PH热榜 | 2026-05-12

Hi Product Hunt 👋

I'm Sachin, co-founder of Kelviq.

For the past few months, we've been building Kelviq, and we're excited to finally share it with you.

Kelviq helps SaaS, AI, and digital products sell globally without rebuilding billing infrastructure from scratch.

It handles payments, checkout, global taxes, compliance, subscriptions, digital file delivery, license keys, usage-based billing, credits, feature access, and payouts in one place.

We built this because we kept seeing the same problem through our other product, ParityDeals.

Adding a payment gateway is easy. Building the full monetization layer around it is not.

Once you start charging for features, seats, usage, credits, API calls, or AI tokens, you usually end up building a lot of custom infrastructure:

• Webhook listeners to sync payment events into your database

• Subscription tables to track who has access to what

• Retry logic for missed or failed events

• Custom usage tracking

• Feature flags tied to pricing plans

• Scripts for upgrades, downgrades, prorations, and grandfathered users

Over time, pricing becomes tightly coupled with your codebase. Changing a plan, moving a feature, adding usage limits, or launching a new pricing model often requires engineering work, database changes, and a deployment.


We wanted to make all of this much simpler. With Kelviq, you can:

• Get started in minutes: You don't have to handle dozens of webhooks, build a database layer to track customers, manage subscription states, or hardcode feature access. Kelviq provides a single, unified system.

• Handle Global Tax & Compliance Automatically: As your Merchant of Record, Kelviq takes on the full liability for global sales tax (VAT, GST, etc.), fraud, and regulatory compliance. You can sell in 100+ countries without ever registering for a foreign tax ID.

• Run any pricing model: Flat fee, usage, seats, tiers, volume, credits, pay-as-you-go, overage charges, or hybrids. Configure them in minutes.

• Sell digital products with ease: Securely deliver files, e-books, or software on purchase and automatically generate license keys to validate customers.

• Meter usage in real time: Count every API hit, data storage unit, or AI tokens in real time. Set soft or hard caps, pick reset cycles, send alerts, and handle overages with one call.

• Control feature access: Define features per plan. Turn them on or off and check access in your app with one SDK call.

• Localized pricing by country: Offer country-specific prices in 135+ currencies with VPN, proxy, and fraud protection built in. You can also set up promotions for specific countries.

• Ship pricing changes without touching code: A/B test plans, roll out changes, or update pricing in real time without redeploys.

Manage the full customer lifecycle: Handle upgrades, downgrades, migrations, overrides, and grandfathering without writing custom scripts.


We are also running a special founder offer right now:
2.9% + 40¢ per transaction for your first $5K in volume. After that, just 3.5% + 40¢ per transaction.

We'd love your feedback, questions, and suggestions.

Thank you for checking out Kelviq 🙏

@alokvats finally its coming out.. so happy for you guys. Kelviq fills such a big gap for us solopreneurs. Keep going 🚀🚀🚀

Congratulations on the launch!

Big fan of your work Sachin, congrats on the launch!

Id use this over Stripe.

I would use it for my next side project ❤️

3.5% flat pricing is interesting curious how margins scale.

Congrats on the launch 🙌

Does Kelviq handle reverse-charge VAT for B2B sales within the EU automatically (valid VAT ID = no tax charged)?

Looks neat. would try it for upcoming side projects! all the best @sachinchoolur

Cool stuff. Congrats on the launch. I hope Kelviq gets featured.

Congratulations with the launch Sachin - been following your and Kelviq's journey. The speed at which you're shipping things is really incredible! All the best 🚀

THIS IS AWESOME!! need this for getoverlay.io

Congrats on the launch! As a founder dealing with stripe integration right now, the idea of having payments, tax, and compliance handled in one place is really appealing. How does migration work for teams already on Stripe?

Let's go Sachin!!

Being able to change pricing without digging through billing code sounds really nice 😂

This stuff always gets messy once products start growing. Congrats on the launch!🚀

Curious how it handles Apple/Google IAP alongside web payments most billing tools are built around web-first and mobile app billing is always an afterthought. Does it support App Store subscriptions?

I think your logo looks very familiar :)

Congratulations on the launch!

Congrats on the launch! Handling global tax is often the biggest hurdle for early-stage SaaS expanding internationally. How does Kelviq handle tax remittance for specific regions like the EU (VAT) compared to US sales tax? Is the 'digital delivery' feature integrated with the billing logic to automate access management upon payment?

