23 Days of Zero Users: I Built an AI Dating Coach That Actually Sounds Like You (Not ChatGPT) — Cold Start Honest Feedback Request

One thing I kept thinking about reading this, the tone categories most dating apps suggest are curious, playful, confident. But honestly the one that consistently works across cultures and personalities is humor.

Not forced jokes. Just the ability to be genuinely funny in text without trying too hard. That's incredibly hard to coach because humor is timing, context, and personality all at once. A generic AI tends to produce safe replies that are pleasant but forgettable.

If your product can actually learn someone's specific brand of humor , the way they naturally make people laugh, that might be the real differentiator. Not just "sounds like you" but "makes her laugh like you would in person."

That's the gap I think most people using dating apps actually feel. The replies are fine. They just don't make anyone smile.

Before deciding anything, run unit economics for ONE specific client archetype, not the product in general.

Pick your most likely buyer. Write down: what they pay, how many hours to deliver and support, your actual hourly cost including tool subscriptions. One row in a spreadsheet.

The question you're answering is whether cost scales 1:1 with clients. If it does, you have a services business with a SaaS UI — fine, but prices differently. If one client's support load doesn't grow meaningfully when you add the next client, you have something real.

I tried positioning broadly first ("for developers"). It didn't work because nobody knew what problem I solved. Rewrote the pitch around one archetype (solo dev studios with recurring delivery) and the conversations changed.

$0 MRR is usually a positioning problem, not a product problem. One honest archetype analysis takes 2 hours.

The “sounds like you” angle is the strongest part. I’d make the first screen a tiny before/after: generic AI reply vs one trained on your voice. Right now users have to believe the promise before they feel it.

For $9.90/mo, I’d also test a lower-friction first win: let someone paste 3-5 old messages and see one personalized reply before asking for the full 20-message setup. The product is personal, so trust and the first “oh, that does sound like me” moment probably matter more than traffic right now.

"the sound like you" seem to be a unique selling point. if i want to adopt this, my objection would be something like: "so i'll integrate this into my Tinder then you will have access to my account/chat. why should i even trust you in the first place?" this is one of the main objections: trust. it's not about sounding like them, the main thing is privacy. do you really want to solve the marketing part?

Same boat — launched 10 days ago, first users came only
after I stopped posting in builder communities and went
where the actual problem lives.

For dating specifically: Reddit threads where someone
posts "what do I reply to this?" are goldmine. Show up
with a helpful reply, not a pitch. Trust first.

The Voice Profile angle is your real differentiator.
Lead with that, not "dating coach."

Zero users for 23 days is actually useful data, especially for something this personal.

I audited the live landing page and the IH thread, and I think you've got a strong "Voice Profile" differentiator, but there's a trust gap right at the top.

A few specific things that might move the needle:

1. Sharpen the Hero: "Stop overthinking" is a benefit, but "Learns your voice" is the magic. Maybe: "Text like yourself, even when you're stuck."
2. Show, Don't Just Tell: The "See it in action" section is good, but the current buttons look like generic examples. Show a side-by-side: "ChatGPT generic" vs "Your voice profile" reply. That's where the value clicks.
3. The Disbelief Tax: People assume AI dating tools are "rizz" apps for pickup lines. You mention Harvard and Gottman lower down—move that trust proof way higher, maybe right under the hero.
4. Friction: 12 free replies is generous, but the Voice Profile setup (pasting 20 messages) is a high-friction first step. If you can let people see ONE "sounds like you" moment with just 3-5 messages, you'll get more people into the loop.

I actually do landing page teardowns for founders at https://roastmysite.io/go.php?src=external_manual_ih_cueple_zerousers_may17_usd_presell_hv (it's US$1 for 3 specific fixes) if you want a more brutal deep dive on the conversion flow.

Either way, "sounds like you" is the winning angle. Don't let it get buried under the "dating coach" label.

Zero users for 23 days is tough, but it's not a verdict — it's a symptom. Usually means one of two things: either the people who need this don't know it exists yet, or the problem you're solving isn't painful enough for them to try something new. Dating advice is a crowded space, but an AI coach that actually sounds human is a genuine differentiator. Most AI dating tools sound like a robot reading a psychology textbook.

