PH热榜 | 2026-04-07

👋 Hey Product Hunt!

I'm Rustam, founder of NovaVoice.

A few months ago, I realized: voice is the interface nature gave us. It's fast, intuitive, effortless. Yet when we sit at our computers, we default to typing — even though speaking is at least 4x faster.

That felt wrong.

So my co-founders and I built NovaVoice — a Voice OS that writes, answers questions, and acts across your entire desktop. Not just dictation.

What makes NovaVoice different:

1. Context-aware writing — speak naturally in your email client, get professional emails. Dictate into Notion, get formatted Markdown. NovaVoice knows where you are and formats accordingly.

2. AI assistant on any screen — hit a hotkey, ask anything by voice. Translate text, get answers, research — no switching to browser or ChatGPT.

3. Voice commands across apps — you're coding. Say "Ask Maria in WhatsApp if design is ready." NovaVoice opens WhatsApp, finds Maria, drafts the message. You just hit send.

4. Smart popover — Reformat any text instantly using preset styles or type any custom instruction on the fly. No switching to Grammarly or ChatGPT to polish your writing.

5. Custom Dictionary — NovaVoice remembers your shortcuts: contacts, addresses, loyalty numbers. Say "email Maria" or "insert home address" — no need to spell things out every time.

6. Cross-language quality — switch between languages mid-sentence. NovaVoice catches proper names, abbreviations, grammar automatically. 

I use NovaVoice daily for prompting AI models, writing emails, code comments, quick Slack messages, repetitive actions in daily apps, and asking the assistant by voice instead of googling — without breaking flow or switching windows.

What's next:

- More app integrations and voice commands

- Near-instant transcription

- Personalization that learns your writing style

We built this for productivity, but realized it has real impact for people with limited mobility. Voice isn't just faster — for some, it's essential.

NovaVoice works on Mac & Windows. 

We'd love to hear your feedback!

I have used Nova Voice (Free Version) and it worked fine with me. I am sharing my experience here:

1. AI assistant is good. It could answer all of my questions. So, it will surely cut one’s time with ChatGPT or similar apps just for finding information or questions answers.

2. I asked questions in English and also in Bengali (my mother tongue) and I was surpised happily with the fact that it worked in Bengali Language and provided me the information I was seeking in my mother tongue.

3. As for transcription, it worked perfectly in English language speech typing or dictation. Auto punctuation is also up to the mark and then after you finish typing you can improve your writing and it also worked well.

4. As for Bengali Language, it worked well if I spoke slowly and clearly but in English, I could speak naturally and it got my words correctly. Well, the thing is that most apps do not provide any quality output in my language at all. So, Nova Voice stands out in this regard in a very nice way. Of course, I want better support.

4. It will surely help writers and bloggers.

5. The only thing that I wish better was speed. I hope that they will improve this in the next version.

Overall, I am satisfied with my experience.

Congrats on the launch!

I'm a big fan of voice and getting things done through dictation. Curious how this compares and differs from @VoiceOS.... which the term itself is also highlighted on your webpage and marketing copy which I found interesting.

been waiting for voice that handles navigation, not just transcription. which apps have you tested beyond the obvious ones?

@rustam_khasanov congrats with the launch! didn't get if it's mac os app or what? does it have full control over my mac os?

Hi PH Community!

I am Anton, co-founder of NovaVoice. Try us and you will see why we earn attention!

As someone who loves WhisprFlow but can't work with it, coz it's not fast and understanding enough (I am suprised too), I am very hopeful here!

I explored your landing page and still have a couple of questions.
- What languages does NovaVoice support?
- Is it possible to use NovaVoice as a speech-to-text tool only, not as an agent? As a replacement for WhisprFlow

Good luck Rustam!

Really impressive work, congrats on the launch! NovaVoice’s transcription is surprisingly accurate. I was especially curious to test how it handles mixed language input, since I often switch between Portuguese and English mid-sentence (which usually breaks most tools), but it handled it much better than expected.

I also like the broader vision of a “Voice OS”, reducing context switching and moving closer to working at the speed of thought feels like a natural next step.

One thing I’d be curious to explore further is how it performs in more complex or noisy real-world scenarios, and how much control users have over formatting and actions. But overall, this is a very promising direction, excited to keep testing it.

This is a nice reminder that I shouldn't always be typing since voice dictation is so much faster. What languages does NovaVoice support right now?

I like the idea of a voice-first workflow, but the real challenge is always consistency once you move beyond demos. Acting across apps, formatting text, and staying context-aware sounds great, but in day-to-day messy usage things usually break or need correction. How close is this to something you can actually rely on without constantly fixing outputs or switching back to keyboard?

Interesting, and looks quite sophisticated!

But it won't be useful for me as I can't conceive what I want to write faster than I can type normally...

I'm using aqua voice. How does your product compare with it?

Congratulations on the launch.

I recently broke my arm, which slowed my writing. The default transcriptionist had all the garbage I'd mentioned, so I had to correct it, and my speed ultimately stalled. Your service helped me get back to my previous pace of communication with partners and maintain my polished style without any additional edits. Thank you!