Congrats on the launch 🚀 The traction today is incredible. Curious — what was the unlock that pushed you past 200 upvotes in just a few hours? Trying to learn from the best today since my own launch is happening 😅

This is really cool 😎

Congrats on the launch! Kelviq looks promising

Curious about the 3.5% + 40¢ number — that's a hard line in the sand vs. Stripe's 2.9% + 30¢, but you're bundling sales-tax remittance and global compliance into the same blob. For SaaS founders the implicit framing matters: is Kelviq pricing as a flat replacement for Stripe + TaxJar + Paddle MoR, or is the AI/usage-billing primitive the part you'd argue is worth a different cost basis?

In my world (project-finance modelling, where I run ModeLoop on the side), the cleanest valuation conversations always start by separating commodity-line items from differentiated ones. The bundling decision tends to be where founders either communicate or obscure unit economics. Where do you see customers landing on that vs. the underlying CPM/CPM-equivalents?

Cool, I use polar and migrate to stripe. how do you handle that kinda low cost? 3.5$ is most low MoR I've ever seen.

what is different form Kelviq and paypal?

Hey Sachin, congrats on the launch! Many Indian founders have been struggling with Paddle and Lemon Squeezy suddenly closing accounts or having strict entity requirements. Since Kelviq supports individuals, can you clarify how you handle payouts for Indian users (e.g., via Stripe Connect or local bank transfer) and if there are any specific compliance hurdles/GST requirements we should be aware of to ensure account longevity?

Vibe coding blindly builds apps you don't understand.

Static tutorials teach with throwaway apps.

With Open Vibe your agent tutors you while you build and ship your SaaS using the tools and concepts of the pros.

What Open Vibe does:
🛠️ Sets up your local development environment for agentic coding for you
😎 Guides you through building full-stack apps (Auth, Payments, AI-integrations)
💻 Explains complicated processes like Git, Deployments, Databases, HTTP, etc along with animated diagrams
🧑‍🏫 Acts as your personalized tutors as you build and ship your own app ideas

Who it’s for:
🧑‍💻 Vibe Coders
🏗️ Aspiring Founders/Entrepreneurs
🔬 Learners experimenting with AI

Quick links
Website: https://OpenVibe.sh
GitHub: https://github.com/wasp-lang/shi...

The biggest problem with vibe coding right now isn't that people can't ship - it's that they ship without understanding what they shipped. You get a working app and zero ability to debug, extend, or make architectural decisions when the AI gets confused. Curious about how it handles the teaching moments. Does it pause and explain concepts in context (like "here's why we're using this pattern"), or is it more of a structured curriculum that happens to use your project as the canvas? Those are very different learning experiences.

This feels like a bridge between learning and doing finally.

Congrats on launch! Wasp is a wonderful framework and perfect for something like this. And, really cool to see a teaching tool!

super cool! Understanding the codebase is one of the biggest issues with vibe coding and it's the reason why most people get stuck. I'd love to see how this looks on a bigger codebase

Tried it and it delivers. Biggest thing for me: instead of blindly approving every Claude Code suggestion, you're walked through what's actually happening and why it matters. Awesome launch, Vince. 🚀

Super excited to see Vinny shipping this! I think that having a tutorial embedded in the agent makes the whole experience so much more interactive. Plus, you can actually build your own thing while learning about all the web dev concepts you need to ship without getting stuck. Huge if true :)

I was just looking for a product like this yesterday! (and honestly, briefly considered building it :P)

I thing this is great! Learning even just the basics of web dev is a big help and will make it much easier to get further along with your app.
Btw I was a bit confused with the messaging because at one moment it is about not getting stuck, another it is about learning, I guess those are connected but also don't have to be, so I am not sure what is this more about, I guess about learning, and not being stuck is just a consequence?

Putting two things together - vibe coding and learning how to develop and code is awesome. Congratulations on the launch!

The "getting stuck" problem is real. For me the bigger issue wasn't technical it was showing up the next day after a bad session. I've been experimenting with extremely small daily logs (like, just one sentence about what I built or unblocked) and it's been weirdly effective at keeping the streak going. Something about having to write even one line means you have to do at least one thing. Congrats on the launch the Claude Code integration angle is smart.

