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
We're building VIQI AI an AI research assistant for the global media & entertainment industry.
It's designed for producers, distributors, studios, investors, and content executives who need answers to complex industry questions without spending hours digging through multiple sources.
The biggest challenge we're solving is turning fragmented entertainment market data into accurate, actionable insights. Whether it's discovering production companies, finding co-production partners, tracking acquisitions, or researching companies, VIQI AI helps users get answers in minutes instead of hours.
Powered by the global entertainment intelligence behind Vitrina AI.
@rutuja_kokate Sounds like an interesting use case, especially because entertainment market research is usually spread across too many sources and takes a lot of manual context-building.
What has been the hardest part to get right so far - finding the right data, keeping it accurate and up to date, or turning fragmented information into insights that are specific enough for producers, distributors, and investors to act on?
@rutuja_kokate Entertainment market data being scattered across too many sources is such a real problem, producers and investors end up spending hours just piecing together context before they can even ask the real question. Turning that fragmentation into answers in minutes instead of hours sounds like a genuine time-saver for that industry.
Valo is an open-source AI security platform that helps organizations discover, govern, and secure AI applications, SaaS platforms, and autonomous agents. Instead of relying solely on detection, Valo combines deterministic policy enforcement, AI-native risk analysis, and enterprise-ready reporting to help security teams reduce risk while enabling innovation.
Our mission is simple: make enterprise AI secure by default.
We're excited to launch the Community Edition, gather feedback from practitioners, and build alongside the cybersecurity community. Whether you're a security engineer, AI developer, CISO, or founder, we'd love to hear how you're securing AI today—and what challenges you'd like Valo to help solve.
Thank you for checking us out and helping shape the future of AI security! 🚀
Website: valosecurity.ai | GitHub: github.com/ValoSecuritySuite | Contact: info@valosecurity.ai
@valosecurityai "Secure by default" is the right framing and honestly the thing most teams skip until something breaks. The gap between detecting a risk and actually enforcing a policy is where a lot of AI security tools fall flat, so the deterministic policy enforcement piece is what stands out to me here.
Excited to see where the Community Edition goes, especially with autonomous agents getting more traction in enterprise.
@nolan_vu Feedback welcomed! Thank you, @nolan_vu
@valosecurityai thank you too, hope that you can produce other good story to follow
@valosecurityai Valo looks very relevant, especially now that AI usage inside companies is spreading.
Curious what kind of feedback you’re hoping to get from the Community Edition first - more from security teams trying to govern existing AI usage, or from AI builders who want to make their own apps safer before they reach production?
Building Faultline Security: manual penetration testing and AI red teaming for startups and SaaS companies. Fixed price, human-verified findings, delivery in days instead of weeks.
Who it's for: teams that need a real pentest for SOC 2, ISO 27001, or GDPR and are stuck choosing between €15k+ enterprise firms with junior testers, automated scanners that miss business logic, or unpredictable bug bounties.
The hardest part: starting from zero, and how little security awareness most small teams have. Most founders don't think about pentesting until an enterprise deal or an auditor forces the question, so half the work is education before it's even a sales conversation. Building the trust and traction from nothing while going up against that awareness gap is the real grind right now.
@beatriz_albernaz Stuck between expensive enterprise pentest firms, automated scanners that miss business logic, and unpredictable bug bounties is such an accurate description of the gap right now. Half the work being education before it's even a sales conversation is a grind a lot of early-stage security founders don't talk about enough, respect for pushing through that.
Building Snipplet, a travel app that turns your messy screenshots, blurry menu photos, and scattered Notes app lists into beautiful visual cards you can actually share.
Who it's for: Anyone who travels and loves making things look beautiful. Think Pinterest meets travel guides. If you're tired of losing your best recommendations across 5 different apps and a camera roll with 600 unorganized photos, this is for you.
Hardest part right now: Getting people to shift from their messy-but-familiar system (camera roll + Notes app chaos) to something intentional. The product works, but the habit change is the real challenge.
snipplet.com
@isabelzav 600 unorganized camera roll photos is way too relatable, I've lost so many good travel recs to scattered screenshots and half-written Notes lists. Getting people to break that messy-but-familiar habit is honestly the harder problem than the tech itself.
Hope Snipplet makes that shift easier for more travelers, upvoted and following.
@nolan_vu Thank you so much for the feedback and support, Nolan! Would love to hear what you think if you give it a try! And incredible work with AI Hive, keep it up. Followed and upvoted as well!
memi
We launched Memi today. It is a macOS workbench for product design teams where Claude, Codex, and Hermes can run against your specs, research, and Figma files. The thing I am most curious to learn is whether teams want one best agent or a way to compare multiple agent outputs side by side
@sarveshsea
Multi-agent comparison on the same spec is the more interesting bet long term, even if "one best agent" is the easier sell upfront.
