When execution becomes cheap

Originally published September 2025. Part of the Archived Essays collection.

“Holy shit.”

Those were the first words out of my mouth when I revisited Replit after Agent 3 dropped.

From ideation to coding to debugging, a swarm of agents collaborated like a relay team passing the baton:

• The Fullstack Agent handled the backend
• The Architect reviewed and tweaked
• The Tester hunted down the bugs

In under two hours, I had a functional app that would’ve cost five figures three years ago.

I’ve been into tech and building for years, but this felt different.

It felt like watching the future assemble itself in real time.

Ideas, knowledge, execution

Every week, new capabilities land and every week, it gets easier to build and ship.

AI isn’t a silver bullet and currently there’s more hype than signal, but ignoring what’s happening now would be as foolish as believing every headline.

We’re just not wired to grasp exponential change.

The “build stack” used to look like this:

• Ideas
• Knowledge
• Execution
• Distribution

Now it’s shifting.

Ideas

Ideas have always been abundant.

Everyone has them, good, bad, brilliant but few are brave enough to expose their ideas to reality.

We all know the person who says:

“I had that idea years ago.”

Cool. But you didn’t ship.

Ideas were never the bottleneck.

Knowledge

Knowledge used to be a moat.

In the early 2000s, it lived behind university walls and paywalled journals.

By 2010, it exploded: YouTube, blogs, online courses, forums. The internet handed anyone motivated enough a near-free education.

Knowledge went from privilege to commodity.

Execution

For the longest time, execution was a huge area of leverage.

You could have the idea and the knowledge, but you still needed capital, talent, and operational skill to turn it into something real. That’s where the founder’s edge lived.

That edge is eroding.

I’ve seen it firsthand: it cost me ~£50 of compute, an hour of planning, and an hour of tinkering to build a functional MVP with AI agents.

No team, minimal risk, just prompts and an agent network doing the work.

Yes, humans still have to be in the loop—especially on hard, technical or domain-specific problems, but the baseline cost of execution is falling fast.

Execution is on a fast track to join knowledge and ideas in the commodity bin.

Which leads to the next question:

When everyone can build almost anything, what still matters?

Discernment & distribution: the new edge

The edge is moving from execution to discernment and distribution.

When building and writing code get close to free, your advantage becomes:

• Knowing what not to build
• Seeing which problems actually matter
• Translating those problems into clear instructions for agents and systems
• Designing environments that feel stable, meaningful, human
• And getting the product in front of the right people

Because in a world where everyone can make, the only thing that matters is whether anyone cares.

Execution and speed will continue to level out.

Discernment and distribution is where the gap reappears.

The implications

When execution becomes cheap, the value moves elsewhere:

• Into meaning – why this product or system should exist at all
• Into trust – why anyone should rely on it
• Into discernment – how you decide where to aim your energy
• Into distribution - how do I get this out to the right avatar at the right time

We’ll need to redefine what “creating value” means when anyone can spin up an app, newsletter, or agent army on a weekend.

We’ll have to untangle identity from output:

• If a machine can do a task faster, what is your role?
• If tools handle the busywork, what do you measure yourself against?
• If execution is easy, what do you choose to build anyway?

That’s where it gets uncomfortable—and where the real work starts.

As tools become more intelligent, the humans who thrive will be the ones who stay intentional:

• Clear on their lens
• Clear on their constraints
• Clear on what they’re optimising for

Where the edge lives now

So where does the edge live today?

It’s still early, but my current bet:

The edge belongs to those who can move fluidly across disciplines, see clearly, and decide precisely.

People who can:

• See patterns across domains
• Frame problems in ways agents + humans can work on
• Design simple, elegant systems around those problems
• Tell compelling stories about why those systems matter
• Distribute those stories to the right people

Well-rounded generalists and multi-skilled operators are going to rise:

• Strategic organisers of systems, data and experience across all slices of business
• Translators between business, tech and humans
• People who weave coherence from chaos and build communities around it

In one of my recent projects, discernment meant ditching a flashy AI feature that demos well for a “boring” one that built user trust.

The result?

Stickier engagement, better retention and much less noise in support.

FAQ: navigating commoditised execution

What is “discernment” in the context of AI building?

It’s the discipline of prioritising what matters over what’s possible.

The temptation in an AI-rich world is to add every shiny capability. Discernment is asking:

• “Does this solve a real problem?”
• “For a real person?”
• “In a way that fits the system and business I’m building?”

It’s turning down 90% of what you could build so you can go all-in on the 10% that matters.

How does this shift impact founders specifically?

It democratises MVP creation but raises the bar on strategy and clarity.

You can build faster. You can test more ideas. But:

• You need a sharper ICP
• You need clearer constraints
• You need better judgment on when to say “no”

You’re no longer rewarded for “I built this”; you’re rewarded for “this meaningfully moved the needle.”

What edges remain for humans in an AI-heavy landscape?

Focus on traits machines don’t replicate well:

• Curiosity
• Empathy
• Cross-domain pattern recognition
• Taste
• Community building

AI investments will keep ramping, tools will keep getting better, but people still buy from, follow, and build with humans they trust.

Should I be worried about my skills becoming obsolete?

Only if you treat them as fixed.

Treat your expertise like software:

Update or obsolete.

The founders who thrive will be the ones who learn, unlearn, and relearn—continuously.

Not in frantic panic, in steady, deliberate cycles.

The beauty of building right now

This is just the beginning of what’s next.

We’re all going to have to figure out:

• How we build alongside machines
• Where our edges move
• How we keep meaning alive as tools accelerate

The world isn’t fixed, you can push it, shape it, and something will move.

That’s the beauty of building in this moment: You don’t need permission.

You need curiosity, intent, and the courage to ship.

—Tariq