Well done @matijash 🙌

"Without getting stuck" is doing a lot of work in that tagline what's the most common place devs get stuck that this specifically solves? Architecture decisions, boilerplate, or something else?

This is a fascinating concept for a 'tutor' in the IDE. How does Open Vibe maintain the balance between giving the developer the 'answer' and actually teaching them the underlying concepts? Does it support multiple LLMs besides Claude, or is it specifically optimized for Claude Code's capabilities?

I like the landing page, we should have you doing our redesign.

Looks pretty sweet. The UI for visualising the data flow is the detail beginners need. It's sometimes easy to get lost in the word soups that LLMs give you... Draw a picture for me and I'm happy. Good job Vinny!

Really interesting approach to AI-assisted development.

Been watching Vinny cook on this one! Just a bit of knowledge can go a very long way when vibecoding effectively, and Open Vibe gives you that, hands on.

Hello Product Hunters 👋

I’m Jeeva from Juspay, the team behind Hyperswitch, the open-source payments platform with 42K+ GitHub stars.

Today, we’re launching Prism: an open-source library that gives developers one common integration layer across payment processors like Stripe, Adyen, Braintree, Checkout, and others.

Most teams start with a single payment processors. That works well early on. But as the business grows, payments get more complex:

Early growth: new markets need new processors, and customers expect local payment methods.

Scaling up: finance teams want pricing leverage, engineering teams need reliability, and maintaining payment processor integrations becomes ongoing work.

Prism helps teams avoid locking in to a single payment processors from the start.

With Prism, you get:

• one schema across multiple payment processors

• a stateless integration layer

• no database and no stored PII

• SDKs for Node, Python, Java, and Rust

• open-source flexibility without infra setup or sales conversations

Our goal is simple: make payment processors integrations easier to build, maintain, and change over time.

When did payments start getting operationally complex for your business?

Congrats on the launch! I like the stateless approach - makes sense for teams that don’t want a full payment orchestration platform yet. One thing I’d look for in a business app is good examples around retries, idempotency, and status changes.

Do you have sample flows for failed payment -> retry -> reconciliation?

Congratulations on the launch @Juspay Hyperswitch team!
This is exactly what we need. Please add golang sdk as well (our stack is on golang)

Since it's polyglot ready, do you provide unified SDKs that maintain consistent error handling across all supported languages, or is the logic primarily at the library level?

How do you handle tracking events and reconciliation if Prism itself doesn’t store anything?

The stateless approach here is pretty interesting. A lot of teams just want the flexibility layer without adopting a whole payment orchestration stack.

Feels like a clean middle ground. Congrats on the launch!

Let’s go!!

The plug-n-switch primitive maps almost perfectly onto how project finance modelers handle module swaps — you don't rewrite the model when the offtake structure shifts from PPA to merchant; you swap the revenue module while the rest of the workbook keeps integrating cleanly. Hyperswitch's redundancy-and-routing layer is the same idea applied at the payments edge. The hard part in both worlds is contract surface: making the swap zero-rewrite means the interface has to anticipate every parameter the new module might want.

This is also the half of project-finance template design that's painfully underrated. I host a small library of these on Eloquens (https://www.eloquens.com/channel/samir-asadov-cfa) and the templates that get reused most aren't the ones with the fanciest formulas — they're the ones where the module boundaries hold up under change. Curious how you decided on what's inside vs. outside the Prism interface contract — did you back-solve from the messiest real-world processor migration?

The "copy any design" angle is bold how does it handle dynamic or interaction-heavy sites? Curious if it captures animations and hover states or mostly static layout.

while the idea is nice, and jotform itself is great, I don't realize how you come up with this need.
Were users asking for working directly in claude or is just a new distribution channel?

Forms-in-a-chat-window is the right shape because the brittle part of form-building was never the fields — it was the loop: build, preview, find an edge case, dig back to the right field, repeat. Collapsing that into a conversation kills the tab-switching latency, and the implicit memory of the conversation context turns "add validation to the phone field" into a one-line ask instead of a five-click navigation.

