PH热榜 | 2026-08-20

Hey PH Fam,

Here we are again! When we set out to build HyNote for Mac, we kept hearing the same two frustrations from users about standard AI note-takers:

First, people hate sending invasive bots into meetings and uploading confidential audio to third-party cloud servers. Second, most tools are built only for live meetings, ignoring the rest of how we actually work.

We built HyNote Mac to solve both:

• 100% Private & Local 🧠: All speech-to-text processing happens on-device using your Mac's local hardware. Your meeting audio, financial discussions, and internal strategies never leave your machine—no cloud uploads, no privacy compromises, and no monthly server-cost markups. You get FREE & Unlimited local transcription without recurring subscription fees or per-minute caps.

• No Meeting Bots 🤖: HyNote captures system audio directly without sending an awkward "AI Notetaker" account into your Zoom, Google Meet, or Microsoft Teams calls.

• The Multi-Source Model ✨: Unlike standard meeting recorders that only listen to live calls, HyNote is a complete knowledge engine. It seamlessly ingests and synthesizes multiple inputs—live meetings, local audio files, PDFs, whiteboard photos, YouTube links, and web clips—stitching them all into one searchable, AI-powered second brain.

Download Now on Mac!

Whether you're bringing a complex PDF into a live meeting synthesis or organizing research across media, HyNote processes it locally, securely, and instantly.

Give HyNote Mac a try today and take back full control of your meeting data and daily context. We’d love to hear your thoughts!

I've always found meeting bots a little distracting. a local transcription tools feels a more natural way to capture conversation.

Congrats on the launch!

Regarding the 100% private and local claim if I have a meeting with a client, does HyNote capture audio from both sides of the conversation? and do I need to get consent from all participants before using it?

On-device transcription for meetings is exactly where this category needed to go, privacy shouldn't be a paid tier 🙌

Really like that it's local-only and free. The cross-device sync makes me curious though: what actually leaves the Mac when it syncs? Does it pass through your servers?

I could see this being especially useful for regulated teams where sending meeting audio to another service is already a headache.

The no cloud stepup iis what caught my attention. It feels much more comfortabke for sensitive work conversations.

Battery usage would be my main question with on device transcription. How lightweight is it when running quietly during a full workday?

Most notetakers meter transcription by the minute. Yours is free and unlimited because it never leaves the Mac. Congrats on the sixth launch and on still giving it away.

If this is free then what's the catch?

Love that HyNote keeps speech-to-text fully on-device, so sensitive meetings never touch a third-party cloud. Privacy-first note taking done right.

Huge congrats on the launch! "Free local transcription that is 100% Private" hits on such a massive pain point right now as privacy-conscious teams look for bot-free meeting tools. Love seeing native Apple Silicon tools that run speech-to-text entirely on-device without locking core utility behind hefty subscriptions. Wishing you tons of momentum today!

I tried Turkish, but it didn't generate a transcript.

Slick product! Particularly appreciate the focus on privacy & local processing which are too often underutilized for use cases like this.

the 100% local processing is a massive differentiator especially with all the recent lawsuits around AI note takers and consent issues. HyNote directly addresses the elephant in the room that many tools are ignoring. excellent work!

The biggest win for me is that it stays out of the meeting itself. I dont want a transcription tool becoming another participant everyone has to notice.

How do you see people using HyNote differently from traditional note apps once all their meetings, documents and ideas live in one place?

Congrats @mia_hello and team!

Curious how good the local transcription is compared with cloud models. privacy is great but accuracy is the thing I’d notice immediately.

This tool seems sleek, simple, and useful.

My main concern is how the security/compliance claims are presented:

• “SOC 2 Type II infrastructure” is materially different from the product/company itself having a SOC 2 Type II attestation.

• “HIPAA-aligned” is materially different from “HIPAA compliant,” yet the badge shown below says “HIPAA COMPLIANT.”

Could you clarify what has actually been independently assessed or audited? As currently presented, I think a prospective customer could reasonably interpret these claims as broader compliance certifications than they actually represent.

I'm always annoyed by random notetaking bots the coworkers keep invite to meeting. But I'm excited to try this since it runs locally on my mac + 100% private.

Does everything stay on the device after transcriotion too, or are summaries sent to a cloud model?