The real question is probably what teams do once they see Claude, Codex, and Hermes disagree on the same input. Do they treat the disagreement as signal (something in the spec is ambiguous) or just noise to be averaged away? That answer probably changes how you'd design the comparison view.
Also curious how deep the Figma integration goes, is it just pulling specs as context, or can agents reference actual component structure and naming when they reason about the design?
What are you building
neuphlo.com, an agentic workflow management system, built and hosted in Europe. We offer a self-hosting option as well. It is BYOM, so the user chooses whatever model of LLM, Cloud or local they want.
Who it's for
Users who do not want to pay premium for tools, where they have to adapt to the tool instead of the tool adapting to their ways of working. Typically small/medium teams in IT development, support management, campaign offices etc. The list of use cases is long.
The hardest part solving
Still working on getting seen in this huge ocean of apps in the category.
There is a free tier and a teams tier, 12 EURO /month when paid monthly and 10 EURO/ month when paid yearly.
Product Hunt Page: https://www.producthunt.com/products/neuphlo
@jan_sorensen The BYOM approach is exactly what serious teams want right now. The pushback against vendor lock-in is getting louder and most platforms are still pretending it's not a problem. Self-hosting option on top of that is a strong differentiator, especially for European teams dealing with data residency requirements.
The "getting seen in this huge ocean of apps" challenge is real and probably the hardest part for any agentic workflow tool right now. Have you tried niching down hard into one specific vertical first instead of going broad? That's usually what breaks the visibility wall faster than competing on features in a crowded category.
@nolan_vu No, the thing is we dont know where to start and how to get seen.
@jan_sorensen honestly that's the most common place to be and nobody talks about it enough. most advice assumes you already have a clear ICP and a channel that's working, which isn't helpful when you're still figuring out where you even fit.
one thing that helped me think through it: forget distribution for now and just ask which specific type of user would be genuinely annoyed if your tool disappeared tomorrow. start there, go deep on that one person, and visibility gets a lot less abstract.
@jan_sorensen I noticed that your product launched quite a few months on PH but no upvote or followers, so I hope you can improve such stats later on the following launches
Don't mind if you can take a look at our AI Hive and share some feedback on it: https://www.producthunt.com/products/ai-hive
@jan_sorensen How much setup do early users usually need before they can get their first useful workflow running?
What I'm building:
Resume MCP — an AI agent that reads a job description, tailors your LaTeX resume to it, writes the cover email, and sends it directly from your own Gmail. No third-party platforms. Your real inbox. Full automation via MCP (Model Context Protocol).
Who it's for:
Developers and job seekers tired of mass-applying with the same generic resume. Engineers who want to apply smarter, not harder.
The hardest part I'm solving:
Getting LaTeX generation to produce ATS-friendly, beautiful PDFs reliably — while keeping the MCP tool chain stable end-to-end without any human in the loop.
Check it out the link from bio and posts— would love to hear how others are solving the resume tailoring problem! 🚀
@anup_ojha Love the angle of cutting out third party platforms entirely and using your own Gmail. The MCP automation chain is what gets me though, that's the part most resume tools skip because it's the hardest to make reliable. ATS-friendly LaTeX generation alone is a problem people have been struggling with for years.
One question on the workflow side: how are you handling cases where the job description is too vague or generic to actually tailor against meaningfully? That's usually where automated resume tools fall apart in my experience. Will check out the link, sounds genuinely useful for the right audience.
@nolan_vu Hi its lovely to know that you like the concept and product
So we are handling the job description cases if anything vague comes up we don't tailor the resume we simply ask the user your score is low you can either update your resume or still can apply without tailoring the resume
So this is how we handle the case
@anup_ojha that's actually a cleaner fallback than most tools go with. instead of hallucinating a tailored version, you just surface the score gap and let the user decide. less impressive on paper but way more honest in practice.
the one thing i'd watch is whether users understand why the score is low when the JD is vague. if they don't, "your score is low" just feels like a rejection, not useful signal.
@nolan_vu Hi yeah cleaner and providing best experience is what we do and also for the
score we give proper reasons for how and why he/she gets the low score.
i will suggest you to try it once and have a look for more questions you questions will help us in improving and getting better day by day
Databox
Building the Skills Marketplace by Databox - a free library of plug-and-play AI analytics skills and workflows.
Who it's for: marketers, founders, and agencies who need fast, confident answers from their data - GA4, Google Ads, Shopify, LinkedIn, Stripe, and more - without building custom reports from scratch.
The hardest part we kept running into: generic AI fills data gaps with plausible-looking nonsense. It doesn't know your lead definition, your MRR calculation, or how you attribute revenue. So we built skills that run against live Databox data via MCP - the AI gets your actual metric definitions and real numbers, not its best guess.