Where I keep landing on these AI-in-existing-tool products is that the win shows up in the everyday flow, not in the marquee demo. I've been building a small AI-powered weekly meal-planning PWA called DishRoll on the side, and the same pattern holds: the value isn't "generate a 7-day plan" (that's a demo); it's "swap Thursday for something high-protein after my workout shifted" mid-week. Curious how Jotform handles state continuity across a long-running form-build conversation — do you expose the form-graph back to the model after each edit, or rely on the chat history as the source of truth?

yesterday i spent 30 mins on a multilingual contact form generating realistic test data. invalid Polish phone formats that pass length checks but fail country-code regex were what kept slipping through (libphonenumber catches it cleanly, but it's overkill for a contact form). with conditional branches test-data setup eats half the session.

Hey everyone!

I'm Ott, co-founder of display.dev.

A month ago @carlrannaberg came to me with a problem. In his constant use of Claude Code, his agents were building him beautiful HTML artifacts – spec sheets, interactive plans, reviews, etc. Sharing them with colleagues, however, was a mess – screenshots or PDFs to Slack, having others open HTMLs and run them on localhost – nothing good really.

The problem clicked immediately. Anyone building with agents long enough hits this wall.

So we built display.dev – one command publishes any HTML or Markdown artifact behind your company's auth. Your team signs in with Google, Microsoft or a one-time password. They see the artifact exactly as the agent built it. No static and inconvenient screenshots, no GitHub accounts, no $320/month Vercel add-ons.

Here’s what makes it actually useful day to day:

> One command, one click or your agent does it for you – CLI, web app or MCP. Publishing happens wherever you already are – your terminal, your browser or inside Claude Code/Cursor/etc. You get back a permanent URL.

> Gated by default, public when you want – Google + Microsoft SSO or OTP. Everyone at your company gets in, nobody outside does. And if you want to share something with the public, simply change the visibility setting.

> Comments your agent can read – Teammates drop inline comments. You and your agent can read them, update the artifact and resolve the thread. The thing stays alive instead of dying as a one-shot output.

> Your published artifacts are natively agent-readable – agents can pull the content as markdown from any published link, so there's no manual back-and-forth copy-pasting or importing-exporting.

> Stats and audit logs – View counts per artifact, plus audit logs of exactly who accessed what. Useful when you actually need to know if your exec opened the doc.

> Publish without an account – agents can publish unauthenticated using “curl”. The response is a public preview URL anyone can open and a single-use claim URL. Later, you can claim the URL for the organization, if you sign up or in.

> Unlimited viewers, flat price – No per-seat cliff when you share with your PM, exec and legal team on the same day. (Also, a free and a solo tier exist!).

We've been using display.dev daily ourselves – privately sharing analysis docs and ideas that Claude has built, collaborating on things fast. It's made a real difference to how we work.

Happy to answer questions – on the product, the problem and alternative solutions!

I like how this is both secure AND lightweight, a rare combination! Cool stuff!

This is a perfect utility for the current 'coding agent' era. Being able to instantly host agent-generated artifacts behind SSO is much cleaner than passing around raw HTML files. Does display.dev support custom domains for these permanent URLs, and is there an option to expire links automatically for temporary artifacts?

Awesome that someone finally built this!

I'm curious about how display.dev handles authentication for different enterprise setups. Is it using OAuth, SAML, or something else? Auth integration is often trickier than it first appears, especially with legacy systems in the mix.

The screenshot-or-PDF-to-Slack pain point is the right thing to fix — agent-generated artifacts only feel like artifacts when they have a stable URL someone can come back to. The SSO/OTP gate is doing two jobs at once: keep strangers out, and (more subtly) tell the agent which audience it's writing for, which I think is underrated.

I hit a related shape building PolyMind, a small alert tool that watches Polymarket and pings me when a position shifts in a market I care about. The artifact problem there is the same: an alert is only useful if it lives somewhere it can be revisited and annotated by the trader after the fact — a one-shot Slack ping is essentially fire-and-forget. Curious whether display.dev's comment layer is intended to drive iteration on the artifact itself, or whether it's more of an asynchronous review channel for humans to triangulate around what the agent produced.

Once the user is logged on, can their identity be passed so that when we fetch the data, we can customize it?

PS: Please add a "Contact us" page :)

Hey, nice tool.

I just tested it on my website maamria.com.

My site is still new, and Google Search Console is currently showing that only a few pages are indexed. I have around 75 pages not indexed, including many marked as "Discovered – currently not indexed"

Your tool gave my site a score of 31/100 and marked it as critical. It also detected things like very low visible content, only one heading, no external links, and no JSON-LD.

