The Jenga Tower Paradox: What AI Is Really Doing to Your Organization

In the middle of the current AI wave, it’s tempting to believe the hard part is tooling: Which copilot should we use? Which vendor? How fast can we roll something out?

But if you listen carefully to the people who sit between technology, people, and commercial outcomes, a different story emerges.

That story came into sharp focus during a recent roundtable hosted by BetterEngineer with three very different, deeply complementary perspectives:

• Barry Marshall – Founding Partner & CEO at P5 and former COO at JP Morgan
• Boris Portman – Co‑founder & CEO at BetterEngineer
• Arthur Woods – Startup entrepreneur, talent executive, and co‑founder of 1080 Ventures

What followed was a candid look at how AI is already reshaping work, why demand for strong engineering talent is rising, and what leaders must rethink if they want to build organizations that are truly agentic‑first, not just “AI‑enabled.”

This article distills the key ideas from that discussion.

1. AI Is an Inflection Point, But Not the Productivity Miracle Many Expected

For Boris Portman, who studied AI at Carnegie Mellon and has spent decades in and around engineering teams, the current moment feels familiar.

The pattern he sees now is the same one he saw then:

• Business leaders look at tools like Claude, GitHub Copilot, and rapid app builders and ask a seemingly reasonable question:“If these tools can generate code and UI in seconds, why can’t we ship this product in days?”
• Engineering leaders know that’s not the full story. Beyond an impressive demo, anything serious must still answer for:
• Governance and access control
• Security and compliance
• Reliability, performance, and observability
• Data quality and data sovereignty
• Long‑term maintainability

This gap between expectation and reality is exactly where many teams are struggling. AI has made “what’s possible” very visible, very fast. It has not magically dissolved the hard operational work beneath it.

Boris used a simple analogy:

“When you add a lane to a highway, it doesn’t actually decrease traffic congestion.”

The same is true with AI. The organizations going all‑in aren’t on the beach. They’re working harder and thinking more broadly about business problems, user experience, and success metrics. More capacity without better design doesn’t fix congestion; it just moves it. AI is no different.

2. The Jenga Tower Paradox: Why Leadership Doesn’t Get Easier with AI

Where Boris focused on the engineering reality, Barry Marshall zeroed in on the organizational one.

Barry has lived through extreme scaling. At JP Morgan, he helped build a global operations center from zero to more than 10,000 employees in five years. That experience wasn’t just about operations—it was about organizational design under pressure:

• How do you build systems that still work when complexity outpaces your ability to manage them manually?
• How do you maintain quality and sound judgment at scale?

In today’s AI wave, he sees many leaders repeating a dangerous pattern.

They’re cutting people, layering in AI tools, and assuming the math is simple: fewer people + more tools = more efficiency.

Barry calls the reality something closer to a “Jenga Tower paradox.”

Imagine your organization as a Jenga tower. Each block represents a person. As you remove blocks—cut layers, reduce headcount—you expect AI to fill in the gaps.

On paper, the structure is “leaner.” In reality, what’s left is more fragile:

The leadership implications are the opposite of what many assume:

• Your ability to lead, retain, and develop people does not decrease with AI adoption; it increases.
• The remaining team members carry more complexity, more decisions, and more risk.
• You have to do more to support them, not less.

From P5’s vantage point, this is the crux of the transformation problem. The organizations that will actually capture value from AI are the ones willing to fundamentally rethink how work gets done:

• Who does what
• How decisions flow
• Where human judgment matters most
• And where it explicitly does not

This is the difference between adding AI to operations and building what Barry calls agentic‑first organizations. Arthur summed up the leadership challenge this creates: AI may increase velocity, but leadership still determines trajectory. As AI speeds everything up, “teams need stronger direction, not less management.”

3. Work Slop, Automation Slop, and the Need for an AI Philosophy

The panel also explored a quieter risk: “work slop”—when teams adopt AI but unintentionally switch off critical thinking.

It shows up when:

• People over‑trust AI outputs and stop interrogating them
• Critical judgment is quietly outsourced to a model
• Tasks get done faster, but not necessarily better

Arthur also pointed to a critical risk in how AI is rolling into organizations:

When teams start to treat AI as an answer engine rather than a reasoning partner, there’s a real danger that critical judgment begins to atrophy.

Barry’s response started with something surprisingly simple: an AI philosophy.

Rather than launching tools and hoping people figure it out, he encourages leaders to clearly articulate:

• Where AI is expected
• Where human review is non‑negotiable
• What “good use of AI” looks like in this organization
• What crosses the line into work slop

In Barry’s words, this isn’t just a leadership philosophy; it’s something you explicitly say to your team. The expectation is that people apply the same intellectual discipline and rigor when they use AI as they would when producing work entirely on their own. In his view, AI should be positioned clearly as an augmentation of what people do, not as an artificial replacement for their contribution.

