4 guests 5 episodes 2,479 words

The T-Shaped Trap: Why Career Advice Needs a Software Update

Should you specialize deeply or stay a generalist in your career?

The standard career advice has been "go T-shaped" for so long that it has become wallpaper. Deep in one thing, broad across many. Everyone nods. Nobody asks the uncomfortable follow-up: deep in what? Broad in what? And for how long?

The AI era is rewriting the answer. Boris Cherny, who leads Claude Code at Anthropic and has not manually edited a line of code since November 2025, is telling CS students to become generalists -- because the specialist coding skills they spent four years acquiring are becoming a commodity faster than they can graduate. Judd Antin, who built the research practice at Facebook and ran research at Airbnb, insists T-shaped individuals composed into teams is the only model that works. Claire Vo went from copywriter to CPTO running engineering at LaunchDarkly by strategically going deep at exactly the right moments. And Laura Schaffer created entirely new roles for herself at Bandwidth and Twilio by staying close to customers and diversifying before specializing.

The real question is not specialist vs. generalist. It is about timing, sequencing, and which skills have a half-life shorter than your career.

Should you invest your career in becoming a deep specialist or a versatile generalist? And does the AI era change the calculus?

The 4 Positions

1 / 4

AI Makes Generalism the Survival Strategy

Boris Cherny

Boris Cherny, head of Claude Code at Anthropic, ships 10-30 pull requests daily with 100% of his code written by AI. That vantage point makes his career advice unusually credible -- and unusually blunt. On his team, everyone codes regardless of title. The PM codes. The engineering manager codes. The designer codes. The finance person codes. The strongest engineers are not pure technologists but hybrids who cross into product, design, business, or user research.

"Try to be a generalist more than you have in the past. For example, in school, a lot of people that study CS, they learn to code, and they don't really learn much else."
Boris Cherny
▶ 00:41:08

Cherny sees roughly 50% overlap between engineering, design, and PM roles already emerging. His prediction: the title "software engineer" will give way to "builder," and by year's end, everyone will be a product manager who codes.

"Some of the strongest engineers are hybrid, product, and infrastructure engineers, or product engineers with really great design sense. Or an engineer that has a really good sense of the business, and can use that to figure out what to do next."
Boris Cherny
▶ 00:41:43

The historical analog he reaches for is the printing press. In 1450s Europe, literacy was a specialist skill held by sub-1% of the population -- scribes employed by lords who were often illiterate themselves. Gutenberg made literacy universal, enabling the Renaissance. Cherny sees coding on the same trajectory. One engineer on his team built a service in Go over a month and still "doesn't really know Go" -- because the AI handled the syntax. If your deep skill is automatable, depth becomes a liability, not an asset.

"I think by the end of the year everyone is going to be a product manager, and everyone codes. The title software engineer is going to start to go away. It's just going to be replaced by builder."
Boris Cherny
▶ 00:00:44

When This Applies

Shift toward generalism now. Boris Cherny's Claude Code team shows what the future looks like: everyone codes, everyone does product, the boundaries blur. The ability to orchestrate AI tools across domains is more valuable than deep expertise in one tool.

T-Shaped Is a Team Design Principle, Not Just Individual Advice

Judd Antin

Judd Antin, who built the research practice at Facebook and was a longtime head of research at Airbnb -- producing alumni now leading teams at Figma, Notion, Slack, Robinhood, and Duolingo -- goes beyond the generic T-shaped platitude. He specifies exactly what the tools are that make a researcher (or any knowledge worker) genuinely multi-method.

"The best researchers have five tools. And those five tools are number one, what we would call formative or generative user experience research... The second type is evaluative, so more like usability testing. The third tool is a basic rigorous survey design... the fourth tool, which is applied statistics... the fifth tool may now be prompt engineering."
Judd Antin
▶ 00:17:44

Most researchers are deeper in one or two of those tools than the rest. The T-shape is not just about individual range. It is a team composition strategy. When hiring, Antin looked for individuals building breadth across all five tools while being deep in at least one, then composed the team so the collective depth covered every gap.

