Roles, Lenses, and the Sushi–Schnitzel Problem

8 min reading time

By Phil Wornath on June 17, 2025

Why PMs and UX need clear lanes at scale—and why generalists still matter

“Schnitzel • Döner • Sushi.” If your org chart reads like that menu, your product probably does too.

To be fair, variety isn’t the sin—lack of cooks is. I worry less about döner‑schnitzel‑sushi as a combo than whether the place is run by a PM or TikTok influencer instead of actual chefs. Still, when a menu has 27 distinct “specials,” my risk sensors light up.

Thesis. PMs optimize why/what/when for the business; UX optimizes how for humans. Collapse those lenses in big orgs and you get politics instead of collaboration. In small teams, the same collapse can be productive—if you’re deliberate about it.

Personal note. I’m a “unicorn” by background (UX + code + data + mobility). That mix means my input gets called in early—from problem framing with PMs to stress-testing feasibility with engineers. It’s not about owning every lane; it’s about knowing the consequences across them.

Yes, roles matter. Also true: indie hackers ship real products and real revenue with one brain and strict constraints. Think food truck—not buffet.

1) Why lane clarity drives performance

Google’s Project Aristotle highlights structure & clarity as a core ingredient of effective teams—clear ownership of who does what and how success is measured. That’s not fluff; it’s operational hygiene.
Source: re:Work by Google — Project Aristotle
Adam Smith’s pin factory is the original specialization case study: dividing labor multiplies throughput. In product, that translates to PMs framing problems and bets, UX owning evidence-based design, and Engineering owning feasibility/quality—tight loops, clean handoffs.
Source: The Wealth of Nations (1776)
Conway’s Law: products mirror org communication. Muddy interfaces between roles → muddy product architecture.
Source: Melvin Conway, “How Do Committees Invent?” (1968)
Escoffier’s brigade de cuisine shows why stations and clear roles make a kitchen fast and safe.

2) Why “structurelessness” backfires

“We don’t do process, we vibe.” Translation: the loudest person ships first. It’s a silent disco where the soundtrack is Jira. Rejecting roles doesn’t create equality; it creates hidden hierarchies. Jo Freeman called this the tyranny of structurelessness: unclear authority invites informal power and lowest-friction decisions. In product terms: if PM and UX are “whoever shows up,” you’ll ship the loudest idea, not the best one.
Source: Jo Freeman — The Tyranny of Structurelessness

3) Generalists aren’t the enemy (they’re context-dependent)

IDEO popularized T-shaped talent—depth in one area, breadth across others. Kent Beck extends this with “Paint-Drip People”: multiple drips of depth accrete over time. In startups and creative labs, those shapes thrive because constraints demand hat-swapping. In enterprises, they still matter—just inside clear interfaces.
Sources: IDEO on T-shaped talent; Kent Beck, “Paint-Drip People”

Field note. Wearing both designer and engineer hats has let me ship working prototypes in days, not weeks—design systems wired to data, AI-assisted flows, and real APIs. That speed earns trust, but only when paired with a ritual that separates exploration from commitment.

Indie hacker rules (the short list):

Pick one problem, one audience, one KPI.
Ship thin vertical slices; charge early.
Time-box hat switches (PM → UX → Eng), and write the decision when you switch.
Automate ops/billing before you automate animations.

4) AI changed the calculus—but not the need for lanes

AI is four interns in a trench coat. Great for drafts—terrible as the CEO.

“When I talk to ChatGPT I don’t always assign it labels — but it still works as PM, Developer, Designer, Coach for me.”

GenAI makes cross-functionality cheaper. A PM can prototype; a designer can query data; an engineer can test value hypotheses—faster. The risk is bias-by-prototype: a quick AI sketch masquerades as strategy and anchors decisions. Use AI to explore, not to pre-decide. Keep the ritual that separates ideas from commitments.

How I work now. I run a parallel concept track in v0 with real-ish data (no lorem ipsum), insights from research/benchmarks, and explicit constraints. v0 gives me a running frontend quickly, so FE/BE issues surface early. I iterate to get the first ~80% “good enough,” then invest the last 20% in brand, interaction, and accessibility.

I jokingly call this the 'Creative AI Director' mode: full of wild, exciting concepts—but like any over-enthusiastic intern, it still needs intervention, writing things down and talking to actual users

AI acceleration, new risks:

Hallucinations/bias: treat outputs as hypotheses; validate with users and data.
Security: speed is the enemy of quality; security is a child of quality. Keep a human-in-the-loop code review.
My stance: I can carry the codebase across FE/BE, but I’m not a security black belt—so I bake in a lightweight threat model, dependency scans, auth/roles review, and an OWASP Top‑10 spot check before calling something “done.”