Congratulations on the launch, team! Super excited to try Nova. I’ve never quite learned to type fast let alone blind and honestly, I’m happy not to have to learn that skill. Typing is one of the main limitations for me and it’s frustrating, so I’m ready to be the #1 fun. Just one quirky question: does Nova support switching between languages in real time? :p

Voice is the way forward. Curious — are you using ElevenLabs under the hood for the voice layer, or have you built your own models? The quality bar they've set is wild and I'm interested in how new voice products are approaching it. Typing is the bottleneck for everything I do, happy to see people building real voice-first tools instead of bolting voice onto a text app.

Is support for other platforms planned (mobile/browser versions)?
Congratulations on the launch btw

My coworkers will definitely think i've lost my mind talking to myself all day:D But 4x speed worth it!

just dropped superwhisper to use this, banger!

I'm an active MacWhisper user and I love the concept. It's definitely worth giving a try!
Congrats guys!

This looks like a massive time-saver. How well does the app handle ambient background noise if I am working in a busy coffee shop or office?

Oh, we put a lot into this one.

Proud of the team. :)

Hope NovaVoice ends up being the thing you reach for every time input text slows you down

👋 Meet Lessie

Hi Product Hunt! I'm Colin, founder of Lessie.

We built Lessie around a simple question: Why is finding the right person still so slow?

Whether it's creators, leads, or candidates, the process is always the same — search across platforms, compare profiles, filter manually, then figure out how to reach out.

So we rethought the whole flow. With Lessie, you can just describe who you're looking for — and it handles the rest: finding, evaluating, and even reaching out. Instead of giving you another list, it actually helps you move from search to connection in one go.

What sets us apart is our State-of-the-Art (SOTA) people search quality. In our head-to-head platform comparisons, Lessie consistently outperforms tools like Exa and Juicebox across Relevance, Coverage, and Utility. By reasoning across the entire web, Lessie identifies the most relevant matches that traditional filters miss, delivering a level of precision and depth that defines the new standard for AI-native people search.

If you're doing:

• 📣 Influencer sourcing

• 🚀 Outbound / Lead gen

• 🤝 Hiring or networking

…or really anyone, anywhere you need to reach, would love for you to try Lessie with a real use case and tell us where it breaks.

We've also open-sourced our core Skills — including our modules for finding people, enriching contacts, and deep company research. We want Lessie to be truly extensible, so we encourage you to plug these skills into your own stack. Whether you're building a custom sourcing engine or a specialized research agent, you can leverage our open-source foundation to automate the heavy lifting.

🎁 Try Lessie today with PH Special: Use code PH50 to get 50% OFF! We're here all day — happy to dive into anything.

👋 Hi PH! I'm Alexia, Growth @ Lessie

Growth starts with finding the right people, but the process is a manual mess. We spend half our day "platform-hopping"—cross-referencing 20+ tabs to verify a single lead.

We built Lessie to kill the tabs. Lessie synthesizes a person's footprint across the web. Whether scouting for partnerships, leads, or niche talent, Lessie handles cross-platform verification in seconds.

Stop the tabs, start scaling. Challenge Lessie today! ⚡️

The useful part here is the context layer. For finding creators who actually fit a niche, knowing why someone matches is way more valuable than getting 500 loose results.

👋 Hi PH! Sharon here, CPO at Lessie.

Most search tools are noisy, static databases. We built Lessie to solve this at the architectural level.

Our AI-native engine reasons in real-time, benchmarked across 119 queries. Lessie leads in SOTA Matching Quality, outperforming traditional tools in Relevance (70.2), Coverage (69.1), and Utility (56.4) by synthesizing identity beyond simple keywords.

Full report: https://github.com/LessieAI/people-search-bench

Search, Reach, Connect. Challenge Lessie today! ⚡️

Hi PH — This is Tina, CMO at Lessie AI

One surprising insight: people search isn't about perfect queries — it's about behavior. Users often write imperfect or vague requests and still get strong results, and a single search usually turns into an iterative exploration rather than a one-shot answer.

A good tool today isn't about helping users ask better — it's about understanding intent, even when it's not fully expressed.

Curious to hear how it feels (and where it breaks). Challenge Lessie today!

This concept is excellent and exactly what I need. Could you use it to source and outreach to suitable YouTube KOLs?

Feels like this could pair really well with founders doing outbound for the first time . Lower barrier to getting started.

Tried once when visiting Bay area with the intention to meet with some frontier lab guys who may have connection with me - Finally turned out to be quite effective. Shout out to Lessie team who built this from scratch.

The personalized outreach follow ups combo is powerful but also tricky .The line between helpful and spammy is very thin here.

This is a meaningful pain point to solve. If I want to identify the right person within a company by the role—but I’m not sure who that would be—is there a reliable way to find them accurately?

Is it something like RocketReach? How the data (+ contact information) for each person are collected?

Tried it out to find developers in a niche I'm working in — described what kind of devs I needed, and it actually found people who are active in that community, not just random name matches. Pretty impressed with the accuracy.

Very interesting for those looking out agentic solutions for sales - congrats on launching!

Will this handle bidirectional sync (with existing CRM) ? Specifically, if a lead is updated in our CRM (say, they are marked "Do Not Contact"), does Lessie's AI agent detect the changes to adjust its own outreach logic?

For me the best part of this category is prioritization. A ranked list of creators with reasons makes the influencer discovery workflow much easier than a raw directory dump.

influencer sourcing is usually a manual nightmare of comparing engagement rates and profile vibes. using ai to evaluate the 'fit' before reaching out could save our agency dozens of hours a week. the ph50 code is a nice touch signing up to test a real use case today.@colin_yu_123

What I like about this influencer discovery workflow is that it seems built around fit, not just volume. I'd rather get a short list of relevant creators than another giant export I'll never touch.