So my question is does this result look normal for a new website that Google has not fully indexed yet? And do you think the score from your tool matches what Google Search Console is showing?

Also, from your point of view, what should I fix first to make the site easier for AI search engines like ChatGPT, Claude, and Perplexity to understand?

Thanks, I’m testing the tool and it looks useful.

Really interesting tool. The recommendations actually feel useful and well-reasoned, not the usual "keep your title under 80 characters" type of advice. Picked up some new insights I hadn't considered before.

Special shoutout to the "Copy Fix Prompt" feature, that's a genuinely clever idea. Makes it super easy to take the suggestions straight into your workflow.

Nice work!

Hey Yahia, congratulations on the launch!

I tried your tool on vaulternal.com and got an 83/100 - which is actually a pretty solid score. The main area that seems to need improvement (according to your tool) is the "Citation Surface Presence" metric, so I'll take a closer look at that. You can see the audit here: https://www.freeaiseoaudit.com/audit/vaulternal.com

Out of curiosity, are you planning to monetize the tool at some point? Having "free" in the domain name might make that a bit tricky down the road.

Good luck with the project!

Bookmarking this — I'm literally in the middle of SEO auditing my own launch today. Will run it through. Congrats 🚀

Tested Free AI SEO Auditor on CambrianEdge.ai and got a solid 93/100.

Really liked the concept overall, especially the “copy full prompt” feature. Makes the workflow much smoother when you want to iterate quickly instead of rewriting things manually.

One feature that could make this even more useful would be the ability to export/download the report as an image or PDF. Feels like something teams would want to share internally or attach to audits.

Great work on this, @yahia_bakour3

This is super relevant for us. We have 30+ programmatic landing pages (airport-specific) and we've been optimizing purely for google. Never thought about how AI search engines parse them. going to run our pages through this today. does the scoring weight structured data / schema markup heavily, or is it more about content signals?

Hey I just found out you website is pretty good and shows amazing resutls.
I just tested my website getdetach.com and it shows results that I really want for my site. Now I will improve my SEO and will again test it.
Thanks!

Hi Yahia, really glad I came across your tool — just gave it a spin on my site falcondrivelabs, which I know could use some SEO polishing. Curious to see the findings and recommended fixes.

One quick general question: when a free tool runs AI-powered analysis, I always wonder — who's footing the bill for the tokens?

This is great timing — I've been thinking a lot about how AI search changes the game for niche tools like mine. I built YTubViral (14 AI tools for YouTube creators) and realized halfway through that optimizing for traditional Google SEO wasn't enough anymore. Creators are starting to ask ChatGPT and Claude things like "best tool for YouTube titles" instead of searching Google.

Quick question: does the audit cover structured data and schema markup, or is it more focused on content signals? That's been the hardest part for me to get right.

"AI search era" is the right framing most SEO audits are still optimized for traditional search and ignore how LLMs crawl and surface content. Does it check for things like llms.txt and structured data specifically?

I just tested my website Clearlist.me and it got 82 out of 100 and got really cool tips on how to even improve more. I do like how easy it is to just copy the recommendations and hand them to my agent.

no more ai slop, just build you nice website with no code !

ask us anything you wants about whale, we are ready to answer.

The “copy any design” part is pretty wild 😂

Can already see this being useful for quickly testing landing page ideas instead of rebuilding everything from scratch.

The "copy any design" angle is bold how does it handle dynamic or interaction-heavy sites? Curious if it captures animations and hover states or mostly static layout.

it would be great to have drag and drop feature for a component, right?

also, do you have a converter from joomla/wp templates or to figma?

Hey! Good job. It may sound weird, but I didn't get how to use the tool. Any plans to add user onboarding?

This is a good one. Voice training angle is what separates this from the generic tools. Training on actual past issues allows to pick up jargon, quirks and sentence patterns. Looks like the right approach! To stress-test - how it handles newsletters that deliberately vary their tone issue-to-issue? Does the model average out your style?

The "in my voice" v. "more human" piece of the prompt is so key. Especially as we see more and more second and third human AI layers like Sinceerly and others start to pop up. Now it just needs the originality that the author brings to the table to be fully autonomous

The "learns from your archive" part is clever. I've seen so many newsletter tools that generate generic content — having it adapt to your voice is what makes it usable long-term.