I like that this solves the awkwardness rather than adding another meeting participant. the less people have to change theur meeting habits the better.

I would use this more for client calls than internal meetings. keeping those conversations completely on device would remove a big concern for me.

The no-bot approach is underrated. Having a random transcription bot appear in the meeting is still a little awkward especially with external clients.

Local transcription is the part that would make me try this. Meeting notes can contain pretty sensitive stuff so keeping audio off third party servers is a meaningful difference.

Hey PH fam 👋

Wanted to bring the latest Grok 4.6 launch to this global builder community today!

Here’s the pattern we keep seeing across the agent ecosystem: most models start strong on step one, but fall apart by step five. Real product building requires long-running endurance—researching unfamiliar domains, structuring applications, and iterating through real feedback loops.

Most AI tools act like a quick-burst sprint. Grok 4.6 is built for the marathon.

It doesn’t just generate code; it sustains context over long horizons and self-verifies its work before moving on.

What stood out most to me:

→ Full-stack first passes: Turns broad ideas into structured, visually polished interactive apps in a single pass

→ Autonomous self-testing: Verifies its own work across codebases before proceeding to the next step

→ Iterative depth: Stays in the loop to refine, debug, and polish based on complex feedback

Big shoutout to the xAI team for pushing the frontier on agentic stamina 🙌

I hope this is not the wrong place to ask about this, but since it's related to the usability/capabilities of Grok, I'll go for it.

I'd love to use Grok, particularly due to its low cost yet high efficiency. However, I currently depend on "tools" or integrations that OpenAI provides but need to be customly developed for Grok (for example, adding issues to GitLab).

An alternative would be, since I use an introductory agent that chooses a sub-agent depending on the requested task, to use Grok as a sub-agent for off the shelf stuff.

Is there any cheap(er) way I can orchestrate this (substituting my introductory agent, I guess?, and route the request either to OpenAI or to X).

Congrats on the release. The upgrade landed and the rate stayed at $2 and $6 per million tokens. Nice to see a better model that doesn't cost more to run.

Saw the news on X (like everywhere on X). Didn't expect to see it on PH 🤔
Does it/you guys need more exposure for the model?

I'm curious to see how it's going to deal with some heavy and complicated questions.

The focus on long running agents is what stood out to me. Better reasoning is important, but reliable execution over longer workflows is where AI starts becoming genuinely useful. Congrats on the launch!

I've always felt Grok is a bit more free compared to other tools. That long-term context is exactly what’s needed when building and iterating on full app experiences. Excited to give this version a spin!

👋 Hey Product Hunt, I'm Gal, founder and CEO of Checksum.

A few years ago at my last startup, I watched our team lose entire sprints to test maintenance. Every time the product changed, someone had to go update selectors, re-triage failures, and figure out which broken tests were real bugs and which were just noise. I'd spent years before that building ML models to detect suspicious activity from satellite data—pattern recognition at scale—and it nagged me that software testing was the same kind of problem.

AI coding tools solved generation and teams can now ship far more code than ever. But they didn't solve verification; every PR still needs to be tested and trusted before it ships. Counterpart, an agentic insurance platform, runs a 10x QA team on Checksum at less than half the cost of one offshore developer, and hasn't had a production outage since. Their engineering manager Ron Alexssen put it this way: "For less than half the salary cost of an offshore developer, I have the impact of a full QA team."

That’s why we built Checksum. Our agentic loop runs in two parts:

🔁 Generate and maintain. On every pull request, an agent spins up in a sandbox, detects what changed, and generates or updates your End-to-end and API tests automatically. No written selectors by hand.

🔁 Run, report, fix. Trigger your suite from a PR, the API, or MCP. When something fails, a second agent triages it: real bug, or broken test by a product change? Real bugs route to Jira, Linear, or Slack. Broken tests get fixed autonomously. Söderberg & Partners went from zero to full coverage in weeks and now reclaims 90 hours of manual testing a month. Postilize cut bugs by 70% and sped up engineering cycles by 30%, with zero flaky tests.

Everything ships as standard Playwright code committed to your own repo. No proprietary format, no lock-in. And the agent doesn't just chase easy passing tests, it goes after the hard cases too: auth boundaries, edge flows, the stuff that's tedious to test manually and easy for AI to skip if you let it.