This seems like a great tool for finding loads of influencers on X all at once. Could be a huge time and money saver for marketing campaigns.

Been using Lessie since the early beta, and honestly the “AI-native LinkedIn” angle is a really interesting direction.

Both personally and for our team, we’ve actually put some budget into it and tried to use it in real workflows — not just testing. What stood out to me is how the agent approaches “finding the right people.” It’s not just filtering or scraping, but more like reasoning through who might actually be relevant based on context, which feels like a step forward compared to traditional tools.

Still early of course, but the thinking behind it is solid. Curious to see how the team keeps evolving this — especially around deeper intent understanding and outreach quality. Rooting for you guys 👍

I like products that reduce tab-hopping. Coach and mentor discovery usually means bouncing between LinkedIn, podcasts, and community directories; pulling that into one flow is genuinely useful.

Cool, this product is amazing. Congrats on this launch!

This is the kind of feature that matters because it protects sender reputation, not because it's exciting on a landing page.

This feels closer to a working research assistant for researching accounts before outreach than a traditional database. Nice if you're tired of stitching together LinkedIn, company pages, and scattered notes by hand.

How fresh is the data that Lessie looks at? Selling now is very competitive with a lot of agentic tools performing a similar search. It would help to have fresh data indexed (or looked at) when we perform a search.

Just curious, which use case is driving the strongest retention right now: influencer sourcing, lead gen, recruiting, or something else?

Congrats on the launch! The search + reach in one flow is a nice UX improvement over the usual "export to CSV, paste into outreach tool". What does the fit scoring actually look at? is it mostly job title and company signals or does it go deeper than that?

 I spent 30 minutes with your website and I was really happy. I was, I particularly looked for two information and I was very happy with the results. It has saved me a lot of time.

 I just used the free version so I didn't spend to need any money but it has saved me at least five or six hours of time and energy so I really wish that this service takes off and many people use it I recommend it to almost everyone first give it a shot as You can try free of cost.

Then if you feel that you want to move ahead then do it. But I am sure that it can give something value, give something to almost everyone who would spend just 10-15 minutes here. Thanks a lot.

What I realized is that this website basically does three things. First, it finds me the people that I am looking for. Secondly, it can create and send email on my behalf. Finally, it can reply to the incoming emails for me.

Just tested it with a real search, looked up Nintendo Switch influencers and the results were actually relevant, not just a generic list of gaming accounts.

That's already better than most tools I've tried, curious how it handles more niche searches though, would 'indie iOS app makers who talk about productivity' surface real micro-influencers, or does it work better with broader categories?

Interesting that you mention multi-agent architecture on the site. Does each agent handle a different source like LinkedIn vs Twitter, or is it more about splitting up the research steps?

I can see this being useful when you're moving fast but still want each email to feel considered.

I really like this idea. I struggle with posture a lot too 😅

I’ve tried using a posture corrector, but it can get pretty uncomfortable after a while.

This feels like a much nicer approach.

this is clever - using existing webcam instead of requiring new hardware. how accurate is the posture detection? we work with wearables data all day and know how tricky body positioning can be to get right, especially with just computer vision.

Me and my boyf def need this!

Looks pretty good, will be trying it out pretty soon :)

I'm assuming it will need the camera access all the time, is that safe

This is the kind of product that quietly judges me all day… and I probably deserve it. Love the idea of turning passive webcam time into something actually useful.

We actually launched on Product Hunt yesterday as well — building Ogoron, an AI system that fixes broken test coverage instead of posture. Different problems, same energy: catching issues before they get worse

Not related to the product, but I like your landing page! How did you make it?

This is sick!

Going to try it soon!

Wow, amazing Concept.

Honestly, I just love the reference to the golden ratio. It made my day! Congratulations on the launch!

Didn't see what versions of MacOS you support.

Legend doing legend stuff.

Hey Product Hunt! Mihir here, maker of OpenOwl.

I built this because of a frustration I couldn't shake:

I'd ask Claude to help me with a task, and it would give me perfect step-by-step instructions... that I then had to spend 45 minutes clicking through myself.

LinkedIn prospecting? "Go to this profile, click Connect, type this message." Great advice. But I still had to do all the clicking.

Updating 200 product prices in Shopify? Claude knew exactly what to change. But there's no API for the admin panel. So I sat there. Clicking. For two hours.

AI can think for you. But it couldn't act for you. That's the gap.

OpenOwl is an MCP server that gives your AI assistant actual eyes and hands on your screen. It sees your screen, moves the cursor, clicks buttons, types into fields, and navigates between apps — not through APIs, but through the actual UI, like a human would.

You describe what you need in plain English. OpenOwl does the rest.

A few things worth knowing:

• Works with Claude, Codex, and any MCP-compatible AI

• Runs 100% locally on your Mac — screenshots and data never leave your machine

• Install with one command: npm i openowl

• Free tier included — 50 tool calls/day, no credit card needed

• macOS only right now (Apple Silicon + Intel)

This started as a weekend project to scratch my own itch. Now it's something I use every single day — and genuinely can't go back to doing these tasks manually.

What repetitive screen task would you hand off to your AI if it could actually click for you? Would love to hear what you'd use this for.