I'm curious about the learning curve. How many past issues does it need before the suggestions start feeling like "you"? I run a bilingual blog (Spanish/English) for a YouTube creator tools platform and voice consistency across languages has been my biggest challenge.

Curious about the voice training minimum. how many past issues does the model need before the output actually sounds like you and not generic AI? most tools I've tested fall apart under ~20 examples

Concentrating a full day labor into about 5 minutes while keeping the operator's voice intact for that price is something! Reach out to mete '' at '' heybe . ai for partnership opportunities with a great product like HeyNews!

The "in your voice" part is what makes this interesting to me. I've tried AI draft tools before and the problem is always that they smooth out the rough edges that made my writing feel like me. Curious how HeyNews handles that does it pick up on style over time or is it more template-based? The Substack import is also a smart touch for people with an existing archive. Congrats on the launch.

Congrats on the launch — the “review-before-send only” choice feels important. For newsletters, the hardest part isn’t just generating words in a familiar style; it’s preserving editorial judgment when a story should *not* sound like the last 20 issues.

Curious how you handle deliberate voice shifts: e.g. a more urgent issue, a more personal note, or a sponsor-heavy edition where the usual cadence needs to bend without becoming generic?

Been watching Cagri and the team build this and it's exciting to finally see it live. The detail that sold me: they used HeyNews to ship 550+ of their own issues before opening it up. That's the kind of dogfooding that shows up in the product. Rooting for you all today!

The "step 3 sounds like AI" failure mode is the real wall for any voice-as-output product — generic LLMs produce competent-but-anonymous prose, and once a reader has caught that smell they can't unsmell it. Training off the archive solves a different problem than prompt-engineering does: it makes the model write like a specific person rather than like "a knowledgeable newsletter writer in general," which is the difference between something a reader subscribes to and something they unsubscribe from.

I hit a parallel problem on the audio side running ModeLoop, a small podcast on financial modeling. When I tried AI-generated show notes early on, listeners called them out almost immediately — not because they were wrong but because they didn't sound like me. Eventually I stopped using them. Curious how you handle the cold-start problem here — how many archive issues does HeyNews need before the voice match feels native to a reader who knows the publication well?

Nice!

Why the 1GB limit?

Hey everyone 👋

We built TrackTalent because job hunting for engineers feels unnecessarily chaotic today.

People track applications in spreadsheets, miss great opportunities, apply to hundreds of irrelevant jobs, and often never hear back. At the same time, referrals and insider knowledge are still one of the biggest advantages in tech hiring, but they’re hard to access if you don’t already have the network.

At Outtalent, we’ve helped engineers land 200+ offers at companies like Google, Meta, Amazon, Stripe, and more. Along the way, we built internal tools to manage applications, referrals, opportunities, and interview pipelines more efficiently. Eventually we realized this needed to become its own platform.

TrackTalent combines:
• application tracking
• intelligent job matching
• community sourced opportunities
• referral workflows
• a structured pipeline from application to offer

The goal is simple: help engineers spend less time on chaos and more time on the right opportunities.

Would love to hear your feedback, ideas, and thoughts ❤️

Recruiting tools that don't feel like surveillance are rare. Curious how you balance candidate trust with employer visibility. Solid launch 🚀

Hi everyone. I’m thrilled to share TrackTalent with you - our job tracking platform!

When I joined Outtalent, this was my very first project: built to provide our clients a smarter way to manage their job search end-to-end. Over the past year I’ve continuously iterated on this platform - adding more features, better workflows and smarter automations aimed at actually helping our clients land jobs. Since launching it internally, the volume of our clients’ job applications and successful referrals has grown tenfold, reflected in the offers they continue to land.

The team and I are now excited to share the platform with the public. You can use it to explore tech job listings and track your applications. If you become part of our Fellowship accelerator program - you will also get access to intelligent job matching, mentorship and referrals to the top tech companies (FAANG & adjacent).

TrackTalent has been my baby since day one, and it's exciting to finally see it fly. Hope it helps you land something great.
I’m happy to answer any questions!

Congrats on the launch! The job market for engineers is quite noisy right now, so a tool that helps centralize applications is very timely. Does TrackTalent offer automated 'follow-up' reminders for applications that haven't responded? Also, how does the 'intelligent matching' handle niche tech stacks compared to broader roles?