4. The Future of Engineering Talent: Smaller Teams, Bigger Surface Area

One of the myths that surfaced in the conversation: “AI will reduce the need for engineers.”

From BetterEngineer’s vantage point in the market, Boris is seeing the opposite.

“We’re not seeing a drop in demand for engineers. If anything, it’s going up. The skills are changing—data, systems thinking, outcomes over years‑of‑language—but the need for strong technical talent in an AI world is only increasing.”

What’s changing is the shape of that talent:

Barry echoed this from the organizational side. In leaner, more AI‑augmented companies, individual engineers often have larger surface areas:

• They touch UX and front‑end decisions
• They understand back‑end and data flows
• They’re part of conversations about where agents should act vs where humans should

That demands what he sees as a modern archetype: full‑stack engineers who can think like product owners, not just implementers.

5. Talent and People Leaders Are Quietly Becoming “Chief People & AI Officers”

One of the more interesting trends Boris brought back from the TRANSFORM conference was how much of the AI agenda is landing with talent leaders.

What used to sit primarily with the CTO or a head of AI is increasingly in the remit of HR and people leaders:

“Talent leaders are also the ones now responsible for implementing AI strategy where it used to be maybe the CTO or the Chief AI Officer… in many organizations that’s merged into the head of talent, both AI and human.”

Barry argued that this is a side effect:

What’s required is a tighter partnership between engineering, business, and people functions:

• Business leaders need more technical literacy.
• Engineering leaders need deeper connection to outcomes, customer experience, and the language of the boardroom.
• People leaders need tools and frameworks to design new roles, career paths, and training that reflect a human+AI reality.

6. What to Look For in AI‑Era Talent

Beyond engineers, the conversation also turned to the kinds of people organizations need alongside them.

Barry looks for curious, system‑thinking problem solvers, people who don’t just execute tasks, but question the task itself.

Arthur extended that point to the organization as a whole. In his view, “we’re going to need to upskill the entire organization… to sort of meet more in the middle around both technical and soft skills that are required for this change.” It’s not enough to hire a few AI‑literate specialists; everyone who touches critical workflows will need stronger problem‑solving, comfort with data, and the ability to collaborate across disciplines.

When you bring AI into the mix, that becomes even more important. You don’t want people who only ask, “Which prompt should I use?” You want people who ask:

• “Should this process exist in this form?”
• “What are we actually optimizing for?”
• “How do we protect the business metrics at the end?”

Boris added that organizations are increasingly evaluating talent on outcomes rather than narrow stacks:

It’s less about ticking every box on a tool list and more about demonstrating the ability to navigate complex, changing systems.

7. Practical Moves Leaders Can Make Now

The roundtable wasn’t theoretical. Several concrete steps emerged that any leadership team can start on, whether you’re an early‑stage company or a global enterprise.

Here are a few:

1) Create bandwidth on purpose.

Use individual‑level AI wins (calendar automation, inbox triage, routine task agents) to free up time—and then reinvest that time in higher‑order thinking and experimentation, not just cost‑cutting.

2) Give people space to experiment.

If AI is truly creating capacity, use it to let people explore and build with AI inside the company. Otherwise, it becomes too easy for that experimentation and value creation to happen elsewhere, not for your organization.

3) Assume constant change and build flexibility.

There is no static playbook. Whatever you come up with this week will likely need to evolve next month, so talent and org design should intentionally build in flexibility, and prioritize people who can adapt as the landscape shifts.

Arthur captured the spirit of the discussion in his closing:

“This is a paradigm shift. We don’t actually have a playbook… we’re all kind of navigating it together.”

None of these require a new platform. They require new decisions.

8. From AI Tools to Agentic‑First Organizations

If there was a single throughline in the conversation between Barry Marshall, Boris Portman, and Arthur Woods, it was this:

AI is not just a new set of tools. It is a forcing function to confront questions organizations have been able to delay:

• How is work actually designed here?
• Where does human judgment make us better and where does it slow us down?
• What would it look like if our systems, people, and agents were designed to create capacity that none of them could create alone?

Barry and the team at P5 focus on exactly that shift: helping founders, boards, and leadership teams move beyond incremental automation to genuine operational rewiring with an agentic‑first lens.

Boris and BetterEngineer see the same pattern from the talent side: companies don’t just need more engineers; they need the right engineers in the right structures, with clarity on how AI fits into their work.

The organizations that will win this era are the ones that are willing to rethink how work gets done—who does it, how decisions flow, and how humans and AI systems can build something together that neither could build alone.

_____________________________________________________________________________________________________________________

This article is adapted from a roundtable conversation hosted by BetterEngineer featuring Barry Marshall (P5), Boris Portman (BetterEngineer), and Arthur Woods on AI, talent, and the future of work.

For leaders who want to hear the full conversation behind these insights, the complete AI Talent Executive Roundtable is live on BetterEngineer’s podcast channel. You can listen to the episode here.