This is cool - can you tell me a bit about where it's more powerful than Claude Cowork or Perplexity's "browse for me" functions, or any major LLM with Playwright or similar installed? Again, really cool either way, definitely going to give this a swing! congrats on the launch :D

most workflow automation tools nail the happy path. partial failures are where they fall apart - one step errors, the rest queue up, nobody notices until the downstream data is wrong.

Cool, Mihir! Currently dealing with that problem cause Calude perfectly set a plan for me but then I want to execute it, which should be made for an agent. Happy to see you helping on this!!

Smart idea,I really wanna give it a try.

Congrats on the launch! The local-first approach is smart, and browser automation that doesn't phone home is a real unlock for privacy-sensitive workflows. How does it handle sites with aggressive bot detection?

Is WIN support coming as well?

Hi everyone!

Google is entering the AI dictation space with AI Edge Eloquent, directly taking on tools like @superwhisper @Willow Voice or @Wispr Flow .

It is powered by Gemma-based on-device ASR models to automatically filter out filler words, "ums," and mid-sentence self-corrections. The interesting part here is the flexibility it gives you: you can keep everything local on-device for privacy, or you can toggle on cloud mode to let Gemini handle the text cleanup for much higher accuracy and formatting.

It is completely free, offline-first, and can even pull custom vocabulary from your Gmail. Have been looking for a solid local dictation workflow? This is definitely worth testing out.

I really like it, but they'll need to enable Keyboard setting on it like Wispr Flow and Typeless so we could use it from within the specifics apps vs having to copy/paste. Not fully baked yet, feels rushed. But offline is a great move.

Offline-first is the right call for any docs with sensitive context. How does accuracy compare to cloud Whisper?

Hey everyone! I'm Charlie. A designer and developer, and today I'm launching Letterbox.

I've always loved typography as a visual art form, not just a way to read words. Letterbox lets you explore that idea: each letter on screen is actually composed of hundreds of tiny characters, creating these dense, textural compositions.

You can pick from a curated set of fonts, dial in your colors, adjust fill density and columns, and every unique design is encoded in the URL so you can share it instantly.

It's completely free, no account needed, and works on desktop and mobile. Would love to see what combinations you come up with!

I’m curious if I could control how the internal text flows, like horizontal vs vertical or even curved layouts.

I like how I can play with fonts and colors but I’d love more control over spacing and density to refine the look.

Love the creative angle here .Tools like this make typography feel more playful instead of rigid .Curious how much control users have over spacing and layout.

WHY ARE YOU SO GOOD AT MAKING PRETTY, FUN THINGS.


Congrats on the launch, @charlie_clark! Now.... how do I get these letters onto my projects/sites.

I enjoy how this concept lets me think differently about type design. It’s not just about readability anymore; it’s about texture and personality.

Just had a play around with this for ages. Love the interactive hover on the letters and the more creative concept, feels like the opposite of every generic newsletter tool out there. The little design details matter way more than people give them credit for.

Looks very cool, any idea why it's blurry for me though? (using chrome, win 10):

Hey everyone! Dan here, cofounder of Caret 👋🏼

We kept noticing the same frustration on our team. Typing in Slack, filling out a Jira ticket, writing a commit message, replying to an email. Our brains already knew the rest of the sentence. Our fingers just hadn't caught up yet.
Autocomplete exists in code editors. Everywhere else on your Mac? Nothing. So we built Caret.

What it does: Caret sits invisibly in the background and finishes your thoughts, anywhere you type on your Mac. Press Tab and it completes your sentence. One keystroke. No copy-pasting into ChatGPT, no switching windows, no waiting for a chatbot to load.

How it works:

• You grant one accessibility permission and Caret reads the context of whatever you're working on: the app, the text field, what's on screen

• No screen recording, no screenshots. It reads text, not pixels.

• No integrations, no plugins, no per-app setup. It just works everywhere.

What makes it different: Behind the scenes, Caret turns your sessions into chains of thought, building a second brain that learns how you write, what you're working on, and what you're likely to say next. Those memories are stored locally on your Mac. The longer you use it, the sharper it gets. It stops feeling like autocomplete and starts feeling like an extension of your own thinking.

We're a tiny team and this is a v1. We love feedback and ship fast. Thanks so much for checking us out! 🙏🏼

trycaret.com

Excited

Caret is my new ⌘V

Excited to see how sharp it gets over time. @dschwartz18

Big fan of small teams shipping fast—keep it up!

Been using it for a while and it’s like it’s reading my mind

@dschwartz18 Congratulations. And happy product launch.

every editor plugin promises AI anywhere you type. the value gap is always timing - knowing when to suggest vs when to stay quiet. availability is the easy part.

Congrats @ron_adin1 and @dschwartz18 on the launch! So happy to see Caret on Product Hunt.

I had the pleasure of meeting them about a month ago, and even then the energy and conviction behind Caret was impossible to miss! You could tell this team was building something they truly believe in.

And the problem they're solving is real: AI is everywhere but context stays siloed in each tool. Caret's approach of building at the OS level instead of forcing yet another integration is genuinely clever. And a huge boost for the productivity!

Excited to see where this goes! Hope to be a great journey!

“Autocomplete exists in code editors. Everywhere else on your Mac? Nothing”

https://cotypist.app/

Awkward…

This is such a real pain point -- the context-switching tax is invisible until you actually measure how much time you lose to it. How does Caret handle sensitive fields like passwords or internal finance docs? Curious how you think about the trust layer when it's reading across every app. Also does the 'second brain' piece work across machines or is it purely local?

