The E-Shaped Product Leader: Why Stacking Skills Beats Specialisation

TL;DR

• The T-shaped model (deep in one skill, broad awareness of others) is giving way to the E-shaped model: deep in one skill, functionally competent in two or three adjacent ones
• AI collapses the cost of acquiring functional competence in adjacent domains
• The career moat isn't depth alone. It's the rare combination of domains that makes you impossible to replace.

---

There's a standoff happening in product teams right now. Engineers believe they can do product management and design because AI gives them the tools. Product managers believe they can code and design because AI gives them the tools. Designers believe they can code and manage products because AI gives them the tools.

They're all correct. And that's the interesting part.

The T-shape had a good run

For two decades, "T-shaped" was the gold standard career advice. Go deep in one domain (the vertical stroke) and develop broad awareness of adjacent domains (the horizontal stroke). A T-shaped product manager understood engineering constraints and design principles without being able to write code or design interfaces.

That model worked when cross-domain competence was expensive to acquire. Learning to code meant years of practice. Developing design skills required thousands of hours of exposure. Understanding pricing strategy meant an MBA or a decade of commercial experience. The horizontal stroke of the T was necessarily thin because building it was slow and costly.

AI changes the economics of skill acquisition. Not mastery. Skill acquisition.

A product manager who couldn't code a year ago can now build functional prototypes in a weekend. A designer who didn't understand data analysis can now run queries and generate insights using AI. An engineer who never thought about pricing can now model unit economics with AI assistance.

The cost of functional competence in an adjacent domain has collapsed. The T-shape doesn't break. It evolves. The horizontal stroke gets thicker, with actual functional competence rather than surface awareness. And the smart play is to grow multiple vertical strokes.

That's the E-shape. Or the F-shape. Or the comb. The metaphor matters less than the structure: deep in one domain, functionally competent in two or three others, with AI bridging the gap between awareness and ability.

The combinatorial advantage

Scott Adams had a framework for this long before AI made it practical. He argued that being in the top 25% at two different skills made you more valuable than being in the top 5% at one skill, because the combination was rarer than either individual skill.

Adams was a cartoonist who understood business. Neither skill alone was exceptional. The combination produced Dilbert, one of the most commercially successful comic strips ever created. A better cartoonist who didn't understand corporate culture couldn't have written it. A sharper business mind who couldn't draw couldn't have created it. The intersection was the product.

The economic term for this is non-fungibility. Larry Summers used to tell people: "Don't be fungible." If you're "just a product manager," you can be swapped for any other product manager. If you're a product manager who can also ship production code and model pricing economics, you're one of maybe two hundred people in the world with that combination. Good luck replacing you.

AI amplifies this combinatorial advantage because it lowers the barrier to acquiring each additional skill. The second and third vertical strokes of your E are cheaper to build than they've ever been. The combination remains just as rare and valuable, because most people aren't building those strokes.

What the E-shape looks like in practice

I didn't set out to become an E-shaped operator. It happened through a sequence of expansions that AI made possible.

The primary stroke: product strategy and leadership. Nine years running a portfolio that grew from three products to eight, $70M P&L, 4,000+ enterprise seats across Tier 1 banks. This is where the depth lives. Strategy, prioritisation, stakeholder management, pricing, go-to-market. Thousands of hours of accumulated context.

Second stroke: AI coding and system building. Starting in late 2025, AI tools let me close a gap I'd wanted to close for years. I went from product leader who could read code to product builder who could ship production SaaS platforms. Two multi-tenant platforms. Fifty-plus AI features each. Six LLMs in orchestration. Stripe billing. Native mobile. Real users.

Third stroke: growth and distribution. SEO, CRO, content strategy, lead generation. I built these as portfolio-wide disciplines at Cotality, taking OnTheHouse from a side project to a million-plus monthly visitors. Give-to-get funnels, organic growth systems, conversion optimisation. Not as a marketer, but as a product leader who understood that distribution is a product problem.

Fourth stroke: pricing and unit economics. Restructured pricing across a portfolio, designed tiered models, modelled AI inference costs, built Stripe billing integrations, analysed margin structures. From flat per-seat to value-based tiered pricing, delivering 15% ARR uplift.

No single stroke is world-class in isolation. The combination is rare. I can identify a market opportunity, build the product, price it, distribute it, and measure the economics. Each stroke reinforces the others. The pricing knowledge informs the architecture decisions. The growth experience shapes the product design. The building experience grounds the strategy in what's actually possible.

That's the E-shape advantage. Not brilliance in one domain, but functional competence across domains that create compound leverage.