5) The feature-factory trap when lanes blur

It’s a treadmill with a Kanban board. “We shipped 18 cards!” Cool—did any move a metric? When PMs “vibe-design” and designers become order-takers, teams drift into feature factories—shipping output without learning. The escape hatch: measure outcomes, run pre-mortems, and keep discovery alive.
Further reading: John Cutler, “Feature Factory” (Cutlefish Substack)

6) A political metaphor: separation of powers

Montesquieu argued liberty is protected when powers check and balance each other. Product orgs aren’t states, but the analogy holds: PM (agenda/priority), UX (evidence/experience), and Eng (feasibility/quality) check and balance one another. Centralize all power—and error concentrates with it.
Source: The Spirit of the Laws (1748)

Historical callback. Apollo 13’s “mailbox” fix worked because Mission Control had clear decision rights, tight feedback loops, and cross-disciplinary trust—evidence that structure and collaboration save missions under stress.

Org–architecture fit: patterns that scale

Microservices + feature teams: cross-functional squads own services end-to-end; modular orgs tend to ship modular systems (Conway-aligned).
Cross-functional teams in complex domains (e.g., automotive/embedded) reduced integration time by moving from siloed components to feature ownership. Fewer handoffs; fewer surprises.

Titles vs. work (and a time‑travel thought)

Some folks parody role names by self‑crowning as “Technoking.” Fun—but titles are theater; outcomes are dinner. Humans are diverse: the same person can be PM on Monday, facilitator on Tuesday, and bike mechanic by Saturday.

If I’d been around 100, 1,000, or 1,000,000 years ago, I’d probably be organizing the campsite, mapping routes, or running a kayak rental—designing experiences at human scale and making a living from it. The tools change; the craft stays: set the conditions, reduce friction, and help people have a good time (and get safely home).

9) The rise of the UX Engineer (and why it’s happening now)

Many teams are hiring for “UX Engineer”/“Design Engineer” role these days - or at least ask for a long list of ingredients. Know how to design while keeping a tight feedback loop with frontends and APIs. It’s a pragmatic bridge: someone who understands how design tokens, component states, performance budgets, and API shapes affect the lived UX.

Pros

Faster learning loops: working code > speculative mocks
Fewer lost-in-translation moments between Figma and prod
Early detection of FE/BE constraints (auth, pagination, latency) that change UX

Risks (mitigate with guardrails)

“Good enough” ossifies without UX craft reviews
Over-reliance on one hybrid profile—create pairing patterns, not single points of failure

7) The operating system for modern product teams

Write the collaboration contract. One page on who owns what, decision rights, and escalation paths. Use RACI/DACI sparingly but explicitly.
Primer: RACI matrix — Atlassian
Meet in the middle on insight. PM brings business model + bets; UX brings behavioral evidence + usability risk; Eng brings delivery risk + architectural options. Decide together, own separately.
Primer: NN/g on PM vs. UX responsibilities
Guard against bias-by-prototype. Treat AI drafts and PM mockups as hypotheses. Time-box exploration; test before committing.
Choose a cuisine. Be the ramen bar, not “schnitzel–döner–sushi.” Teams with a clear value proposition and clean role interfaces ship better food—and fewer stomach aches. No raw pork with soy sauce. Clarifier: variety is fine if you have real cooks and mise en place; the red flag is influencer menus with dozens of one‑off specials and no kitchen craft.
Name the bridge. If you rely on a UX Engineer/Design Engineer pattern, write the contract: where they pair (PM, UX, Eng), their decision rights, and the handoff back into the craft lanes.
Don’t skip security basics. Quick threat model, scan dependencies, check auth/roles—call a specialist when stakes are high.

Solo OS (indie/consultant mode)

Declare your hat per block: AM = PM/strategy, PM = UX/flows, Eve = Eng/build.
Keep a one-pager: problem, audience, offer, price, single metric.
Weekly release: one outcome, one email, one metric check.
Kill scope fast; keep one “daily special,” not six cuisines.

8) Where I land

Meet in the middle on insight—not in the deep fryer.

Big orgs: strong lanes, shared context, explicit decision frameworks.
Small teams/indie: food-truck energy. Own the stack end-to-end, but keep constraints, a single KPI, and a ritual separating ideas from commitments. Use AI/v0 to draft; spend taste and craft on the last 20%.
Either way, clarity beats vibes. When PM and UX meet in the middle on insight (not ownership), you get speed, creativity, and execution. When lanes blur without intent, you get politics—and a product no one truly owns.

Related: Methods over tools — over‑specialization and tool obsession drift teams from basics. The larger and niche‑ier the team, the more crucial clean information flow becomes so each person can truly own their slice of the product.

Fun fact: Recently I visited Lviv, Ukraine. They sell Sushi Kebab - and it was delicious

Sections