Congrats on the launch, rooting for you 🙌

This is super cool, we're very excited to start using this at @Basedash

Hey Product Hunt! Community 👋

I'm Nilesh. I was a researcher at Yale University.

I calculated once how much time I spent that year not doing research.

Managing citations. Fixing compiler errors. Reformatting for a different journal. Emailing .tex files back and forth.

It was months.

Not hours. Months of a PhD that I will never get back. I did not want to lose my entire train of thought every single time.

That's why I built Bibby for the scientific community.

What it does:
✍️ Autocomplete that understands your paper as you write

🔁 Reword any sentence with one click

✅ Grammar and academic tone fixes, right where you're typing

🤖 Bibby Chat: an AI co-author that knows your entire paper and helps you draft, restructure, and think

📸 Snap a photo of any equation, get it formatted perfectly in seconds

📚 Add citations from 200M+ papers with one click, no copy-pasting

🔍 Deep research mode: finds sources, spots gaps in the literature, and traces full citation trails

📝 Writes your entire literature review with real cited references

📄 Generates your abstract from your full paper in seconds

📊 Describe your data in plain English, Bibby builds the table

✅ AI paper reviewer trained on top conference standards, NeurIPS, ICML, ICLR, CVPR

🔧 Catches writing and formatting errors in real-time and fixes them with one click

📄 Upload any PDF, Bibby reads and analyses it.

👥 Collaborate with your whole team in real-time, no conflicts, no versioning chaos

🔀 Sync with GitHub, with automatic commit messages handled for you

⏪ Full version history with visual diffs and one-click restore

📋 5,000+ publisher-approved templates including IEEE, Nature, APA, NeurIPS and ACM

🔐 Your research never trains our models. Ever.

Trusted at Yale, Harvard, Stanford, MIT, Oxford and others.

Built by a researcher, for researchers.

How many hours did you waste last month on something that had nothing to do with your actual research writing?

As someone who interact daily with grad students, the number of hours I've watched them waste on LaTeX errors instead of actual research is painful.

Hopefully this will encourage everyone to author more research papers.

Really curious about the photo-to-equation feature. How well does it handle complex multi-line equations? And does it work well with matrices or piecewise functions?

I've been using Overleaf for 5 years and the lack of AI features has been increasingly

frustrating. Every other tool I use has gotten smarter except my paper editor. Excited to

try this. 🎉🔥

This is very interesting, I would be on steroids if this has it's own memory too. So I can fetch context across my research lines.

Curious to know, what LLM is powering the autocomplete and reviewer? Is it a fine-tuned model or are

you using something like GPT/Claude with custom prompting?

If this existed earlier, i might have actually enjoyed my PhD instead of just survivin it) Thats a gem

Co-authoring is often messy — different writing styles, different comfort levels with AI assistance. Can individual authors set their own suggestion preferences within a shared document, or does one setting apply to the whole team?

the real test for a research AI co-author is when your sources directly contradict each other. citation conflicts are where the actual thinking happens.

How difficult is it to edit the equations? Is it possible?

What databases does the citation search pull from? Is it just Semantic Scholar or does it

also search PubMed, arXiv, CrossRef?

This really resonates! writing is honestly the most draining part of research sometimes. Love the idea of having something that not only drafts but actually understands literature and catches issues before reviewers do. Feels like the kind of tool you’d want next to you at 3am when the deadline is way too close.

We launched our own product on Product Hunt yesterday, so we definitely feel the launch day nerves, sending you lots of support. Congrats, this is a super meaningful direction

I work in tech and had no idea researchers still use tools from the 90s to write papers.

This seems like it should have existed 10 years ago. The research community deserves

modern tool

Hey PH! 👋

I built UltraMock to solve a simple problem: making screenshots look cinematic shouldn't require hours in Photoshop or a photography degree.

UltraMock is a browser-based tool that turns any screenshot into a stunning 3D mockup with real depth-of-field blur, the kind of look you see in Apple keynotes. Drag to rotate, pick your blur style (tilt-shift, radial, lens, directional), and export a high-res image. Takes about 30 seconds.

Free to use daily. Pro is 40% off right now at $29.99 once, forever.

Excited to launch beta of video exporting with this launch too.

V2 is in the works already and will be a Chrome Extension that adds cinematic animations to ANY website or APP for instant mockups. 200+ already on the waitlist.

Would love to hear what you think and how you'd use it!

What I like about Ultramock is that when you're building out your own app, you look at it all day and after a while it starts to look boring and you start questioning what you've built..

And then you simply copy and paste a screenshot of it into Ultramock, and all of a sudden it looks really cool and you feel motivated again to keep building!

I've been using UltraMock since the day Josh launched and it's been great to reduce the time needed to create a quick social post to just a few minutes that would have otherwise taken a few different tools (and more time) to pull this off. This is a fun way to experiment with your app/socials.

Also Josh is a badass and continuously shipping great new features all the time and ya'll should support him. Keep up the great work!

Hello! Wow, this looks really great - congrats on the launch! I'm also building a mockup tool, launching soon :)

Hey Product Hunt! Kalyan here, head of content and marketing at sync.

We've been building AI lipsync for a while now, and today we're launching sync-3, our most advanced model release ever.