How AI enables the stacking

AI isn't just a tool for doing work. It's a tool for learning how to do work. This distinction is underappreciated.

When I was learning to build production systems, I didn't just have AI write code. I had AI teach me why it was writing the code that way. "Explain this database migration." "Why did you choose this state management pattern?" "What are the security implications of this approach?" The AI was simultaneously doing the work and training me to evaluate the work.

This creates a compounding loop. You use AI to produce output in a domain you're learning. You interrogate the output to understand the reasoning. You develop judgment about what good looks like. That judgment lets you direct the AI more effectively. Better direction produces better output. Repeat.

Within six months, I went from "I need AI to write every line" to "I write the architectural scaffolding and let AI fill in the implementation." The AI accelerated both the doing and the learning. In the old model, acquiring coding competence would have taken years of deliberate practice. The AI compressed the timeline without eliminating the depth, because I was learning through production-grade output, not toy exercises.

This same loop works for any adjacent skill. A designer learning data analysis. An engineer learning pricing strategy. A marketer learning product management. AI lowers the floor (you can produce functional output faster) and raises the ceiling (you can access expert-level feedback on demand). The combination makes skill stacking viable for anyone with the curiosity and discipline to pursue it.

The Mexican standoff resolves

The standoff between product, engineering, and design is real, and it's going to reshape organisations. But it doesn't resolve in one role "winning." It resolves in the roles blurring.

The product builder who can code, design, and think strategically doesn't need the traditional handoff chain. The designer who can build prototypes and understand business models doesn't need a product manager to translate their vision into requirements. The engineer who can talk to customers and evaluate business impact doesn't need a PM to tell them what matters.

This doesn't eliminate specialisation. Deep expertise still matters. The database engineer who understands query optimisation at the kernel level is still essential. The interaction designer who's spent ten years studying how humans process visual information brings insight that AI can't replicate. The product strategist who's navigated three market transitions has pattern recognition that no LLM training set contains.

What changes is the floor. The minimum viable competence in adjacent domains rises from "I've heard of it" to "I can do it with AI assistance." And that higher floor creates individuals who can operate across wider scopes, ship faster, and make better decisions because they understand the full system, not just their slice.

The career strategy

If you're planning your career for the next decade, here's the E-shaped playbook:

Pick your depth. What's the one domain where you want to be genuinely excellent? Not good. Excellent. This is your primary stroke, your identity, the thing that anchors your credibility. Protect it. Keep going deeper.

Choose two adjacent domains strategically. Don't pick randomly. Pick domains that create combinatorial value with your primary skill. A product manager adding coding and pricing creates more leverage than adding coding and graphic design. An engineer adding product strategy and data analysis creates more leverage than adding project management and copywriting. Choose adjacencies where the combination is rare and commercially valuable.

Use AI to compress the learning curve. Don't try to learn the adjacent domains the traditional way. Use AI as a tutor, a pair programmer, a sounding board. Build real things in the adjacent domains, not exercises. Ship a prototype. Run a pricing model. Design a user flow. Production exposure teaches faster than theory.

Compound the strokes. The real advantage emerges when you start using your skills in combination. Don't do product strategy in one context and coding in another. Build a product that requires both. Price something you built. Distribute something you designed. The compound value is in the intersection, not in each skill independently.

Stay uncomfortable. The moment all your skills feel natural, you've stopped expanding. The E-shape advantage comes from continuously adding functional competence in new domains while maintaining depth in your primary one. AI makes this possible at a pace that would have been absurd five years ago.

The T-shaped career was a strategy for scarcity. The E-shaped career is a strategy for abundance. AI gives you the tools to acquire skills at a fraction of the historical cost. The question is whether you'll use them, or whether you'll keep polishing the single skill you already have while the people around you build combinations that make them impossible to replace.

Don't be fungible.

---

Frequently Asked Questions

Doesn't this create "jack of all trades, master of none" generalists?

No, because the E-shape requires maintaining genuine depth in at least one domain. The difference from a generalist is the primary stroke: deep, credible expertise that anchors everything else. The adjacent skills are functional, not superficial. You can ship in those domains, not just talk about them.

How do you maintain depth in your primary skill while expanding into others?

Time allocation matters. Roughly 60% of your development time should stay in your primary domain. The adjacent domains get 20% each (or 15/15/10 if you're building three). AI compresses the learning curve in adjacent domains, so 20% of your time goes further than it would without AI.

What if my organisation still values specialists?

Many do, and specialists will remain valuable in domains where depth is irreplaceable (security, machine learning research, regulatory compliance). But organisations increasingly need people who can operate across boundaries. If your organisation doesn't value the E-shape today, the market will. Build the combination anyway.