Something similar for marketing doesn't exist? :)

Hi everyone!

MiniCPM-V 4.6 is a 1.3B open MLLM for image and video understanding, built for phones and consumer-grade hardware. It is the smallest MiniCPM-V model to date, and probably the cleanest efficiency play in the series so far.

Visual understanding can get expensive very quickly, especially with high-res images, video inputs, and on-device use cases. MiniCPM-V 4.6 focuses on making that workload lighter, faster, and more practical to deploy.

It also has a pretty complete developer path: mobile demos across iOS, Android, and HarmonyOS, Apache-2.0 weights and code, quantized versions, and support for frameworks like vLLM, SGLang, llama.cpp, and Ollama.

Small multimodal models are getting a lot more interesting when they are designed around real edge constraints!

Hey Product Hunt!

I'm Igor, and today I'm launching Habitvs - a tiny virtual pet for iPhone and Apple Watch whose mood is shaped entirely by your real Apple Health data.

Why I built it
Most health apps shout numbers at you. Rings. Charts. Streaks that punish you for one bad day. I wanted the opposite - something that just cares. So I built Kova, a pixel companion who quietly reacts to how you're actually living. Move and sleep well, and Kova thrives. Burn out, and Kova gets a little worried.

How it works
Kova has 5 moods - Lively, Rested, Balanced, Attentive, Concerned - driven by your activity rings, sleep, HRV, resting heart rate, mindfulness, daylight, and even wrist temperature. No goals to hit. No guilt trips. Just a small friend who mirrors your wellbeing.

What's inside
• Native iOS + watchOS apps (SwiftUI + HealthKit)
• Home Screen, Lock Screen, and Apple Watch complications - including a pixel-art "Retro" style
• A weekly letter Kova writes to you, generated on-device with Apple's Foundation Models - nothing leaves your phone
• Streak system with a weekly "shield" so one missed day doesn't break your momentum
• Milestones at 3, 7, 14, and 30 days - including a crown for Kova
• Customization: 7 colors, hats, glasses, bowties

Privacy
Your health data never leaves the device. No accounts. No tracking. Ever.

I'd love your honest feedback - what feels delightful, what feels missing, and what would make Kova feel more alive. This is just day one.

Thanks for stopping by, and go check on Kova.

Hi Igor!

First off, I love the philosophy behind Habitvs. As someone with ADHD who appreciates a low-stakes way to stay mindful and active with some extra gratification from it, the idea of a virtual companion that mirrors your wellbeing rather than shouting numbers and guilt trips at you is so refreshing. The pixel-art complications are a nice touch.

I spent a little time with my Kova today (I named mine Goofy!), and I want to share some creative input that I hope will provide you with some decent feedback.

Onboarding Experience

Feature Tour - I think the app could benefit from a more in-depth guided tour upon first opening. A step-by-step walkthrough with info blurbs highlighting specific features would help new users feel oriented right away

Clarity on Stats - For example, I found myself wondering what exactly the energy bar represents at first

Deepening the Connection with Kova

Visual Evolution - I'm interested to see how Kova could change visually based on health data. Things like lack of sleep/not meeting a sleep goal reflect with little eye bags or other features of fatigue.

Moodlets & Check-ins - Having small "moodlets" appear while the app is open or doing occasional mood check-ins with Kova would make the relationship feel more interactive and two-way if that is your goal

Gamification & Customization

Activity-Based Rewards - I love the current customization, unlocking even more colors, hats, or accessories by hitting personal or app-based activity milestones would be a great motivator and gratifier. I think it would also increase the longevity of a user who may want to "collect" such items, especially if any were limited edition or only obtainable by specific goals and/or activities

Mini-Games - Adding small, health-related mini-games, like a 1-minute breathing exercise, brain-boosting games, or a quick movement challenge, could give users more ways to interact with Kova during the day

Streaks

If you're interested in leaning even further into the "no guilt" mindfulness vibe, consider a system that doesn't just protect streaks but prioritizes rewarding momentum. Continuing to reward positive progress, even if a day is missed, keep the focus on the journey rather than just a number without eliminating it completely. You already lean into this with the weekly shield, a feature I really like!

the on-device Foundation Models for the weekly letter is a smart call - that's exactly the kind of feature that would feel creepy if it phoned home and delightful because it doesn't. how did you decide which health signals to weight most heavily for Kova's mood?

