I built an AI that rates your outfits honestly. Here's what I learned after 5,000 monthly visitors.

Congrats on the launch, looks solid. How are you currently thinking about acquiring early users and gathering feedback?

the honest scoring insight is real. working on AI image generation tools and we found the same thing - when our quality checker flags bad outputs instead of passing everything through, users trust the pipeline way more. they know that when something scores well, it actually is good.

on the vision API side - have you looked at cost per analysis? if youre using gpt-4o for each outfit scan, that adds up fast at 100+ daily users. there are cheaper vision models now that could handle style analysis without the full reasoning overhead. could meaningfully improve your unit economics at scale.

Did u add suggestions tab

The insight about honest scores converting better is really interesting. I've noticed something similar in dev tools — when you give people sugar-coated metrics, they stop trusting the dashboard entirely. But when the numbers sting a little, they actually pay attention and take action.

Curious about the freemium tiers though. The jump from 5 free scans to unlimited at $9.99/mo feels like there might be room for a middle tier. Something like 20 scans/mo for $4.99 could catch people who don't need unlimited but want more than 5. Have you looked at where most users hit their limit and churn?

Also wondering how you handle edge cases in the scoring — like cultural differences in fashion norms, or formal vs casual context. Does the user specify the occasion before uploading, or does the AI try to infer it?

Hey Saad,

Love the honesty angle — building an AI that doesn’t sugarcoat outfits is refreshing in a sea of “you look amazing” apps. Congrats on 5k monthly visitors and 100+ DAU as a solo founder; solid traction.

How long did it take you to build your MVP, did you ever experience feature creep?

Looove this idea!!

Brutal honesty paying off in conversions is a refreshing take. Most apps sugarcoat to keep users coming back short-term but OutfitScore proves real trust from tough love drives upgrades way better. Hitting 5k visitors and 100+ DAU with that model shows the approach works. Smart freemium setup too, letting people taste the value before paying. Keep leaning into those improvement tips on low scores, it turns criticism into motivation and revenue. Solid build and great lesson shared.

This is such a goldmine of an insight, Saadsella! 🏄‍♂️

The fact that honest (low) scores actually drive higher trust and conversion is a counter-intuitive win. Most people are afraid to give users 'bad' news, but you proved that utility beats ego every time.

Regarding your question on monetization: have you thought about a 'Consultation' tier? If someone gets a 43, they might pay a one-time fee to get a 'fix' recommendation.

I actually lean into a similar 'honesty' philosophy with a tool I just finished called SaaSurfer. It scans Reddit to give indie hackers a 'Demand Score' (0-10) for their SaaS ideas. If there's no demand, it gives a RED badge. Like you found, builders actually appreciate the blunt truth before they waste months building.

I’m actually selling the project right now ($2,200) because I’m a builder at heart and ready for the next challenge. It’s pre-revenue but fully deployed on Vercel with all the payment UI and documentation ready for someone to take the reins.

If anyone here is looking for a 'validation-as-a-service' stack to add to their portfolio, I’d love to chat!

Congrats again on the 5k visitors—that’s a huge milestone!

5k monthly visitors with 100+ DAU is solid — especially in a consumer niche.cThe honesty → trust → conversion insight is interesting. A lot of AI products optimize for “feel good” outputs, but it sounds like credibility became your real moat. On monetization — have you considered leaning into event-based use cases (weddings, interviews, dates) instead of just unlimited scoring? Feels like high-intent moments might convert even better than general usage. Also curious if you’ve seen users come back repeatedly, or if it’s more of a “one-off before an event” behavior.

I struggle with getting the ability to post here so this is impressive!

This is a great example of trust > vanity metrics.

I built free tools too, and I noticed the same pattern — when the result is honest (even if it’s not flattering), people respect it more and come back. Fake-positive feedback might increase short-term usage, but it kills long-term trust.

One suggestion: show “how to improve” along with the score. People don’t just want a 55 — they want to know how to reach 75. That makes the product more actionable and increases the chance they upgrade.

Brutal honesty is actually nice

The brutal honesty angle is brilliant. I built something similar in a different space (Healthien, AI calorie tracking via photo) and ran into the same temptation to make the AI "nicer."

Users kept saying the calorie estimates felt too low or too aggressive. My first instinct was to adjust the algorithm to be more forgiving, but the ones who stuck around were actually the people who wanted the honest numbers, even when they stung.

Your point about trust driving conversions makes total sense. When everything else online tells people what they want to hear, being the one source of honest feedback becomes valuable.

Question about your AI prompting: How do you balance being constructive vs just harsh? I struggled with this - honest but helpful vs honest but discouraging.

This is actually a solid idea! I could see why people would be invested in this!

Hi there! First of all, huge congrats on the launch and getting your first users !

I was checking out the website and I was thinking that maybe the landing page could be a bit cleaner to help convert even more users !

Quick question: have the ads started generating any revenue already?

Interesting concept! Brutal honesty as positioning is bold, and the traction numbers are solid.

From a security and privacy perspective, a few questions:

• Since users upload photos of themselves, how are you handling image storage? Are photos deleted after scoring or retained?
• Are you stripping metadata like EXIF to prevent accidental leakage of location or device info?
• Are images ever used for model training, or strictly processed for inference?

Apps that process personal photos need strong privacy guarantees. A clear statement about retention, deletion, and model usage would likely increase trust and conversion.

Concept is strong. Privacy transparency could become part of your competitive edge.