🎁 Product Hunt community gets a free 30-day trial with code PHLAUNCH

We're here all day, ask us anything. 🙏

AI-generated code is making development faster, but it also makes verification more important than ever. Tools that help developers catch problems before they reach production feel like a natural next step.

The interesting part of AI coding isn't just generating more code—it's being confident that the code actually works. Anything that closes that verification gap has huge potential for modern engineering teams.

The stale test detection is the part that caught my attention. AI generated code is only useful if the tests dont become another maintenance job.

I work at checksum so grain of salt, but my take:

What surprised me: I came in assuming the hard problem was generating tests. It isn't — models will write plausible-looking tests all day. The hard problem is that "plausible" and "green" are both terrible proxies for "actually proves the feature works." A big part of my job turned out to be making the system distrust its own passing tests.

The moment that sold me: a test broke after a routine frontend change, and the obvious move — the one every human on every team makes ten times a week — was to update the assertion to match the new behavior and get back to green. The agent refused. It walked the diff, decided the new behavior wasn't an intentional change but a regression, and filed it as a product bug instead of "fixing" the test. It was right. A human reviewer would have rubber-stamped the assertion update, the suite would have gone green, and the bug would have shipped with a passing test standing guard over it.

Why verification is the unglamorous problem: when it works, nothing visible happens. There's no demo moment for "this green check is real." So everyone builds the flashy generation demo and quietly ships suites that decay into checkmarks nobody trusts. But an untrusted suite is worse than no suite — you keep paying the maintenance cost and get none of the confidence. Solving that is the actual product, and it's the part nobody wants to put on stage.

Playwright code staying in the repo is a nice choice. i do much rather have tests i can inspect and edit than another black-box- QA layer.

Would you suggest using this for more elaborate claude code projects (full stack, hosted)? I'm the owner of a few of our internal tools and I'm trying to figure out where I can/should fit something like Checksum into the stack. I'm frequently worried about random bugs and keeping the tools running as smoothly as possible.

As a marketer, I'm just here to say that I love your tagline ☺️ it makes a pretty intimidating category feel approachable. Congrats on the launch!

The take test detection caught my attention. False failures can waste so much time that fixing those automatically could be really useful.

How do teams build confidence in auto healed tests without manually reviewing every change?

I like the idea of keeping everything as normal Playwright tests. makes adopation much easier for an existing engineering team.

Auto-healing Playwright tests for coding agents is a brilliant fix for test suite maintenance. Huge congrats on the launch!

Shipping the generated tests as standard Playwright in the team’s own repo is a really strong call. Makes it feel like part of the engineering workflow, not some vendor-owned test layer. Also love the real bug vs stale test distinction, that’s exactly where a lot of CI pain comes from. Congrats on the launch!

@benln can it test changes across multiple services in the same PR?

The New Calendly includes:

Before the Meeting
Callie, an AI assistant, handles scheduling coordination right from your inbox. Just add callie@calendly.com to an email thread, and Callie finds a time that works for everyone.

During the Meeting
Calendly Notetaker automatically records and transcribes meeting notes, so you can stay fully present and engaged. It works seamlessly with Zoom, Google Meet, and Microsoft Teams.

After the Meeting
Notetaker generates automatic recaps with summaries, key points, and action items — plus pre-drafted follow-up emails ready to send. You can also ask Callie questions about past meetings to surface context instantly. Here's a canned sample.

Why This Matters
Your entire meeting lifecycle is now connected in one platform. It's Calendly. It's easy. And it works where you already work!

Connecting the entire meeting lifecycle is a meaningful step beyond scheduling. I upvoted the product, and having scheduling, notes, recaps, and follow-ups in one workflow could remove a lot of unnecessary handoffs for teams.

Eleventh launch, and the first one landed back in 2014. Congrats to the whole team on that. Twelve years in and you are still adding to it rather than coasting.

Love the holistic focus on handling the entire workflow before, during, and after meetings. Congrats guys!

Hey Product Hunt! I'm @sidhdharth, co-founder of MeetStream AI.

Here's the bet we've staked the company on: meetings are about to stop being human-only rooms.

The people who own the platforms already believe it. Zoom's CEO says he wants to send a digital twin to his meetings. Microsoft is reorganizing Teams around human-agent teams. Fireflies hit a $1B valuation and gave its notetaker a voice. Gartner says 40% of enterprise apps will ship task-specific agents by the end of this year, up from under 5% last year.