Looks pretty cool im gonna try it. Im creating a video for my new app i made and i really dont wanna do it manually, this looks very promising!

one of the best mac app i have in recent time (plus its designs are lit 🔥)

Hey, I love Klack - what's the difference? Why should I consider to switch to Keeby? :)

Recently I saw friends and people with incredible stories losing a project they did with great effort because of data leaks and an unsafe app! This is happening more and more! and looking at the audience of founders who use AI to create apps, I feel that we already have a lot of demands to deal with, from creating the app, marketing, customer acquisition, etc., one of the pillars is a safe and scalable project! So it was based on this pain that we created DeepFrame, an agent system focused on security, from end to end to reveal the biggest points of security flaws in any application! and one of the biggest differences is that there is no code base required, just submit your product link and authorize the service.

Visit https://deepframe.xyz/startups to join the startup program and receive security support.

The fact that no code base is required is a huge plus—just submitting a link and authorizing the service makes deep pentesting feel way less intimidating for fast-moving startups.

this looks slick!

Hey Product Hunt 👋

Today we’re launching Hopper, the first agentic development environment for mainframes.

Mainframes running on COBOL are the 60-year-old computing platforms that still quietly run much of the modern economy: banks, payments, insurance, airlines, government systems, and more.

But they were built for expert humans using terminal screens, function keys, batch jobs, datasets, and highly specific workflows, not AI agents.

Modern AI coding tools assume GitHub, shells, files, package managers, and test runners. Mainframes are a completely different computing paradigm.

Hopper combines a real mainframe terminal, context panels for datasets and jobs, and an AI agent that can safely operate across mainframe workflows.

Our goal is simple: bring AI agents to the legacy systems that still run the world without pretending they are modern codebases.

You can request access to a mainframe on our page, and start playing with Hopper to see what an agentic mainframe environment feels like.

Would love to hear any feedback or comments!

Never thought I would see a COBOL AI agent, this is awesome!

Hey Product Hunt! 👋

I'm Alex, the maker behind Vexilo.

Two months ago I started using Claude Code seriously and hit a wall immediately — not with the AI, but with discoverability. I kept finding powerful features by accident, months after I should have known they existed.

/context alone would have saved me hundreds of dollars in wasted tokens. I found it on week 6.

So I spent 3 weeks mapping the entire ecosystem. What started as a personal doc became Vexilo — an interactive, bilingual index of everything Claude Code can actually do.

What makes it different:

• Organized by what you're building, not alphabetically

• Auto-generates a CLAUDE.md you drop into any project

• Works offline (PWA)

• One-time purchase, no subscription

Use code PRODUCTHUNT for $5 off today.

Happy to answer any questions about Claude Code or how I built this. What's the most useful Claude Code feature you discovered way too late? 👇

Beyond happy that we’re finally sharing Seer with everyone.

As someone who has always been more into hardware than software, I’ve spent years wanting to build useful physical things and constantly running into the same exhausting process around it all. And with deadlines and time pressure, you end up settling for less than what you originally imagined.

That always felt wrong to me. Ideas are valuable. They shouldn’t be abandoned just because the process of building them is too difficult or fragmented.

So seeing Seer not only solve that frustration for us, but also help other inventors bring their ideas to life, has been incredibly meaningful.

Really excited to hear what everyone thinks.

While most agentic platforms today are focused on automating software development, Seer is focused on something else entirely:

Hardware.

Seer is built for inventors, makers, and hardware visionaries who have ideas but lack the specialized skills needed to bring them to life — things like CAD design, circuit design, prototyping workflows, and manufacturing logistics.

Our goal is simple:

Remove the barriers between imagination and physical creation.

With Seer, users can:

• Generate hardware concepts
• Create CAD and electronics workflows
• Manage BOM sourcing and component ordering
• Access prototyping and manufacturing tools
• Perform market and patent research
• Prepare patent filing documentation

And we don’t stop at the prototype.

We want to accompany inventors from idea → prototype → production → protection.

Software had its “vibe coding” moment.

We believe hardware’s next step is:

“Vibe crafting.”

Would love to hear what the Product Hunt community thinks about the future of AI-native hardware creation.