Here's the short version: previous lipsync models (including our own) processed video in small, isolated chunks and stitched them together. sync-3 takes a fundamentally different approach. It builds a global understanding of a person across an entire shot and generates all frames at once. The result is consistency and realism that closes the gap between real footage and dubbed footage.

A few things sync-3 handles that nothing else does well:

- Close-ups and partial faces (the full face doesn't need to be visible)
- Extreme angles including side profiles, over-the-shoulder, non-frontal
- Obstructions like hands, mics, scarves - detected and handled automatically
- Speaker style and emotion are preserved, not flattened

- Low lighting and varied lighting scenarios

It's 32x larger than our previous model (16B vs 400M parameters), supports 95+ languages, and outputs in 4K.

You can use it right now at sync.so, through our Adobe Premiere plugin, or via API.

We think of this as the leap from perfecting lip sync to unlocking facial reanimation, the model doesn't just match mouths, it understands performances.

Would love for you to try it and let us know what you think. We're here all day answering questions.

Hey Product Hunt!

Super happy with the launch, sync-3 is much more powerful than any previous models we've released, my favorite feature is how you can upload a video with the lips closed and have that be lipsynced without issue and with the highest of quality.

We want you to be able to try it so if you sign up with code SYNC3LAUNCH, you get a free month on the Creator plan and $25 in credits.

Can't wait to see what you create!

95 language in 4K is wild .Feels like this could seriously change dubbing workflows if quality is production ready and not just demo-level.

How are you handling edge cases where emotion and lip movement don’t quite align across languages, especially with big differences in sentence structure?

We've been building lofi.town for over a year now. What started as a little lofi music app turned into something bigger - a virtual third space where people show up every day to work, study, and just hang out together.

The idea came from a simple problem: body doubling works. Having someone next to you while you focus genuinely helps you stay on track. Every app doing it was either paywalled or put you on a video call with strangers, which felt like too much. We wanted something more like your local coffee shop - no camera, no pressure, just knowing other people are locked in around you.

So we built a free multiplayer world where you pick a spot, turn on the lofi radio, and focus alongside real people. We've got customizable Pomodoro timers, ambient sounds, habit tracking. It's all built in and all free. We don't paywall any core feature and never plan to!

Between sessions you can fish off the pier, decorate your burrow, or just hang out with the community. That's what keeps people coming back. We also have users who've been logging in daily for over a year. it's become their spot.

If you want a place to sit down and get things done with good people around you, come check it out! There's always a spot for you :)

p.s. if you're a UX designer, we'd love feedback on our productivity panel. Don't hesitate to reach out :D

-lofi.town team

Steven & Trevor

I was a very early user of lofi.town, and I am STILL using it regularly. The community is growing, but the vibes are still carefully maintained by the team. The updates and new features the team is constantly working to introduce to the community have been amazing!

I started using the app as I was studying for the MCAT. I needed to stay accountable while studying for this massive exam. Being on lofi.town made me feel like I was studying with other people, and honestly, and productivity improved. Months later after taking the MCAT, I’m starting medical school, and I’m 100% sure I’ll be using lofi.town even more now that I’ll constantly be hitting the books!!

I don’t use ALL the features of lofi.town (burrows, etc.), but I think that is what makes the app so approachable and usable for everyone! Whether you’re using it for productivity, socializing, gaming, or listening to music, lofi.town’s vibes are immaculate!

This past few months I've been experiencing a lot of academic changes, finishing and starting new stages in my career, including the need of creating a whole new routine that I can actually stick to and help me thrive and get a job. In the middle of all of those changes, I reached Lofi town. I thought of giving it a try, the indie game aethetic gave me a cozy feeling of those games I used to play as a child, but with the addition of task managment tools that help me as an adult. The thing is not only it helped me a lot with all of the above (specially with body doubling), but also got me in a community that basically shape the app as something new and different; the "third space" in the description definitely makes a ton of sense.
I work from home, which sometimes can feel a bit lonely, but is the community in lofi town (and the involvement of the team with said community, listening and learning from them on what the app needs) that pushes me to work a bit harder and feel seen at the same time.
I hope to stick around for a long long time and I'd love to see more apps and tools made by you guys;) I never get tired of telling you; great job!

Been using lofi town for a couple months, and it has been a game changer for me. Especially with recent additions being added to see stats and track daily habits. As someone with ADHD and Autism, it has given me a sage space to keep up with my work and keep it feeling rewarding while helping me build good daily habits.

Body doubling with actual people has been such a game changer. And you never feel forced to talk to anyone, you just can if you want to. The mini games are also cute, fun, and not hard to do. They are also completely optional. This is very worth it!

The idea of focusing alongside real people without the pressure of a video call is something I really get - it's the digital version of studying in a library.

I've been building SelfOS - a minimalist life planner (tasks, habits, goals) with a similar philosophy: no pressure, no guilt, just gentle support. Even added a little bonsai tree that grows with your streaks 🌳
Totally different angle but same energy.

Curious do your users who come for the focus sessions also engage with the habit tracking? Or is it mostly the coworking vibe that keeps them coming back?

This honestly feels like my kind of space: cozy, a bit playful, but still about actually getting things done. I already opened it and can totally see myself working there with lofi in the background. Also had a phase of making lofi music myself, so part of me is like… I could’ve been on your radio

We actually launched our own product on Product Hunt yesterday, so we really feel the launch day nerves — sending you lots of support. Congrats, this is a beautiful concept ✨

Hi everyone!