But almost all of it is still capture: record the meeting, summarize it afterward. The agent reads the minutes. It never sits at the table.

So we built two things, and you need both to change that.

An unified capture engine. 50+ data points per meeting in real time: per-participant audio and video, live transcripts with speaker attribution, participant events, the full meeting lifecycle over webhooks. Zoom, Google Meet, and Teams through one API.

A voice infra layer. Your agent joins as a real participant with scoped permissions, speaks while the conversation is still happening, and calls tools mid-call. Ours update CRMs while the customer is still talking.

Here's what I keep coming back to: every agent company is chasing the same scarce resource, and it's context.

The context that matters most isn't sitting in a CRM field or a doc. It's in the conversation. Decisions get made in meetings. Objections surface in meetings. Most of it is never written down anywhere.

That's what we let them capture. Today MeetStream runs underneath CRM platforms, customer support tools, productivity apps, and yes, notetakers (we're not one, we power them!). 30+ AI products in production. In every one of them we ship as a feature: meeting data flowing in, and a voice going back into the room.

I didn't set out to build this either. I wanted a sales agent that could speak in meetings, and found the hard part wasn't intelligence, it was presence: lobby states, per-speaker streams, reconnection logic, platform changes that break things at 2am. Every AI team hits that wall. So we became the rails.

Voice is how humans have always worked together. Now agents are joining the conversation, literally. You build the agent. We keep it in the room.

We've been in private beta for a long time, building with a small group of teams who were patient with us while we got the hard parts right. Today we're opening it up to everyone. No waitlist, no sales call. Sign up and put a bot in a meeting in a few minutes.

What we actually want from today is the feedback. Tell us what's missing, what broke, what you'd need before you'd trust this in production. I'm reading every comment.

So, honestly: how critical do you think agents in meetings will be? I'd love to hear where you land.

@navaneeth_jawahar here, co-founder and CTO. @sidhdharth covered the why. Let me cover how it's built, because the architecture is the actual product.

Most "voice agent in a meeting" setups today are three vendors stitched together: a meeting-bot API to get into the room, a hosted voice platform somewhere else to run the agent, and a widget or iframe injected to bridge the two. Three integrations, three billing relationships, three sets of licenses, and latency that compounds at every hop.

We built MIA (MeetStream Infrastructure Agents) so the orchestration lives inside the same platform that holds the meeting seat.

What that means concretely:

One integration surface: The bot that joins the call and the agent that speaks in it are the same system. No external voice host, no injected HTML, no separate license stack to manage.

Bring your own models: STT, LLM, and TTS are all pluggable. We orchestrate the loop; you pick the providers. Deepgram, AssemblyAI, OpenAI, Gemini, ElevenLabs, Sarvam.

Wake word or proactive: Run it pipeline-mode with a wake word ("Hey MIA"), or realtime-mode where the agent decides when to speak.

In-meeting tool calling + MCP: The agent can call functions or MCP tools mid-call and report back by voice while the meeting is still going. Your tools, our tools, MCP - your pick. Or your agent’s pick, if you trust it that much.

Now the part nobody talks about, which is where most of the engineering actually went.

Every bot is a live media workload. A machine that joins a call, holds a real-time audio and video pipeline open for the length of the meeting, separates streams per speaker, and tears down cleanly. That is not a request-response API. It is closer to running a hyperscaler: we spin up over 100,000 servers a month, and the hard requirement is that none of them fall over mid-meeting, because a dropped bot is our customer's product failing in front of their customer.

Then there is platform drift. Zoom, Google Meet, and Teams each ship SDK updates, DOM changes, auth changes, and admission-flow changes on their own schedule, usually without notice. A meeting bot is permanently downstream of three roadmaps you do not control. Absorbing that so nothing changes for the teams building on us is, honestly, most of what this company does.

The rest of the unglamorous list: per-participant audio separation, speaker attribution that survives rejoins, lobby state handling across three different admission models, and reconnection that does not drop the media pipeline.

That is the layer we maintain so you don't have to.

Happy to go deep on any of it: architecture, real-time media orchestration at scale, latency, cost optimizations, and finally: why diarization is harder than it looks. Ask away.

When i read this, it just made so much sense! There are so many agents that would benefit from joining meetings, but why should all of them and their companies spend time and money on building that infra!