"You have experts who are t-shaped, but maybe deeper in one or several of those ways. But when I built a team at Meta and at Airbnb, that was my goal, is individually as researchers build up those tools and then as a team build deep expertise that would fill all the gaps."
Judd Antin
▶ 00:20:44

His interview technique reflects the philosophy: give a candidate a juicy, open-ended research question and see if they propose a multi-method approach. A candidate who defaults to one method is a hammer looking for nails.

"If all you have is a hammer, then everything looks like a nail. And so if you give in the context of an interview, let's say, a researcher, a pretty juicy, open-ended research question... a good answer is usually multi-method."
Judd Antin
▶ 00:20:05

The nuance that makes Antin's position distinct from generic T-shaped advice: your depth makes you valuable to a team. Your breadth makes you functional within one. Neither alone is sufficient. And the T-shape must evolve as teams change -- the researcher who was invaluable for quantitative depth may need to develop qualitative skills when the team composition shifts.

When This Applies

Diversify aggressively. Try different domains, company stages, and functional areas. Laura Schaffer went from sales at Bandwidth to growth at Twilio to VP at Amplitude -- each move building on the last. You do not know enough to specialize wisely yet.

Follow the Fun, Go Deep When It Matters

Claire Vo

Claire Vo, CPO at LaunchDarkly and creator of ChatPRD -- likely the most-used PM-specific AI tool -- has been chief product officer at three companies and once ran engineering at Color despite starting her career as a copywriter. Her path is the strongest existence proof that generalism with strategic depth beats either pure breadth or pure specialization.

"When it comes to career growth and that progression from, I actually started as a copywriter of all things, copywriter all the way up to a CPO or CPTO that runs an engineering organization, it boils down to something really simple, which is know what you want out of your career, be clear and ask for it, and then make it easy for your boss or whoever can support or champion you to get you from here to there."
Claire Vo
▶ 00:05:56

Her career moves were not random generalism. They were moments where she recognized an organizational gap and stepped in. When the head of marketing left at her e-commerce company, she drew up an org chart with her name on top, walked into her boss's office, and proposed merging product and marketing under her leadership. She got the job. At Color, she expanded from product into leading the engineering organization because she saw a need and was confident she could fill it.

"PM is such a generalist role, it's okay to go a little left and a little right to go up. I took this marketing growth role. That was actually my first director role. It wasn't only for product, it was for marketing."
Claire Vo
▶ 00:14:32

What makes Vo's career unusual is not the breadth -- it is the willingness to go uncomfortably deep at specific moments. Each lateral expansion (into marketing, into engineering, into building ChatPRD) became a foundation for the next. The generalism gave her context. The periodic deep dives gave her credibility. That combination -- breadth as a platform for strategic depth -- is different from either pure generalism or pure specialization.

"Know what you want but do the work and produce the results, and you can have a career like mine."
Claire Vo
▶ 00:08:44

When This Applies

Deep specialization still wins. Judd Antin's five-tool framework shows that even in research, depth in one method is valuable -- but only when combined with breadth across the other four.

Diversify Early, Specialize Later

Laura Schaffer

Laura Schaffer, VP of Growth at Amplitude, previously built the growth engineering team at Twilio where she cracked the top-50 employee tenure list. Her career framework is the most operationally specific: the typical strategy of working hard and hoping your manager advocates for you is fundamentally limited. Her alternative is to stay close to customers, surface insights executives cannot see, and use those insights to create roles that did not previously exist.

"Career growth is definitely not a straight lineup, but there's definitely some frameworks and methods that have worked really well for me."
Laura Schaffer
▶ 00:06:11

At Bandwidth (50 people, her first real job), Schaffer joined in sales and noticed she was repeating the same information to customers. She proposed putting it online in a self-serve flow. The GM said yes, and she became "e-commerce manager" -- a role that did not exist before she created it.

"I think we should put that online. I think we should make that available so they can just see it and then buy it... And he goes, 'Wait a minute, tell me more about that.' And by the end of the conversation he was like, 'Hey, why don't you go do that?'"
Laura Schaffer
▶ 00:08:46

At Twilio, she applied the same pattern at a larger scale: shared customer insights in a digest, grew it into quarterly voice-of-customer sessions that CEO Jeff Lawson attended, and used the resulting trust to pitch and launch the growth engineering team.

