to build faster; it’s also breaking their processes in significant ways that need to be reckoned with.

When the phrase “AI brain fry” started making the rounds a few weeks ago, I was already deep in the weeds reporting how this is manifesting for developers in particular for a story in LeadDev, which published yesterday. What I found was utterly fascinating: many developers who have embraced agentic coding feel literally addicted to it. They described being unable to walk away from their computers, losing sleep over it, and noticing their weekends disappear. They cite their excitement, but even more so, the very way these AI coding tools function, comparing them to slot machines.

The result is that they’re working at unprecedented speeds and carrying a growing number of open tasks. And this is where the real “break” comes in: they’re working on so much so rapidly — including managing multiple agents simultaneously and constantly context-switching — that they’re often getting lost in their projects and struggling to keep up with the “invisible decisions” their agents are making. The cognitive overload is so significant that one computer science professor who focuses on the social aspects of software engineering is warning the concern for the field is shifting from technical debt to “cognitive debt.”

For Fortune, I was simultaneously writing about Kilo, a startup building AI coding tools that’s taking speed to the max. At Kilo, ever engineer is expected to operate like a “mini CEO” and independently conceive of, build, and ship features in mere days, skipping many of the typical product development steps and not even requiring leadership approval. Company leadership made a specific decision to not bring on product managers — shaking up not only the org chart, but the typical way tech companies function. It’s all about getting the “velocity killers” out of the way, CEO Scott Breitenother told me.

Even for a startup, this seems fast and like a striking level of autonomy. It also flies in the face of the advice I so often hear about AI and especially agentic AI, which is that bolstering governance and ensuring guardrails is paramount.

Taken together, I think these stories pose (and start to answer) interesting questions about productivity-at-all-costs and the trade-offs of AI-powered speed. There is so much more to dive in here, and I’m actually already working on a feature that will explore this even deeper. I have a feeling this conversation is just getting started.

I’ve primarily been writing about AI for business and technical audiences as of late, but I’ve always really enjoyed unpacking complicated technologies for everyday readers. So when Reader’s Digest approached me to tap my own experiences covering AI for over a decade to break down everything the unacquainted would want to know, I jumped at the chance.

The story explores what AI really is in the first place, ways to use it in your everyday life, the risks to be aware of, and what’s next for this technology.

That’s the question I set out to answer for my recent collection of stories in Fortune. The aforementioned story on Kilo was a part of this. I also wrote about…