With meetstream, i am not going to get October agents to join dev calls!

what's the wildest use case or strangest thing you saw a uzr build with @Meetstream.ai ? (either hackathon or actual prod customer)

Congrats on the launch! Giving agents live presence rather than only post-meeting summaries is a compelling leap. The combination of per-participant real-time data with wake-word and listening controls seems especially important for usable turn-taking. Wishing the MeetStream team a great launch!

Big fan of MeetStream AI! We have been using it and the experience has been top-notch. The team is incredibly responsive to feedback and quick with fixes. For a small team, what they have built is seriously impressive and works great. Highly recommend checking this out! 

Hey @sidhdharth I see this is the best product on the market. I'd definitely love to try it!! Best of luck to you and your team.

Legends, awesome product! Can't wait to add this as an MCP

@Meetstream.ai is awesome and has some incredible capabilities! and, AI workforce is real - who wouldn’t want a clone of themselves joining the same meeting and answering questions on their behalf? 🤯

Great going guys, it’s a great usecase for meeting bots entering multi participant conversations. I can think of a variety of usecases that can be built. Cheers and congrats on the launch 🤘

Hey Product Hunt! 👋

I'm Marinos, co-founder of Prized (with Hudson).

We built Prized because AI made building easy, and the people who understand workflows best (ops, support, finance) stopped waiting on engineering. They point Claude Code or Cursor straight at company data and ship a working tool in an afternoon with no permissions or audit trail. Companies end up choosing between blocking the behavior and accepting the risk.

Prized lets you describe the tool you need (a customer lookup, an admin panel, an approval flow) and get a real full-stack app, built in a sandbox where company data comes pre-connected and scoped to what you're allowed to see. Shipping is one click, and the finished tool lives behind your company's sign-in.

The part I'm most excited about is that security is enforced in the infrastructure. The sandbox never holds a credential. Secrets live in a broker and get injected outside the generated code. Every tool gets its own database role, every data access is audited, and destructive changes wait for a human.

We're YC-backed, have a free tier with no card required, and are already powering teams at Avoca, Onyx Odds, and more.

Would genuinely love feedback, especially from anyone who's built internal tools or fielded the requests for them. And tell us the internal tool you've been waiting on engineering for 👇

Hehe loved the video guys! But I'm impressed by the value proposition even more. I'm sure you'll rock it! All the best!!!

Congrats!

Really cool, congrats Marinos. Since the builders aren't engineers, who actually holds the human-in-the-loop approval on destructive changes - is that role fixed, or configurable per tool?

looks useful, where does it deploys the code?

This is so cool! Can folks having different permissions in the team get different access to data?

Hello! 👋 So excited to share this with you all.

When my grandfather could no longer walk, my father took down two walls in our family home so a hospital bed could fit through the doorway and a wheelchair could reach the bathroom.

He changed the shape of the house because the care system stopped at the front door.

What stayed with me was how much of health falls to families: the records, medication lists, appointments, small changes, and the job of remembering which specialist said what. That work usually lives across portals, group chats, folders, and one person’s head.

So Gurleen, Param and I built Lifelong: one place for a whole family’s health.

You can keep records, medications, symptoms, appointments, activity and sleep organized by person, connect the wearables your family already uses, and use Alo in the app to log updates and take simple actions.

Most health apps are designed around an individual. Lifelong is designed around the household. It’s for families to use together, not for one person to monitor everyone else. Sharing is user-controlled, and Lifelong does not diagnose or prescribe.

Check us out on the App Store: https://apps.apple.com/ca/app/lifelong-family-health/id6758003611

Or visit our website: trylifelong.com

If you’re the person in your family who remembers which specialist said what, I’d genuinely love to hear what would make Lifelong useful for your family.

I appreciate the focus on reducing the mental load for the person who usually remembers everything. That small part of family care can become a huge responsibility over time.

love that! what privacy measures have you implemented?

Hi everyone!

@MiniMax already shipped H3. Design is the product around it.

It gives you an agent team working on the same canvas. Describe what you want, and they can turn it into a brief, reference frames, H3 shots, music, subtitles, and eventually a finished cut. Skills let you keep a good setup and use it again like a small production line.

