The Full-Stack Role Collapse
In the AI era every role needs a minimum baseline in the adjacent two — deep in one, dangerous in the rest.
- Difficulty
- Moderate
- Time to result
- ~months to results
- Steps
- 4
- Confidence
- 90%
AI tooling rewards polymaths. Success with these tools comes down less to which function you're in and more to individual attitude plus the ability to cross over into adjacent disciplines. The model is an upside-down T: go deep in your specialty, but reach a minimum baseline of competence in the other two functions of your triangle (PM/design/eng — and analogously sales/SE, or the marketing sub-roles). As everyone's Venn diagrams converge, teams can do more with fewer dependencies and move in a far more leveraged way.
Origin
Howie Liu's synthesis, which he grounds in long-standing multidisciplinary cultures he admires: Google's original PM spec requiring technical fluency, Apple designers understanding hardware constraints, and Stripe engineers who take the product lead (where the DRI isn't always the PM).
Core principles
- 01It comes down to individual attitude and polymathism more than role title.
- 02There's a strong advantage to anyone who can cross into the other two roles — the hybrid unicorn.
- 03The best PM and design cultures have always been multidisciplinary; AI just raises the baseline and lowers the cost of acquiring adjacent skills.
- 04Being outcome-oriented means collapsing dependencies so you could, if needed, do the whole thing yourself.
- 05Any human with a growth mindset can learn this; the disciplines are not innate 'you have it or you don't' skills.
How to run it
- 1
Name your deep specialty
Identify the one dimension where you go deep — engineering, product/UX, design, etc. This remains the vertical of your upside-down T and your primary source of leverage.
Watch out Don't abandon depth in the name of breadth; the model is deep-in-one, not shallow-in-everything.
- 2
Get 'dangerous' in the adjacent two
Build a minimum baseline in your triangle's other roles. A designer learns enough about how models, tool-calling, and technical constraints work to prototype interactive concepts; a PM gets hands-on and technical instead of only writing PRDs; an engineer learns to think about product and business requirements.
Pro tip Use AI prototyping tools as the shortcut — they let a designer or PM build interactive prototypes without the long hoops of formally learning CS.
Watch out Static, flat design deliverables are increasingly inadequate because the product's value lives in its interaction — the real design of a chat product happens under the hood in how it responds.
- 3
Extend the collapse beyond the product triangle
Apply the same pattern to every function. An account executive becomes SE-fluent enough to demo the product themselves; marketers collapse the performance-marketing / ad-copy / positioning hand-offs so one person can do the whole campaign if needed.
Pro tip It's really hard to sell an AI product now without being fluent enough to demo it live — AE's must absorb SE skills.
Watch out Leaving hard dependencies between roles (AE waiting on marketing assets and an SE to demo) slows outcomes that a more full-stack operator could deliver alone.
- 4
Reset expectations and lean on growth mindset
As a leader, reframe roles: a PM isn't becoming irrelevant, they're becoming a 'hybrid PM prototyper with good design sensibilities.' Make growth mindset a core value and tell people to invest nights/weekends — or take full days off — to build the adjacent skills.
Pro tip Frame proficiency as learnable: everyone can reach 'good enough to be dangerous' at software engineering with a bootcamp or side projects, even mid-career.
Watch out Not everyone will become a Hemingway-level specialist in a second discipline — aim for good-enough baseline, not mastery, in the adjacent roles.
In the wild
Liu points to Google's original PM spec requiring technical fluency, his co-founder Andrew reading visual-design and color-theory books in Google's APM program, Apple hardware designers who understand technical capabilities, and Stripe product groups where the engineer — not always the PM — takes the product lead as DRI.
→ These cultures produced consistently strong product execution, which Liu uses to argue the role-collapse isn't new — AI simply raises the baseline everyone must hit and cheapens the cost of crossing over.
Common mistakes
Staying a single-discipline specialist
A designer who only produces flat static mocks, or a PM who only writes PRDs, can't prototype or judge what's technically one-shottable, so they become a bottleneck as interactive AI products demand cross-functional thinking.
Chasing breadth at the expense of depth
Trying to be equally good at all three loses the deep specialty that makes you valuable; the upside-down T requires one strong vertical plus baseline reach, not uniform mediocrity.
Is it for you?
Best for
PMs, designers, engineers, and go-to-market individual contributors trying to stay relevant and high-leverage on an AI-native team.
Not ideal for
Highly regulated or safety-critical specialties where deep single-discipline rigor matters more than cross-functional breadth.
From the transcript
“I think you need to get like decently good at all three. Like there's just a minimum baseline of like if you're any one of…”
“there's a strong uh advantage to any of those three roles who can kind of cross over into the other two right like kind of…”
“as a PM you need to start looking more like a hybrid PM prototyper who has some good design sensibilities”
From the episode
How we restructured Airtable’s entire org for AI
Howie Liu (co-founder and CEO)