Aytekin here, founder and CEO of Jotform. We are excited to share Jotform ChatGPT App!

We’ve all seen how powerful LLMs are at answering questions. But at Jotform, we believe the next step for AI is helping you ask them. Jotform ChatGPT App turns passive conversations into active data collection.

We built this, because:

ChatGPT is incredible for generating ideas and text. But when it comes to structured data, like collecting registrations, feedback, or orders, you usually have to leave the conversation and open another tool. We wanted to change that.

What it does:

Jotform ChatGPT App allows you to manage the entire form lifecycle conversationally:

- Build: "Create a feedback form for my beta launch" generates the form instantly.
- Refine: "Help me edit this question" lets you tweak it naturally.
- Analyze: "Summarize recent submissions" gives you a visual snapshot of your data.

It maps every interaction to a clear, single action , so you never get lost.

Who it’s for:

- Teams needing quick polls or internal feedback.
- Researchers creating surveys on the fly.
- Event organizers tracking RSVPs without leaving the chat.
- Power users who wants instant insights from ChatGPT by using the collected data.

What's so special:

It transforms ChatGPT into a real-time dashboard. By connecting conversational AI with Jotform’s persistent storage, your chat isn't just temporary text, it becomes an ongoing way for tracking business data over time.

Hope you like it, let us know what you build! 🚀

Hi Product Hunt 👋

ClipMark started from a simple idea:

everything I copy should still be there when I need it.

Sometimes I want to search and organize things — with tags, filters and context.

But most of the time, I just want them back instantly.

That’s where Quick Recall comes in:

hold a shortcut, scroll through your clipboard with previews (great for screenshots), release → paste.

ClipMark combines both:

fast recall when you need speed, and powerful search when you need control.

Happy to answer any questions 🙂

— Mario

first time launching here… didn’t expect this at all.

really appreciate the support so far 🙏

Hey Product Hunt. I'm Ken, a 20-year-old Stanford CS student. I built Sup AI.

I started working on this because no single AI model is right all the time, but their errors don’t strongly correlate. In other words, models often make unique mistakes relative to other models. So I run multiple models in parallel and synthesize the outputs by weighting segments based on confidence. Low entropy in the output token probability distributions correlates with accuracy. High entropy is often where hallucinations begin.

My dad Scott (AI Research Scientist at TRI, PhD from UCLA) is my research partner on this. He sends me papers at all hours, we argue about whether they actually apply and what modifications make sense, and then I build and test things. The entropy-weighting approach came out of one of those conversations.

In our eval on Humanity's Last Exam, Sup scored 52.15%. The best individual model in the same evaluation run got 44.74%. The relative gap is statistically significant (p < 0.001).

Methodology, eval code, data, and raw results:

• https://sup.ai/research/hle-white-paper-jan-9-2026

• https://github.com/supaihq/hle

Limitations:

• We evaluated 1,369 of the 2,500 HLE questions (details in the above links)

• Not all APIs expose token logprobs; we use several methods to estimate confidence when they don't
  

We tried offering free access and it got abused so badly it nearly killed us. Right now the sustainable option is a $10 starter credit with card verification (no auto-charge). If you don't want to sign up, drop a prompt in the comments and I'll run it myself and post the result.

Try it at https://sup.ai. My dad (@scottam) is in the thread too. Would love blunt feedback, especially where this really works for you and where it falls short. If Sup ends up being useful, we added a Product Hunt offer that expires in a week: 20% off your first month with code PRODUCTHUNT.

If you're unsure what I meant by entropy and output token probability distributions, whenever an LLM outputs a token, it's choosing that token out of all possible tokens. Every possible output token has a probability assigned by the model. These sum to 1, forming a probability distribution. APIs typically return these probabilities as logprobs (logarithms of the probabilities) because raw probabilities for rare tokens can be so small they underflow to zero in floating point, and because logprobs are the natural output of how models actually compute their distributions. We use these directly to calculate entropy. Entropy is a measure of uncertainty and can quantify if a token probability distribution is certain (1 token has a 99.9% probability, and the rest share the leftover 0.1% probability) or uncertain (every token has roughly the same probability, so it's pretty much random which token is selected). Low entropy is the former case, and high entropy is the latter.

There is interesting research in the correlation of entropy with accuracy and hallucinations:

• https://www.nature.com/articles/s41586-024-07421-0

• https://arxiv.org/abs/2405.19648

• https://arxiv.org/abs/2509.04492 (when only a small number of probabilities are available, which is something we frequently deal with)

• https://arxiv.org/abs/2603.18940

• tons more, happy to chat about if interested

Looks interesting, especially when you dealing with sensitive docs and information. The pain is real for me.

Quick tech question: do you use deterministic functions, like code execution, calculation, aggregation, SQL, etc, or is the output completely determined by the LLM consensus and confidence score?

I'm asking because for number sensitive tasks, raw LLM output even with consensus, usually does not match the reliability of using deterministic tools. How do you solve this?

The Token Expenditure might be really high here! How have you optimized Sup AI for that?

I built Highlight Studio because every screen recorder I tried was either too simple (no editing) or too complex (full NLE for a 2-minute demo).
I wanted one app that records, edits, and exports something polished — without switching tools.

