What are you building? Drop your AI product below
by•
Doing one of these because I get more from reading what others are building than from any newsletter.
I'll start.
Building AI Hive, an enterprise AI agent platform that helps mid-market and enterprise teams get from AI pilot to production in weeks instead of quarters. The hard problem we keep solving: compliance, model flexibility, and the lack of in-house AI talent on the customer side.
Product Hunt page: https://www.producthunt.com/products/ai-hive
Your turn. Drop:
- What you're building
- Who it's for
- The hardest part you're solving right now
Will read everything and upvote what resonates.
2.4K views
Replies
Thanks a lot @nolan_vu for your great initiative.
Today is my 1st Launch!
Lazar AI, a Visual AI that instantly solves golf on-course doubts with just a photo.
Simple, Isn't it? No text, No speech. Just snap a good photo of your lie on the golf course.
It is for golf players that didn't read the rule book (72% as per surveys) but need the best advice to save strokes.
The hardest part I'm facing now is to fine tune a local model so it becomes more and more accurate.
Check it out on my launch page!
Any feedback is gold for me. You already know that.
Have a great Saturday.
@dani_trylazar Congrats on the first launch! The photo-only input is a bold UX call and honestly makes sense for the use case. Golfers on course don't have time to type anything.
The fine-tuning challenge is the real game though. Curious how you're collecting ground truth data for edge cases like unusual lies or local rule variations.
DS PHP Edit - https://www.producthunt.com/products/ds-php-edit a deepseek v4 powered PHP editor. Made it for my own purposes for doing some smaller php projects up and didn't want to wait for CoPilot to integrate deepseek, so I just made my own more refined version. Its open source, so I'm pretty sure no one cares unless I slap a number on it. Solves problems for me, and that's what I care about. I've also gone the extra mile and slammed it with features that I wanted, and also made sure i can see what is going on behind the scenes easily enough.
@g023 Built it for yourself first, open-sourced it second. That's usually how the most useful tools get made, honestly.
The "see what's going on behind the scenes" part caught my attention. Most editors treat the model as a black box and that gets frustrating fast. Is that more of a debug log view or something closer to a reasoning trace?
Second Brain for AI
What I'm building: Second Brain, an open-source AI memory layer that runs entirely in your own Cloudflare account. Every memory you save automatically links itself to related ones. The graph builds itself. Ask a question and recall expands outward, surfacing context you didn't think to ask for.
Who it's for: Developers who are tired of re-explaining themselves every time they open a new chat. If you've ever had Claude, ChatGPT, and Cursor all confidently remembering different things about the same project, this is for you.
The hardest part I'm solving: Memory that's actually trustworthy. Storing things is easy. Knowing which memory to trust when two of them conflict — and making sure the wrong one doesn't silently overwrite the right one... that's the hard part. v2 has an explicit trust model for exactly that.
https://github.com/rahilp/second-brain-cloudflare
@rahilpirani The conflict resolution problem is the one nobody talks about. Everyone's excited about storing memories but when Claude and Cursor have confidently diverged on the same thing, which one do you trust? The explicit trust model in v2 sounds like the part worth reading first.
Running it on Cloudflare is a solid architectural choice for latency. Does the graph expansion happen at query time or is it precomputed?
What I'm building: Maleu — India's first life OS. A social platform that brings together your scattered digital life: sharing moments, learning, fitness tracking, communities, AI-guided photography, and raw reflections — all in one place with one social graph.
Who it's for: People who feel like different versions of themselves exist across 10+ apps and none of them talk to each other. If you've ever had Instagram for photos, Strava for runs, Reddit for communities, and a notes app for thoughts — and felt scattered — Maleu is for you.
The hardest part I'm solving: Convincing people that "one app for everything" isn't a compromise. The default assumption is that a focused app always beats a bundled one. Breaking that mental model while actually delivering quality across every feature — that's the real challenge.
Solo founder, 22, building from Hyderabad. Would love any feedback from this community.
@dharan_tej_reddy Respect for taking this on solo at 22, that's not a small bet to make.