The 3D Director is a good example of why the canvas matters. If a jump across a river or a camera move keeps drifting, you can stage the space and motion first, then generate the shot from that reference.

macOS and Windows builds are live now!

Exactly what I needed! Excited to try it out! Congrats on the launch, Kingsong and team! 🥳

Great product, I was looking for this!

I wish i had something like this back when I was building B2B SAAS 😭😭😭 it would've made the UX 10x slicker and probably been a huge boost to conversion rates.

Cool - looking forward to trying this out for branded emails!

So useful!

Super cool to see this in action! I tried the demo with my work email and loved how quickly it pulled everything together. Such a simple and clever idea. Congrats!

Super exciting! Having worked with B2B brands in the past. This would be super useful.

This is such a smart product. We tried tackling this before. Brand consistency is one of those problems that seems simple until we tried to scale across all channels. Congrats on the launch! 🚀

love this, does this only get the colours or brand elements as well?

Love this, going to have my CTO look into implementing this. This should add another level of personal customization that should keeps client retention higher.

We built a version of this at my last company. It didn't work very well and I wondered whether it was worth investing in. So glad someone is creating a SaaS for this so it can be excellent and maintained and easy to integrate.

Love this launch and the direction you’re taking Glasp.

Bringing Firefox into the ecosystem feels like more than “just another browser support” — a lot of power users, researchers, and knowledge workers live there, so it’s a strong signal you’re serious about serving that audience deeply.

A few things stand out to me as a maker:

- Cross-browser continuity: The “nothing to migrate, everything just appears” detail is huge. Most tools underestimate how much friction syncing and setup create. You’re leaning into an “it just works” experience across Chrome / Firefox / web / mobile.

- AI tightly woven into the workflow: Summarizing pages, PDFs, and YouTube directly in-tab with multiple models (ChatGPT, Claude, Gemini) is a smart move. Instead of building “yet another AI wrapper,” you’re meeting people where they’re already reading and thinking.

- Treating highlights as a long-term asset: The export options (Notion, Obsidian, Markdown, CSV, API) show you respect that users own their knowledge. That’s how you earn trust over time.

- Philosophy behind the product: The story about wanting to leave a meaningful legacy and turning that into “Greatest Legacy Accumulated as Shared Proof” gives Glasp a clear north star. It’s rare to see a highlight tool with such an explicit human motivation behind it.

Curious about two things from a product perspective:

1. How you’re thinking about surfacing past highlights at the right moment (beyond search and daily reviews) so the knowledge actually compounds.

2. How the AI Clone evolves over time as people highlight across very different domains (e.g., technical papers + philosophy + business blogs).

Big congrats on the Firefox release — excited to see how you expand into mobile Firefox and deepen the AI features next.

The MCP connector is the part I want to dig into. When Claude queries Glasp through MCP, does it get access to all highlights across every source, or can you scope the context to a specific tag or date range? I tag things heavily by project, and for agent workflows the difference between all 5,000 saved clips and just the 40 tagged for a current research thread matters a lot.

Congrats on the Firefox launch. Since Glasp bills itself as a social highlighter, I'm curious how the default privacy works in practice - if I highlight a paragraph on a public article, is that highlight private to me by default, or is there a layer where other Glasp users browsing the same page can see what I marked up unless I opt out? That distinction matters a lot for how comfortable people are highlighting sensitive or work-related reading.

I like such new things that are coming for my lovely Browser! Thanks, Guy!

Congratulations on the launch! Is it kind of like browser bookmarks, but more convenient and modern, with AI?

Congratulations! It's nice to see Firefox supported :) How do daily highlights resurface things: by relevance, randomly, in some other way?

Making knowledge easier to capture is one thing, but making it searchable and useful later is the real value. Great to see Glasp finally available for Firefox users. Congrats on the launch!

Very interesting! Is it limited to Whisper, though? I started using Raycast dictation, and it’s pretty reliable.

This is intriguing but I'm confused. I do this process with Whisperflow and tell Claude Code all the things I want to have changed on screens and use Fable to orchestrate Opus agents to make the changes. Since it can see the screen shots on the local dev server for the screen I'm talking about it knows already and I just tell it which button or components to change (not by component library name, just natural language) and it does the updates? Perhaps I'm missing something.

@wojciech_dobry Congrats on the launch. The pointing is the part I'd use. Not having to describe where something is, top left, bottom of this panel, that saves a hell of a lot of time. Definitely worth a try. Thanks for making this.