What it does:

- Smart Zoom — AI watches your clicks and generates zoom keyframes automatically. No manual keyframing.
- Virtual backgrounds with custom images
- Camera branding
- Cursor effects — 3-level smoothing, post-recording resize, click sounds, and visual highlights.
- Brand kit — Logo, colors, fonts, watermarks, and 6 animated intro/outro templates. Set once, apply everywhere.
- Device frames — 27+ pixel-perfect Apple devices with real finish colors. Record your iPhone via USB and wrap it instantly.
- AI transcription — Word-level subtitles in 50+ languages.
- Multi-track editor — Trim, split, speed ramp, freeze frame, annotations, blur masks. Separate tracks for video, camera, audio, subtitles.
- AI agent support — 40+ CLI commands over TCP. Works with Claude Code and any AI agent framework.

Everything runs on the GPU. Under 100 MB app size.

It's native Swift, renders on Metal, and the entire editing pipeline is GPU-accelerated. Smart zoom alone saves me 20+ minutes per video.

Lifetime access is $55 during launch (60% off). No subscription required.

Highlight Studio offers extensive Help Center with AI chat bot that will answer all of your question about product features and how-tos.

Video has been recorded, and edited only using Highlight Studio it self.

Really like this: the native Metal approach + no Electron makes a big difference for video workflows, and keeping it under 100 MB is impressive. Smart Zoom based on user interaction is also a great touch — exactly the kind of automation screen recording tools need.

We launched on Product Hunt yesterday as well — building Ogoron, an AI system that generates and maintains test coverage as products evolve. Different space, but very aligned in spirit of automating the tedious parts :)

Native Swift + Metal rendering, under 100MB, no Electron — you're speaking my language. I'm building a Mac-native video editor with the same stack (Swift + Metal + Rust core) and the GPU-accelerated pipeline makes all the difference for real-time playback. Smart Zoom from click tracking is clever — that's the kind of feature that only works well when you have direct access to system-level events, which is exactly why native beats web wrappers. Great launch.

I like all the personalisation for interactions. It seems a lot more in depth and user friendly than previous screen recorders I've used! I'm building marketing videos for a product I'm working on and always end up frustrated by the polish gap between "raw screen recording" and "finished clip."

Does Highlight bridge that, or is it still more of a recording tool than an editing one?

The Smart Zoom feature alone would save me so much time, I've been manually keyframing zoom on every app demo I record for social.

As an indie iOS maker about to launch, I need exactly this: record, edit, and export without juggling 3 tools.

Two questions: how does it handle portrait mode recordings for iPhone app demos, and does the device frame work when recording directly from Xcode simulator?

How are you finding the balance between automation (like smart zoom and AI edits) and giving users control, especially for people who want more precise editing?

Something similar could be created for board games. Actually a good idea :)

OMG I love this.

I do a lot of remote movie nights with friends and it's always kind of hacky. This feels like what we needed. Really excited to try it out. Thanks for making it happen🙌

Finally something that doesn’t feel like a work meeting. Been using Couchspace with family and it’s the closest thing to actually hanging out together. Smooth, cozy, zero setup. Highly recommend for anyone doing long-distance with people they care about.​​​​​​​​​​​​​​​​

I just checked it out and it looks really nice!

I’ve always had issues with platforms like Netflix blocking screen sharing — how are you handling that?

Oh wow, that's so interesting. My bestie and I live live in different countries and can't see each other at all, but if this tool help us hanging out together (watching series, or movies) - that will be so cool. Can't wait for the launch to test it

I’m honestly impressed that it works offline. I travel a lot with my kids and I’ve struggled to find apps that don’t rely on constant internet.

How often will new games be added to keep kids engaged over time?

Hi everyone!

Khan Academy Kids has a very interesting competitor...!

Netflix Playground is a standalone kids games app for children 8 and under, included with a Netflix membership, with no ads, no in-app purchases, and offline play. That alone is already a pretty strong product shape for parents.

What makes it more interesting is that Netflix is clearly trying to turn its kids IP into a full watch-and-play loop. Peppa Pig, Sesame Street, StoryBots, Dr. Seuss — instead of just watching those worlds, kids now step into them.

Hey Product Hunt!

After having fun with SlapMac last week, I wanted to take the idea further — so I built MacYaps, a menu bar app that makes your Mac talk back with funny sounds (or serious if you wish) for real events like low battery, WiFi drops, charger changes, high CPU, and USB connections (more triggers to come).

Multiple voice packs + full custom sound support included.

Short demo found here: https://macyaps.com/demo.

This is my first Product Hunt launch — would genuinely appreciate any feedback. Or let me know if you'd like any other triggers in the app.

Discount Code: `PHLAUNCH` to get it for $4.49 (limited to 50 activations)

The voice packs sound like the main attraction .Savage New Yorker reacting to to low battery is something I didn’t know I needed 😂

Do you plan to add context-aware responses, like different tones depending on urgency?

do you think you could also add triggers for when mac sleeps/wakes up?

Very smart! Love the idea! Congratulations on the launch!

idea: you could check the item's charge cycle count and use that to find optimal charging patterns

i was ready to ignore another ai design tool until i saw the part about encrypted secrets and local context. one quick question "does it pull from the global design system file automatically or do i need to point it to specific files? @bowlcutwiz @AIDesigner

skeptical about fully autonomous UI generation - every agentic system I run hits edge cases where the output breaks existing design contracts. what's the human review point before it commits changes?