The "one app isn't a compromise" pushback you're expecting is real, I've felt that exact fatigue with juggling Instagram, Strava, and a notes app for stuff that's honestly all just "my life." The risk is usually quality dilution when you spread across that many features, so if Bloom and Wrex and Circles each hold up on their own without feeling like afterthoughts, that's the whole game. Following the launch, want to see how this lands.
I'm building Margn, a real-time AI-powered profit dashboard for Shopify stores.
It connects Shopify with Meta Ads, Google Ads, and TikTok Ads to automatically calculate true profit after product costs, fees, shipping, and ad spend—all in one clean interface.
It’s built for Shopify founders and DTC brands who know their revenue, but don’t actually know what they’re earning.
The hardest part right now is normalizing messy ecommerce + ad platform data in real time and turning it into a single reliable profit number that founders can trust without spreadsheets or manual reconciliation.
Drop your email to test it risk-free on your store today.
@samuelcrotti Real pain point, props for tackling it head on.
Ran a small Shopify store for a bit and the ad spend vs actual profit math was always a mess of half-updated spreadsheets, never trusted the number by the time I looked at it. If Margn can pull that into one reliable figure without me reconciling three platforms by hand, that's worth a lot more than another analytics dashboard nobody opens. The "normalize messy data in real time" part is the hard 20% though, curious how you're handling refunds and returns messing with the profit calc since that's usually where these tools fall apart.
@nolan_vu Thanks! That pain point is exactly what pushed me to build Margn.
You're also right that refunds and returns are one of the hardest parts. Our approach is to continuously sync Shopify order events and treat refunds as changes to profitability rather than static transactions, so the profit figure updates as the underlying data changes. The goal is that if something happens today, a refund, a return, or an order edit, you shouldn't have to wait for a report or manually reconcile anything to understand its impact.
There's still a lot to refine, but accuracy and trust are the priority. If founders don't believe the profit number, the rest of the dashboard doesn't matter. Really appreciate you bringing this up!
@samuelcrotti you're welcome then, hope that you and your team can manage to fix and alter the product in a positive way to deliver quality and satisfied experience to your users
@tony_kim8 This is such a clean idea
The privacy-first spot thing is the part that gets me. Every fishing buddy I know guards their spots like state secrets, so building that into the product from day one instead of bolting it on later is the right call. Honestly think the "game" layer is what'll make people actually log catches consistently instead of letting it die after week two like most tracking apps do. Gonna check out the beta, good luck with it.
SEOKRATES the SEO toolkit in 13 languages, like semrush but for normal people for normal price..
for all ecommerce owners from small to big. marketers, bloggers.
Find users... Also wanted to have great todays launch here but everyone just upvoting same stuff everyday / its sad for me because i really build that because i hate how much are costs for same platforms...
@hustlerv Really feel this comment, thanks for being honest about it instead of just posting another polished launch update. Building an SEO toolkit in 13 languages at a normal price while everyone else charges Semrush money is genuinely a good mission, that frustration is valid.
Discovery is brutal right now, the feed does favor whoever posts loudest that day, not necessarily the best product. One thing that's helped others I've talked to is going deep into ecommerce and blogger communities directly instead of relying on PH traffic alone, since that's your actual audience.
What I'm building:
reFrame. Most people can't see their own communication patterns until the damage is already done. Contempt, defensiveness, stonewalling, gaslighting. reFrame catches them in real time, both in the message you're about to send and the one you just received, and shows you the pattern before you hit send. There is also built in two way detection to read what someone has sent you. You know, the message you screenshotted and sent to 5 friends and said, "am I the crazy one here?" Someone being gaslit finally gets proof it isn't in their head. Privacy is real so messages are never stored.
Who it's for:
Anyone who's ever sent a text they wanted back. Right now we're going in through divorce and co-parenting, where one bad message ends up in a custody file and the kid is the one stuck in the middle.
