The doorbell rings, and I glance at my phone to see who it is. On the screen, a Girl Scout in a green sash is standing on the porch, a folding wagon piled high with cookie boxes behind her. I can’t help but smile.

These days, there are lots of ways Girl Scouts can avoid the toil of going door-to-door in pursuit of cookie sales. Online ordering, social media and enterprising parents canvassing the office break room are all easy, effective ways of making big sales in record time.

As an entrepreneur, I’m all for these other tactics. But knocking on doors, introducing yourself and making a pitch, one household at a time, does more than sell Thin Mints: It builds resilience, confidence and sets you up with invaluable interpersonal skills. In my opinion, every founder needs these traits.

It’s true that launching a business has never been easier than it is today. AI has made it possible to do everything from automating marketing to building products in record time. Even so, I still believe in the seminal advice offered by Paul Graham back in 2013: Do things that don’t scale. Especially when you’re out finding your first customers, the analog work of pounding the pavement can be the difference between your business taking off and falling flat.

Related: These 3 Simple Steps Will Help You Expand Your Network in Your Base City

Don’t wait for customers to come to you

AI is changing how startups engage with customers in transformative ways — optimizing for Answer Engine Optimization (AEO), for example, is essential for getting your product noticed and promoted by LLMs.

But it can’t be the only tack you take. Even in the age of AI, you still need to ensure product-market fit — the point where a product naturally generates strong market demand. A great idea will stay just that — an idea — until it’s actually being used by its target audience. The early feedback you gather is key to shaping a successful business.

Take Stripe, the now-ubiquitous payment platform whose founders recruited their first users not by passively emailing them a link to try, but by physically taking their laptops and setting it up on the spot. They weren’t the only ones to undertake such dramatic efforts — not unlike Girl Scouts selling cookies, Airbnb’s founders initially went door-to-door to recruit new users and help existing ones improve their listings. Grubhub’s co-founders began by digitizing takeout menus by hand.

Understandably, most founders aren’t overly excited about the prospect of something as labor-intensive as knocking on doors or as uncomfortable as asking someone to fork over their laptop. But as Graham puts it, “marketplaces are so hard to get rolling that you should expect to take heroic measures at first.”

This doesn’t mean you have to follow this blueprint exactly. One founder-friend is based in Silicon Valley, where he organizes meetup events for small businesses and startups. Running this meetup has given him a solid network of friends and acquaintances — and a built-in user-base to pitch the new email marketing tool he’s currently developing. It’s not scalable, but having an established rapport with even a handful of users allows him to gather in-depth feedback in real-time and understand their needs far better than he would otherwise.

Related: Don’t Wait For Customers to Find You — Here’s How to Go to Them Instead

Scaling while testing

Obviously, there’s a limit to how many doors you can knock on, acquaintances you can leverage and menus you can digitize by hand. The tactics that help you win those first critical customers are often the very ones that become impossible to sustain as your business grows.

There’s an art in knowing when to make the shift. Research shows that premature scaling dramatically increases the odds of failure — according to Harvard Business Review, scaling within the first 12 months of founding raises the risk of failure by 20%. On the other hand, staying too scrappy for too long can slow momentum and frustrate early adopters who expect your business to mature.

Rather than rushing to scale, take those same boots-on-the-ground skills you used to find your early customers and continue to A/B test and gather feedback. I started my company, Jotform, almost 20 years ago, but we still interview our users regularly. We do this by offering people a $50 Amazon gift card for 45 minutes of their time, which allows us to target the exact demographics whose feedback and insights we most want. Of course, AI tools have made this process simpler, but it’s critical not to skip it — the HBR research also found that startups that incorporate experimentation through A/B testing are more likely to survive the perils of rapid scaling.

Ultimately, there’s no shortcut around the early hustle, and the work that doesn’t scale is often what sets you up to succeed down the line. AI will continue to reshape how we build and grow companies, but it can’t replace the human grit required to understand your customers and earn their trust. Do the unscalable work first, learn from it and only then let technology help you amplify what works.