AI is lowering the cost to build — but raising the bar to win
Everyone's celebrating how cheap it's become to build with AI.
I think they're missing the point. Yes — a team of 3 can now ship what used to take 20. That's real.
Everyone’s celebrating how cheap it’s become to build with AI. That’s not the interesting part.
Yes, three people can now do what used to take twenty. Code is cheaper. Prototypes that once needed full sprints now come together in hours. Design, copy, support, analytics, a lot of what used to require teams is now compressed into software. I’ve seen this up close at Deutsche Telekom, where AI isn’t a side tool anymore, it’s reshaping how product and engineering actually operate.
But here’s the part people gloss over: your competitors have the same leverage. The floor just dropped. The ceiling moved up. AI makes it easier to build, and harder to win.
What AI is actually compressing
To understand what’s really happening, you have to separate signal from noise. AI is compressing execution. Turning a clear idea into a working product is faster, cheaper, and accessible to much smaller teams. That’s the visible shift, and it’s getting all the attention.
What it doesn’t compress is judgment. Knowing what to build, for whom, and why now, that’s still human work.
There’s a gap opening between teams using AI as a tool and teams redesigning around it. Most are in the first bucket, they’re faster, that’s it. The second group is rethinking how work even happens, not layering AI on top, but rebuilding workflows from scratch. That’s where the real advantage sits.
We’ve seen this pattern before. The internet didn’t win because websites improved, it won because operating models changed. Amazon redefined retail. Netflix rethought distribution. Airtel changed how content reached users. AI is doing the same thing, just deeper and faster.
The constraint was never coding
Here’s the uncomfortable part. Coding was rarely the bottleneck.
Across two decades of building products, TVs, set-top boxes, OTT platforms, large-scale apps, the constraint was never engineers writing code. It consistently came down to three things:
ICP clarity: Who exactly is this for? Not broad labels like “streaming users,” but a specific person, with a specific frustration, in a specific moment. AI will build anything you tell it to. The decision is still yours.
Distribution: How does your product reach people? AI doesn’t solve reach, trust, partnerships, or channel strategy. A faster product that nobody sees is still invisible.
Positioning: When dozens of teams can build similar products over a weekend, what you say and how you say it becomes the differentiator. Narrative is no longer support, it’s strategy.
When we built AirtelXstream, the tech was hard, but it didn’t decide the outcome. Clarity did, which customer mattered, where they were, and what moment we were solving. That hasn’t changed. Everything around it just got faster.
What this means for product leaders
For product leaders, this shift is blunt. If your value was tied to execution, that ground is eroding. Speed is expected now. Judgment is the job.
The most valuable PM isn’t the one who ships fastest, it’s the one who defines the right problem before anything gets built. Discovery is the scarce skill.
Three capabilities start to matter more:
Market intuition: Reading what people actually need, not just what they say. This comes from exposure, pattern recognition, and staying close to the problem.
Narrative: At scale, the bottleneck isn’t technical, it’s alignment. Getting engineering, business, and leadership moving in the same direction requires a clear, compelling story.
Prioritisation: When building is cheap, options explode. Saying no becomes harder, and more important.
At a senior level, the shift is even sharper. Your role is defining the brief: what to build, for whom, and why now. The “how” is increasingly handled by AI-augmented teams. Your advantage is the clarity you bring before they start.
The uncomfortable implication
A lot of teams aren’t ready for this. If you’ve defined success by output, features shipped, velocity, sprint metrics, this shift will feel disorienting.
When output becomes cheap, it stops being the metric. The real question becomes: was it the right thing to build? That requires honesty about whether you’re solving something real or just something buildable.
A useful question to sit with: if AI can replicate most of your execution work within a year, what are you building that it can’t? Your taste, your market read, your trust with customers. That’s the real job now.
The question that matters
AI is real. The efficiency gains are real. Smaller teams, faster cycles, all of it.
But none of that is the advantage. It’s the baseline.
The question isn’t whether you’re using AI. It’s whether your strategy is sharp enough to make that speed matter. The teams that win won’t be the ones who shipped the most. They’ll be the ones who knew what was worth shipping, before anyone wrote a line of code.
Deepak Upreti