Does ‘Doing Things that Don’t Scale’ Apply For AI Startups?
March 23, 2026 · 4 min read•
By: Chintan Zalani
In the dynamic world of startups, sometimes doing counterintuitive things can reap huge rewards.
One of these strategies is doing things that don’t scale, which has made waves in startup circles.
This is where a lot of initial effort and personalized engagement with early users are emphasized.
This allows startups to build an early customer base, gain relevant insight and then move on to the iteration phase.
Let’s hear its breakdown from the horse’s mouth: Paul Graham.
One of the most common types of advice we give at Y Combinator is to do things that don’t scale. A lot of would-be founders believe that startups either take off or don’t. Founders don’t focus enough on individual customers because they worry that approach won’t scale.
But when founders of larval startups worry about this, I point out that in their current state they have nothing to lose. Maybe if they go out of their way to make existing users super happy, they’ll one day have too many to do so much for. That would be a great problem to have. See if you can make it happen.
Joe Gebbia, Airbnb co-founder, emphasizes making initial users happy while noting down their pain points.
Try to manually recruit users; it will teach you more than anything else. We made an extraordinary effort inititially- we flew to New York, rented a camera, and went door-to-door, watching customers use our product in real time. This allowed us to see their pain points firsthand, stepping into their shoes. That’s the heart of innovation: empathy combined with a unique perspective. Because of this approach, people began saying they ‘loved’ Airbnb.
Y Combinator is one of the most well-known startup accelerators that provide seed capital to startup founders.
Let’s see how things run there.
At YC, doing things that don’t scale is a mantra. So at the beginning, we were the delivery drivers – we would go to class and then afterwards deliver food. We were the customer support too; I sometimes had to take phone calls during lectures. We spent afternoons just going down University Avenue passing out flyers.
We didn’t have any dispatch system, so we had to hack together solutions like using Square to charge customers, Google Docs to track orders, and Apple’s Find My Friends to track where our drivers were. Doing these unscalable things allowed us to become experts in our business.
Personal engagement with users is crucial, according to Bram Louwers, Director at BrainManager.
We personally engaged with every early adopter to gather detailed feedback and refine our product. This hands-on approach, though not scalable, provided keen insights and fostered stronger relationships with our initial users. Their feedback helped us iterate quickly and build a product that truly met market needs.
The principle of ‘doing things that don’t scale’ is still incredibly relevant to Dinesh Agarwal, CEO at RecurPost.
He says this is especially relevant in early-stage AI startups.
When we first launched RecurPost, one of the non-scalable tasks that contributed to our growth was personally onboarding our initial users. Instead of relying solely on automated systems or broad marketing, I dedicated time to engage with our early adopters directly. I walked them through the product, listened to their feedback, and even helped them set up their first content queues.
Both of these successful businesses emphasize garnering feedback from their audience.
From Scrappy Beginnings to Scalable Growth
According to Marc Israel, now is the time to act, as the window to gain a competitive advantage is shrinking fast.
He goes on to say that tech CEOs have a 12- to 24-month window to embed GenAI into their operations before it becomes a baseline.
He goes on to cite Gartner as the source for this insight.
Think of autonomous operations as running a business where intelligent agents (both human and AI) manage specific functions in real time. You’ve got AI agents acting as your product managers, financial planners, marketers, and even HR coordinators. Each has a defined role, and they communicate seamlessly to optimize workflows, allowing you to scale without needing an army of employees.
As a reality check, Walker Williams says that the product you launch with will almost certainly not be the one that scales.
He goes on to mention that the focus should be on rapid progress and iteration during the early startup days.
As engineers, your instinct is to build a scalable, clean-code platform, but you need to prioritize speed over scale, not perfection.
For instance, when some major clients showed interest but requested key features, we realized building them the ‘right’ way would take a month—time we didn’t have. So, we chose quick solutions to validate demand, accepting some technical debt along the way.
Scale can actually be the biggest disadvantage to big corporates, according to Illai Gescheit, who founded his own firm.
“Early-stage startups succeed by doing things that don’t scale, focusing on building a great product before worrying about growth,” he says.
Yet corporate innovators tend to think at scale from day one, envisioning global reach before validating product-market fit. If corporates adopt a ‘do-things-that-don’t-scale’ mindset, they might succeed in creating something impactful. Scaling too early, however, often leads to spreading too thin. The real trick is knowing when to switch from scrappy innovation to scalable growth.
I'm the insight architect behind Bot Memo. I have spent the last decade building media assets on the internet. Bot Memo started as a simple project covering industry deep dives. Then I built a data pipeline for it. And now I love analyzing and covering all things AI startups and trends on top of our own data infrastructure.