Result: analysis you can share with your team without having to verify it first.
Launching on Product Hunt June 30 - happy to share a link if you want early access.
@zigapotocnik the MCP angle for live data is what makes this actually different. generic AI giving plausible-looking numbers is a real problem and most people don't catch it until they try to present it to someone who knows the actual figures.
"analysis you can share without verifying first" is a strong line btw, that's the real pain point for anyone who's been burned by AI making up metrics. following the June 30 launch.
anyway congrat on your upcoming launch, I wish to examine further on that day
@zigapotocnik Interesting to see analytics data becoming usable directly in AI tools! congrats :)
Love this thread! 👏
We're building Myspec, a tool that helps founders and developers turn rough product ideas into structured, build-ready specifications before jumping into AI coding tools.
We're solving the problem of unclear requirements and missing context, because we've learned that AI can generate code incredibly well, but it can't fix unclear thinking. Our goal is to help builders start with clarity, not just better prompts.
🌐 https://myspec.dev
🚀 https://www.producthunt.com/products/myspec
We're still improving the product every week, so if this sounds interesting, we'd genuinely love for you to give it a try. Any feedback, whether it's something you love, something that's confusing, or something you'd build differently, would mean a lot to us.
Looking forward to hearing your thoughts and checking out everyone else's projects too!
@nancyy_choo Thanks for the kind words on the thread, and genuinely nice to see Myspec here. "AI can generate code incredibly well but it can't fix unclear thinking" is such an underrated point, most people blame the model when really the brief was never clear to begin with.
Turning rough ideas into build-ready specs before touching a coding tool is exactly the missing step in most AI workflows I've seen. Will check it out, curious how it handles specs that keep changing mid-project since that's usually where documentation tools fall behind reality.
Would love your feedback on AI Hive too if you're open to it, an automation platform I'm building for teams. Any review helps a lot:
@nolan_vu Thanks so much, Nolan! That really means a lot.
You're absolutely right, we've found the biggest bottleneck is usually the context, not the model. That's exactly what we're trying to solve with MySpec.
À nếu đúng là người Việt thì xin chào cậu nhé!! I already upvoted AI Hive on Product Hunt! I also read the story behind it, and I really like how it evolved from ChatX. That's such a cool journey. I'll definitely spend some time trying AI Hive and share my thoughts soon.
Wishing you and the team a fantastic launch! 🚀
@nancyy_choo Chào cậu, tiếng Việt là một dấu hiệu rất chill khi bắt gặp trên PH, cảm ơn cậu nhiều nhé. Nghe cậu và team nói về việc context, không phải model, là bottleneck lớn nhất mà cậu đang giải quyết với MySpec, mình cực đồng cảm. Đó đúng là gap mà rất nhiều team đang cố dùng LLM để cover nhưng bị quên phần context.
Cảm ơn cậu vì đã upvote AI Hive và tìm hiểu cả story đằng sau nữa, mình rất trân trọng điều đó. Rooting for MySpec trong hành trình tiếp theo, và mình sẽ theo dõi progress của cậu. Nếu có dịp gặp offline hay online tiếng Việt thì gọi mình nhé, luôn thích được kết nối với người Việt cùng đang build trong không gian này 🚀
T đã để lại review cho sản phẩm MySpec nhé, mong là sản phẩm sẽ phát triển tốt trong tương lai
Building Analyse (analyse.net)
What: a unified growth platform combining cookieless product analytics, an AI SEO content engine, and a data copilot in one dashboard. It also ships an MCP server so you can pull your live analytics into Claude or Cursor.
Who it's for: product teams who want Mixpanel-grade funnels without SQL, and content teams who want SEO on autopilot instead of managing four separate tools.
Hardest part: getting people to believe one tool can genuinely do all three well. The default assumption is that bundled products are mediocre at everything. So we have to earn trust on each pillar individually before the unified value clicks.
@wesley_breukers Thanks for laying this out so clearly Wesley, the framing around "earning trust on each pillar individually before the unified value clicks" is honestly the sharpest thing I've read about bundled products in a while. Most teams pretend the unification is the sell, but you nailed why that rarely lands cold.
The MCP server for pulling analytics into Claude or Cursor is a really smart move for developer-heavy teams. Cookieless product analytics is also going to age well as third-party tracking keeps tightening. Rooting for you on this one.
Wishing Analyse continued growth and hope you keep making users happier as it scales. Just upvoted and followed to stay on top of updates 🚀 If you have a sec, I'd love a look at AI Hive too, a platform I'm building for enterprise AI agents and automation flows: https://www.producthunt.com/products/ai-hive. Any upvote or review really helps our small team.