"Your executive team and executive teams at companies are often very sharp, but the nature of their day-to-day just does not link them with customers. And so that means that over time, especially as a company grows, they often lose access to some of the best insights."
Laura Schaffer
▶ 00:09:34

The through-line is not generalism for its own sake. It is diversified experience driven by customer proximity. The specialization comes later, once you have enough diverse experience to know where your natural energy lies and what the organization actually needs.

When This Applies

Pick your depths. Choose two or three areas where you have natural energy and go deep. Claire Vo chose the moments (engineering at Color, marketing at her e-commerce company) -- not a predetermined specialization plan.

Evidence from the Archive

LaunchDarkly / ChatPRD

Expanding from product leadership into running the engineering organization at Color

Drawing up an org chart with her name at the top and proposing a product-marketing merger to her boss -- getting the job

Claire Vo · Chief Product Officer at LaunchDarkly, creator of ChatPRD ▶ 00:05:56

Anthropic

The printing press: literacy went from sub-1% of the population to universal, enabling the Renaissance

The Claude Code team: PM, engineering manager, designer, finance person, and data scientist all write code daily

Boris Cherny · Head of Claude Code

Airbnb / Meta

Antin's research teams at Meta and Airbnb: composed of T-shaped individuals whose deep skills complemented each other

The research alumni diaspora: Matt Gallivan (Slack), Janna Bray (Notion), Celeste Ridlen (Robinhood), Louise Beryl (Figma), Hannah Pileggi (Duolingo) -- all products of this team composition model

Judd Antin · Former Head of UX Research at Airbnb and longtime research leader at Meta ▶ 00:20:44

Amplitude

Bandwidth: joined in sales, noticed repetitive customer interactions, proposed self-serve e-commerce flow, created a...

Bandwidth: joined in sales, noticed repetitive customer interactions, proposed self-serve e-commerce flow, created a new role for herself

Laura Schaffer · VP of Growth at Amplitude ▶ 00:06:11

The Synthesis

The specialist-vs-generalist debate has a hidden variable that most career advice ignores: the half-life of your deep skill.

Skill Half-Life

What hidden variable makes the specialist-vs-generalist debate solvable?

Builder Before Optimizer

Should you specialize early or stay general?

The Comb Model

Is the T-shaped career still the right metaphor?

The hidden variable is the half-life of your deep skill. If your depth is in a skill with a long half-life -- systems thinking, design judgment, strategic narrative -- specializing is safe. If your depth is in a skill with a short half-life -- specific coding languages, manual data analysis -- generalism is your insurance policy.

Growth teams need Innovators, Builders, and Optimizers -- and the first hire should almost always be a Builder (a generalist). Optimizers come once the growth model has traction. Careers follow the same sequence: generalist first to build the system, then specialize once you know what needs depth.

The T-shape needs an update. Think of it as a comb: multiple teeth of depth, connected by a bar of breadth. In the AI era, you want three or four areas of real depth and the ability to connect them. The connecting is the new superpower.

Which Approach Fits You?

Answer 3 questions about your situation. We'll match you to the right approach.

Question 1

Where are you in your career?

Question 2

What is the half-life of your deepest skill?

Question 3

How do you want to create value?

Notable Absences

The Bottom Line

The T-shape metaphor needs an update. Think of it as a comb: multiple teeth of depth, connected by a bar of breadth. In the AI era, you want three or four areas of real depth -- and the ability to connect them. The connecting is the new superpower.

Elena Verna's growth hiring framework, from Lenny's Newsletter "Six rules of hiring for growth," offers a structural parallel. Growth teams need three profiles: Innovators (big thinkers), Builders (generalists who create systems), and Optimizers (deep specialists). The first hire should almost always be a Builder -- a generalist. Optimizers come second, once the growth model has traction. Careers follow the same sequence: generalist first to build the system, then specialize once you know what needs depth.

Sources

Judd Antin — ""The UX research reckoning is here"" — January 4, 2024
Claire Vo — ""Bending the universe in your favor"" — April 7, 2024
Laura Schaffer — ""Career frameworks, A/B testing mistakes, counterintuitive onboarding tips"" — March 9, 2023
Lenny Rachitsky — ""Navigating your early career"" — August 11, 2020
Elena Verna (via Lenny Rachitsky) — ""Six rules of hiring for growth"" — November 9, 2021