A few things that might move the needle: find where your potential users are already complaining about their dating struggles — Reddit threads, Discord servers, TikTok comments — and show up there not with a pitch, but with genuine advice that happens to align with what your tool does. People trust the helper before they trust the product. Also, a cold-start honest feedback request like this one is the right instinct. Keep asking. Keep refining based on what actual humans say, not what you assume they need.

What's the most common piece of feedback you've gotten from the few who did try it?

Hi Peterjack.
Just want to understand how your idea; customer looks like?

23 days of zero users is actually useful data.
Have you tried posting in subreddits where your target users already hang out, rather than builder communities?
The feedback loop is slower but the signal is much cleaner.

23 days is exactly where I am right now — just shipped FindAlert into beta and sitting at ~0 users too.
The framing you chose is interesting. "AI Dating Coach That Actually Sounds Like You" is a specific promise. The cold start problem for something that personal is probably trust, not awareness. People need to believe it before they try it.
One thing I keep coming back to at day ~1: the first user doesn't come from distribution. It comes from a single person who has the problem bad enough to try something unproven. Have you identified anyone who fits that description, even one person?
What's been your main channel so far — IH, Reddit, something else?

The Voice Profile idea is actually the most defensible thing you have -- most dating-AI tools are one-shot suggestion engines with zero memory. The longer someone uses Cueple, the more the model learns their voice, and the harder it is to switch to a generic alternative. That's a retention flywheel that most cold-start products don't have. Worth making that explicit in your positioning.

On the cold start / zero users problem: the friction usually isn't the product, it's that you're trying to intercept people mid-frustration when they're not searching for a product, they're just texting someone and freezing. Where are your target users the moment before that freeze? Probably not googling 'AI dating coach' -- they're staring at their phone screen. The channels that convert tend to meet people in that exact moment: Reddit threads where someone posts a screenshot saying 'what do I reply to this?', Discord servers for specific dating app communities, TikTok creators who do 'rate my reply' content.

On the WTP question: .90/month feels right but stated willingness to pay is unreliable. The real signal is whether someone will hit their credit card before they've had a 'wow' moment with the product. Given that your free trial is 12 generations, I'd make sure at least 3 of those generations are extremely high-quality personalized outputs that feel nothing like what any other tool would produce. The 'sounds like you' moment needs to happen in generation 1 or 2, not generation 8.

What's your current sign-up-to-Voice-Profile-completion rate? That's probably the metric worth obsessing over right now.

Appreciate you posting the actual numbers instead of dressing them up — 23 days of zero users is where most of us actually live, and the "honest cold start" posts are way more useful than the survivorship-bias success stories.

On the positioning: "sounds like you, not ChatGPT" is a strong differentiator, but it's a differentiator from a category the user has to already know exists. The hardest part for me on a similar cold start was that my positioning made sense only after someone had tried a generic competitor and felt the pain — so I was paying for a pre-aware audience that didn't exist yet. Worth checking whether your target user has already burned out on Rizz/YourMove, or whether you're the first dating-AI they'd ever try.

One small question: are the 12 free generations gated behind sign-up, or can someone feel the "sounds like me" moment before committing an email? That first-touch friction has been the single biggest lever on conversion in every cold-start product I've shipped.

The strongest angle here is not “AI dating replies.” That category already feels crowded and slightly gimmicky. The more interesting wedge is personal communication confidence: helping someone understand their own tone, avoid generic replies, and build a better pattern across conversations over time.

I’d probably move the positioning away from “dating coach” and closer to “your personal communication style coach for dating.” The Voice Learning part is the differentiator, but it needs to be the headline. “Sounds like you” is much more believable than “AI helps you date better.”

One thing I’d watch is the Cueple name. It fits dating, but it may keep the product locked into a dating-app frame. If this becomes broader emotional/communication coaching, Lyriso .com would feel softer, more human, and more expandable than a dating-specific name.