The 10x engineer is a lie (once more)
Every other year, the myth of the "10x engineer" pops up again. A new technology arrives, and suddenly everyone claims it will make engineers 10 times more productive. This time it's agentic engineering. Let me explain why - once more - the 10x engineer is a lie. At least for your company.
But wait - they do exist ¶
Here's the thing: 10x engineers actually exist. Just not where you would want them to be.
Many skilled, senior+ engineers have recently started solo businesses and go absolutely crazy with agentic engineering. They build products that would have needed a team of 10 or more people before. One person, shipping what used to require an entire company. That's real and it's happening right now.
So what makes them 10x?
It's not about coding speed ¶
Let's look at what these people actually do. Yes, they run multiple AI agents to implement code for them. But that is the least interesting part about their workflow.
While those agents do the cumbersome work of auto-completing the thoughts into working code, the engineers work on everything else: They think about what their product should actually do - product design, or at least product thinking. They break that vision down into implementable portions. They plan in which order to feed work into the agents. They keep an eye on the full architecture - scaling, deployment, security. They foresee side-effects and check quality constraints. And on top of all that, they handle marketing, sales, support, networking, accounting …
The 10x does not come from writing code 10 times faster. It comes from three things working together:
-
Zero coordination overhead. A single person doesn't need to "share state" - knowledge, task splits, work style, priorities - with anyone. There are no syncs, no hand-offs, no waiting for decisions, no misunderstandings, never multiple visions that collide. Every bit of mental energy goes into the actual work.
-
A rare combination of skills. These people aren't just great engineers. They have product sense, business knowledge, and enjoy handling everything from architecture to marketing. That combination is very rare, and it's what makes the whole thing work.
-
AI amplifying all of it. Agentic working doesn't mean to just speed up coding. It augments every skill these engineers bring to the table. Product thinking, marketing copy, support responses, even bookkeeping. AI acts as a multiplier across the board. And it only works as an amplifier for an existing skill set.
Remove any one of these three legs and the 10x collapses. You need all of them. And that's exactly where it falls apart for companies.
Why your engineers won't become 10x ¶
The moment you have more than one person, coordination overhead comes back. That's not a bug - it's structural. A company of 50 has enormous coordination tax, and agentic engineering doesn't reduce it. It only accelerates the parts between the coordination points.
On top of that, specialization strips away the role-consolidation gains. In a team, the engineer doesn't do marketing, sales, or support. Dedicated people or teams handle those functions. That removes a huge portion of the effectiveness the solo entrepreneur gains from consolidating everything into one head.
And even if your engineers get significantly faster at their part - what then? There's input flowing into engineering, typically from product management and support teams. The output flows into QA, marketing, support, documentation. To actually benefit from faster engineering, all of these departments would need to keep up the pace. Otherwise the overall system drowns in the side effects of accelerated engineering output.
In a classical company environment, good engineers can probably reach 2x, maybe 3x efficiency compared to before. And even that requires them to pick up new habits beyond the actual agentic engineering practices: getting more involved in product thinking, working on what comes into and out of their team, filling the gap that opens up when "writing code" is no longer the bottleneck.
The 10x engineer is incredibly rare, too ¶
It's worth noting: even outside of companies, the "real" 10x engineer is a unicorn. How many engineers genuinely have the product sense, business skills, and - maybe most importantly - the interest to run everything solo? Great engineers who also excel at marketing, sales, and customer support? That's not a common profile. The 10x solo entrepreneur is real, but it's not a career path you can just train people into.
So what's the actual takeaway? ¶
The 10x engineer exists - but only as a 10x one-person-company. The 10x employee is a lie. Inside any organization with teams, specialization, and coordination overhead, the structural reality caps your gains far below 10x.
But that doesn't mean there's nothing to gain. 2-3x is real and meaningful - if you're willing to train your engineers, your engineers are willing to go down that path, and if you can rethink how your organization works to actually capture it.
That's a broad topic, probably worth its own post or, better yet, a book. For now, the important thing is to stop chasing the 10x myth and start asking the right question: not "how do we make our engineers 10x," but "how do we change our organization to benefit from what AI actually offers?"