Hardest part right now:
Distribution, and it's hard for the same reason the problem is hard to see. Nobody sends themselves an invoice for a bad conversation. You get charged a little at a time, one clipped reply, one eye-roll, one thing you can't take back, and none of it feels like anything until it compounds. Then the bill comes due all at once...the blow up, the tit-for-tat statements, and you're wondering how you got here. The people who need this most are the ones still being charged in small amounts who don't feel it yet. Reaching them before the blowup instead of after is the whole mountain.
Launched on Product Hunt today if you want to see it. No signup, no cost: reFrame: The X-ray of any relationship, situation, or conversation | Product Hunt
@wereframe Thanks for putting this out there, honestly this hits differently than most launches. The two-way detection piece, being able to check a message you received and not just one you're about to send, is the part that feels genuinely useful and not just a gimmick.
The line about nobody sending themselves an invoice for a bad conversation is painfully accurate. That slow compounding is exactly why people can't see the pattern until it's already a mess, especially in custody situations where every message can end up as evidence. Real problem, real product.
If you're open to checking out something else in the AI space, I'm building AI Hive, an automation platform for teams. Would appreciate any feedback or review if you have the time:
What you're building: Naxely — turns a CSV or Google Sheet into a branded PDF report with AI insights, anomaly detection, and charts in under a minute.
Who it's for: Freelancers and small agencies who send the same kind of client report every week and are tired of formatting it by hand.
Hardest part right now: Getting AI-generated summaries to reliably follow a consistent structure (lead finding → context → business implication → one action) instead of reading like a stats dump. Turns out prompting alone isn't enough — ended up moving to a delimiter-based parsing approach so the structure is enforced in code, not just requested in the prompt.
Product Hunt page: producthunt.com/products/naxely
@deepanshu_garg9 Appreciate you being upfront about the hard part, that's the most useful kind of update honestly. "Prompting alone isn't enough" is something more people building with AI need to hear out loud.
Moving to delimiter-based parsing to force structure in code instead of just asking nicely in the prompt is the right call. I've hit the same wall building automation flows, the model will follow instructions 90% of the time and that last 10% is exactly where trust breaks down for clients. Good instinct catching it early. Already upvote and wish your product launching the best
Would love your feedback on AI Hive if you have a minute, we're tackling a similar reliability problem for enterprise agents. A review would help a ton too: https://www.producthunt.com/products/ai-hive
We had 600 clients before we built the app.
My wife runs a trichology clinic in a premium hair salon. Eighteen months of tracking real people's hair and scalp conditions on a medical-grade trichoscope before we wrote a line of consumer code. The thing she kept seeing was blood results the GP had marked normal while the scalp imaging told a different story. No tool existed to connect them.
That gap became Órga Hair. Android is live now.
@gerard_brandon This is such a strong founding story, thank you for sharing it. 18 months of real tracking before writing a single line of code is rare, most people rush straight to building.
The part about blood results being marked normal while the scalp imaging told a different story really stuck with me. That's the kind of gap you only catch by actually watching real patients over time, not something you'd stumble on building in a vacuum. Congrats on getting Android live, that's a big milestone.
Thanks Nolan, that means a lot. You've put your finger on exactly why we did it this way round.
The blood-normal-but-scalp-not moment happened often enough in the clinic that it stopped feeling like an anomaly and started feeling like a pattern worth building a solution around. You're right that it's not something you'd design for on a whiteboard. It only shows up when you're watching the same people come back month after month and the numbers on paper don't match what you're seeing under the trichoscope.
Android live is the first proof point. With more than 600 clients and 18 months of trichoscope data behind it, the pattern wasn't a hypothesis by the time we started building, it was already established. The app is really just the natural next step, taking what we'd already validated in one salon and making it available to anyone, anywhere.
Appreciate you taking the time to read the whole thing.
@gerard_brandon you're welcome, I hope you can keep developing and upgrading the product Orga Hair so clients can experience the best
Already left an upvote to congrat, hope that you can support our AI Hive product too
@nolan_vu done, but also looked at AIHive and it looks impressive. It certainly would have been something I would consider were I back in my executive position