Rich Mironov says "Code Isn't Product" — AI lets teams ship 100x more code, but customer attention didn't scale. Lenny Rachitsky's survey data shows mid-career engineers are the unhappiest cohort in tech. And a wave of product writing warns that many internal projects are "self-fulfilling." Three pieces, one uncomfortable theme.

Code isnt product , self-fulfilling projects , on mid-career

Rich Mironov dropped a 4-minute essay on May 31 arguing that AI is creating a "market-driven confusion between creating code and creating products." The same week, Lenny Rachitsky's tech-worker survey data shows mid-career engineers are the most burned-out, most pessimistic cohort in the org chart. And there's a new wave of essays warning that a lot of the projects getting greenlit right now are "self-fulfilling" — they exist because someone wanted to build them, not because anyone asked. Three pieces, one uncomfortable theme: the cost of confusing activity for outcome just went up.

What You Need to Know: Mironov's "Code Isn't Product" argues AI lets teams ship 100x more code, but customer attention, budgets, and buying processes haven't scaled to match. Lenny Rachitsky's "How tech workers really feel about work" survey identifies a pronounced mid-career slump: senior individual contributors reporting the most burnout and the most pessimism. A third strand of product-management writing (and a CS Prof paper circulated this month) points to a wave of "self-fulfilling projects" — initiatives that get resourced because of internal politics, not customer pull.

Why It Matters

For engineers: Throughput is no longer the bottleneck. The discovery and positioning work upstream of code is what determines whether anything ships, sells, or sticks.
For product managers: Forward-deployed engineers are booming. Mironov says that's not a feature — it's evidence that you shipped a tool, not a product.
For managers: The "next-bench syndrome" — building for the person sitting next to you — is back, and it's responsible for an awful lot of technically impressive, commercially dead projects.
For senior ICs: The mid-career slump data says the work itself is the problem, not the title. Manager-track isn't the only answer, and for many isn't an answer at all.
For founders: If you can't explain who would miss your product if it disappeared, the project is self-fulfilling — and it will be the first thing cut in the next reorg.

What Actually Happened

Mironov: "Code Isn't Product"

On May 31, 2026, longtime product leader Rich Mironov published "Code Isn't Product" on his blog. The thesis: AI now lets product teams ship 100x more code than before, and boards and CEOs are using that throughput as the stated justification for layoffs at Meta, Amazon, Intel, Microsoft, and Alphabet. But customer attention didn't get 100x bigger. Pitch volume went up, buying processes didn't, and budgets didn't. Mironov quotes April Dunford on positioning: in a world with 100x more offerings, the dozen words you describe a product with are what determine whether anyone even hears about it.

He flags two specific failure modes. First, the forward-deployed engineer (FDE) tell — when a vendor needs a full-time engineer embedded at the customer to explain the product, that's a product problem, not a deployment problem. FDEs are expensive, hard to hire, and don't scale. Second, the next-bench syndrome, a Bill Hewlett-coined term from HP's internal history. Engineers building for someone whose problems, language, and buying behavior they understand (literally, the person at the next bench) get tight feedback loops and reliable intuitions. The moment you extrapolate that pattern to veterinarians, teachers, firefighters, or government procurement officers, you ship things that are technically impressive and commercially dead. Mironov's sound byte: "AI is giving us 100x the code. Nobody is giving us 100x the customers or 100x the revenue." Expect a "summer of DOA products."

"Self-fulfilling projects" in product and engineering culture

A second strand of writing this month — much of it recirculating the same Wardley Mapping and Doctolib "Golem and Pygmalion effects" essays — warns that a lot of internal projects right now are self-fulfilling: they get staffed, scoped, and resourced because of organizational inertia, founder preference, or a manager's career path, not because of customer pull. The signals are familiar: no clear "who would miss this if it disappeared" answer; an SOW that reads like a resume, not a customer problem; success metrics that measure shipping, not outcome. Doctolib's Golem and Pygmalion effects essay frames the same dynamic from the engineering-management side: low expectations become low performance, high expectations become high performance, and managers who don't see themselves in the loop are running a self-fulfilling experiment whether they mean to or not.

Lenny's "How tech workers really feel about work right now"

Lenny Rachitsky published a large-scale survey of tech workers in May 2025, with a follow-up wave in 2026. The headline finding: there is a pronounced mid-career slump. Mid-career workers (typically 8–15 years of experience) report the lowest job enjoyment, the highest burnout, and the most pessimism about the industry's future. The pattern is consistent across company size, geography, and role family. The data also shows 67% of regular AI users say AI is improving their job satisfaction — but the improvement is concentrated in junior and senior cohorts, not in the mid-career band. Lenny's read: the slump isn't about AI specifically, it's about a career stage where ambition has outrun the ladder. The mid-career cohort is large enough that it's now showing up in retention dashboards. Lenny's Newsletter — "How tech workers really feel about work right now".

The Take

These three pieces are saying the same thing from different angles. Mironov: "your code is not the product." Lenny: "the people with 8–15 years of experience are the unhappiest in your company." The self-fulfilling projects writing: "half your roadmap is internally motivated." The throughline is that the AI era isn't just making engineers more productive — it's making the cost of skipping discovery, positioning, and customer proximity visible at a speed that wasn't possible before.

The practical implication is uncomfortable. If throughput is cheap and discovery is expensive, the high-leverage work for the next 12 months is not "writing more code faster." It's picking the right thing, talking to the right people, and writing the right twelve words. That shifts budget and headcount, which is going to be politically ugly. But the data on mid-career engineers is the canary: the people with the experience to do the hard upstream work are the ones most likely to leave. If your roadmap is full of self-fulfilling projects, you don't have a retention problem. You have a strategy problem that's about to become one.

For mid-career engineers reading this: Lenny's data is the headline, but the subtext is that the manager track isn't the only out. The "senior individual contributor who owns a P&L" path is real at a growing number of companies, and the FDE trend is essentially that role being invented under a different name. Pick the version that matches how you actually want to spend your days.

Quick Summary

Three essays, one uncomfortable theme: AI is exposing — not creating — a gap between shipping code and shipping products. Mironov says throughput is no longer the bottleneck; Lenny's data says the mid-career engineers who can do the hard upstream work are the unhappiest; and a growing body of product writing warns that a lot of internal projects are self-fulfilling and should be killed.

Sources:

Source: TLDR | mr.technology — The Master Skill Index