Week 4 - Speed & Precision: Why I’m Still Choosing Fast Over Perfect in Early-Stage Building

Right Back to Work After the Half-Marathon

Fresh off a half-marathon this weekend, I jumped straight back into AI Grammar Mentor.

The last post about prioritizing speed and precision was still ringing in my head — and today proved again why that mindset matters so much.

Login & Register — Surprisingly Smooth This Time

The main task was finishing the login/register flow.

Thanks to Clerk (chosen in the previous step), implementation was straightforward. No endless AI back-and-forth, no security rabbit holes. I’ve gotten much better at prompting different models and quickly evaluating what they produce.

The real skill isn’t just getting code — it’s understanding why something works, how it connects to what came before, and whether it will still make sense next month.

The “Army of Indians” Mental Model

I once told a colleague I basically want “an army of Indians who implement all my half-baked ideas quickly” so I can see which ones actually stick.

It’s half-joking, half-serious: speed of experimentation beats perfection every time in the very early days.

The fastest way to learn is to ship rough-but-working versions and watch real behavior — not to polish features no one has asked for yet.

Next: Paid vs. Free Differentiation

Now that logged-in vs. logged-out is working, the logical next step is splitting free vs. paid capabilities.

I’m thinking carefully about:

• Which features feel truly valuable enough to pay for
• What motivates someone to create an account in the first place
• How to make the upgrade path feel natural instead of pushy

But I’m deliberately not over-planning every edge case. In the past I’ve fallen into the trap of designing bullet-proof security, blazing performance, and pixel-perfect UX… for features only I would ever test.

That’s fun as a coding exercise. It’s terrible for business momentum.

Payments: No Custom System This Time

I’m not building my own payment & subscription engine.

Why? Because I already lived through that pain with a previous project (PromptIn + ExtPay). It was simpler there, but still a headache.

Lemon Squeezy looked like the better choice for this stage — even though the setup involves more steps than I expected:

• Separate demo vs. live product configuration
• Multiple places to define the same products
• Lots of small decisions about webhooks, taxes, EU VAT, etc.

I almost wasted hours projecting how “it should” work instead of just reading their docs. Once I did the research, it clicked: they solved these problems years ago. I just had to follow the path they already paved.

Lesson reinforced: when in doubt, search + follow proven integrations instead of reinventing.

The Real Goal Right Now

Get the second Google Ads campaign running with proper tracking.

Success at this stage isn’t immediate sales. It’s:

• More account creations
• Better understanding of who signs up and why
• Early signals whether people see enough value to consider paying

I’m not expecting purchases on round two. Round three — after more iteration and clearer data — is when real willingness-to-pay should start showing up.

The Core Principle That Keeps Me Moving

It was precisely this mindset that helped us move so quickly from the initial idea to our first paid campaign. You can read the full story here: The Sleepless Night That Launched AI Grammar Mentor.

Speed remains the #1 constraint.

Even with zero revenue so far, momentum is everything. Every day spent over-engineering non-differentiating parts delays the moment I get honest user feedback.

I’m holding onto that tightly.

If my journey building AI Grammar Mentor (and especially the hard-earned lessons from my earlier project PromptIn) sounds useful to you, I’m happy to share the full key-learnings privately.

Just send a quick message through the contact form with #PromptInStory — I’ll send you everything I wish someone had told me earlier (honestly worth way more than 500 CHF to me in hindsight).

And if you’re curious about the tool itself — whether you’re writing emails, blog posts, reports, or just want clearer English without losing your voice — come try AI Grammar Mentor.

No pressure, no long onboarding. Paste your text and see the difference for yourself.

I’d love to know what you think.