@wojciech_dobry The brief-quality framing rings true, but spoken feedback leans hard on pointing: "move this up", "that button". When the transcript says "this" and the capture shows where my cursor was, does the task your agent receives carry the actual element or file reference, or just the words?

Windows first for an agentic IDE is the thing I noticed. Almost everything in this category ships Mac and gets to Windows months later, so that is a real choice and it is the reason I can actually try it.

You asked where it falls apart. The seam I would look at is between the visual editor and the agent. When I change something on the running app, does that land in git as an ordinary diff? And can I tell later which lines came from me dragging things around and which the agent wrote?

Also, if the agent is halfway through editing the file I am clicking on, who wins?

We’ve been using bitdrift since January, and they’ve been fantastic. As a very small team migrating from Firebase, we needed an observability solution that was easy to adopt, had minimal performance overhead, and wouldn’t increase our costs.

bitdrift integrated smoothly into our existing pipelines, helping us identify and solve user problems in real time. Before using bitdrift, sampling and processing costs meant we were missing important data. Now, we have access to the information we need to quickly understand issues and focus on solving the problems that matter most.

They move fast and as the same time it feels we have a genuine personal relationship with the team. Keep rocking!

Hello Product Hunt! 👋 I’m Pete, co-founder and CEO at bitdrift.

Today, I'm really excited to announce the release of bitdrift AI: a real-time, full-fidelity observability system that lets AI agents query mobile user behavior and act on it autonomously. It's built on the bitdrift Public API and bd skills, which give your AI agents access to everything that happens on any of your customers' devices in real time. bitdrift AI is the world’s first agentic mobile observability platform.

We've all seen agents become an integral part of the observability world, and it's only going to accelerate. Agents can autonomously triage, investigate, debug, and even fix issues as they happen. This allows teams to focus on what really matters: your customers and their experiences.

But those agents, regardless of the model and harness, need access to high fidelity data to do their job. More importantly they need to be able to iterate in tight feedback loops as they investigate issues. In the past that's been impossible on mobile devices, with multi-day app release cycles, unreliable network connectivity, and scale and pricing structures severely limiting the telemetry being collected. As a result, all other observability tools rely on heavily sampled and stale data. Agents can only be as smart as the data they see, and on mobile, most have been operating blindly. Working with minimal visibility and interpolated data means they’re fixing the wrong problems.

The bitdrift approach to mobile observability is revolutionary: by coupling a real-time control plane with local on device storage, we're able to send 1000x the data when you need it and none when you don’t. We layer together extremely powerful observability and RUM features that let engineering and product teams debug and fix issues faster than ever. All without having to wait for an app release.

We’ve been working our way toward this all year, rearchitecting both our underlying client SDKs and our powerful workflow engine, releasing a new public api, and a revamped CLI.  This isn’t just some AI lipstick slapped on top of an existing product, but a ground up rethink of what it takes to give AI agents real-time access to millions of connected devices so they can find and resolve issues before users notice them. No sampling, no waiting on multi-day releases.

If you’re building a mobile app, you need to check this out. We’ve already had customers beta test this and 10x their defect fix rate, in less than a week.  We’re very excited to see what you can do with this.

Read our blog post for more details, or go directly to bitdrift.ai to learn more and install.

Ask us anything, we’ll be hanging out in the comments to answer any questions you might have. Thanks for checking us out!

--pete

Good one

Pete, the guessing after a crash report has stolen too many of my afternoons, so getting the full story of what actually happened sounds like a real relief.

Product makes a ton of sense. What's the origin story?

Hey, 
I'm Pierre from NobodyWho 👋

We've spent the last months getting local inference to be production-ready across six platforms and frameworks, not just a cool demo that works on one device.

With NobodyWho you can:
- Get answers from any open-weight AI models: Gemma, Qwen, LFM...
- Analyse images and audio through multimodal input

- Transcribe speech to text with any Whisper models
- Generate natural-sounding speech with Supertonic, Pocket TTS and Kokoro models

- Tool calling with guaranteed schema-valid output, the grammar is built from your function signature so the model can't return malformed JSON

- Run long conversations without hitting a hard message-length wall, thanks to preemptive context shifting

Wanna try our work on your device? We've built a few demo apps: iOS, Android, Apple Watch & Vision Pro.
We've also built starter examples to get started in 5 minutes and a model selection page.

NobodyWho inference engine is open-source & free, please leave a star to support us on Github 👈

Happy to answer any questions :)

Local inference on-device sounds obvious but almost never ships clean. I keep hitting setups where the demo works, then you wait twelve minutes for the model to load and the novelty dies. What does first inference look like on a mid-range Mac with something like Qwen 1.5B? Trying to figure out if this is ready for daily use or still more of a weekend experiment.

niceeee, Godot support is unusual. is the intended use NPCs that stay local or tools that game clients aren’t supposed to phone home?

Running AI models fully on device is becoming more important than ever. Love the focus on privacy, offline capability, and broad platform support instead of relying on cloud APIs. Great launch!

Six platforms including Godot is a wild spread - most on-device inference projects stop at one and call it a day.

We went the other way on a consumer app I'm building: on-device only for the narrow stuff (Apple Vision for photo classification, SFSpeechRecognizer for voice, both tiny and purpose-built) and kept anything needing real context on the server. The deciding factor wasn't output quality - it was that the AI features need months of user history in the context window, and on a phone that's either impossible or unbearably slow.

So the honest question: at what tokens/sec and what context length does local actually replace a cloud call for you? Not for a demo - for a feature someone hits ten times a day without thinking about it. That's the number I keep failing to find in these projects.

And half-joking, half-not: if this takes off, the memory story gets interesting fast. We'll end up with phones shipping 32–64GB of RAM because a photo app wants a 12B-35B model resident. :)

Hot on the heels of Buzz, Block has another very strange launch...! It's like an infinite space with trippy agentoids that do stuff for you.

Seriously cursed.

WHAT IN THE CHUMBAWUMBA.

@anupamsingh0211 I love fun stuff like this.

Hey Product Hunt community! 👋

I am excited to introduce Zoho Cliq 7.0, our most comprehensive release yet! 🚀

We built Cliq 7.0 to help teams collaborate, automate, and communicate without context switching or breaking focus.

• 🤖 Multi-Vendor AI & Multimedia Summaries: Bring Claude, ChatGPT, Gemini, Cohere, or Zia into your workspace. Summarize videos, audio, and recordings, run AI tasks, and build workflows via your preferred model.

• 📌 Chat Tabs & Advanced Search: Pin vital docs, links, and widgets inside channels with chat tabs. Locate files and past discussions fast with upgraded advanced search.

• ⚡ Cliq Mini & Collaborative Sheets: Carry conversations across CRM, Mail, and WorkDrive using the redesigned Cliq Mini. Co-author spreadsheets directly inside your chat.

• 🌐 Guest Access: Securely collaborate with external clients and partners via scoped channel access.

• 🎥 Inclusive Meetings: Live captions, pinned Sign Language Mode, and built-in availability checks to eliminate scheduling conflicts.

• 🛠️ No-Code Dev Tools & MCP: Build automations with a drag-and-drop Flow Builder, pin frequent actions via Chat Actions, and connect external AI agents via MCP and OpenAPI.

• 🛡️ Security & Accessibility: Granular AI privacy controls, admin message reporting, and full WCAG 2.0 alignment.

We would love for you to explore Cliq 7.0 and share your feedback below!

Hi Product Hunt 👋

Every riding app treats a ride like a workout: pace, distance, leaderboards. But a ride isn't a workout. It's the road, the weather, the day — and nothing was keeping that.

So I built Roveri, a riding journal for iPhone. One tap and it captures everything worth remembering: the route painted by your speed, the climbs, the weather you rode through. By December you're not looking at a spreadsheet of numbers — you're re-opening a season.

My favorite thing in it is the Atlas: every road you've ever ridden, drawn on one map as squares that deepen each time you ride them again. Watching your ground grow is stupidly addictive.

Free: recording, full history, ride details, share cards. Pro: the Atlas, a where-to-ride weather map, deeper stats, GPX export.

So, riders: what does your season deserve to be remembered by?

Almost like GitHub for travelling. :)

@p12se Congrats on the launch. Good to see a journal rather than another tracker. Everything else in this space just turns the day into statistics.

Does it keep recording properly with no signal? Spend a lot of time out of range and most apps quietly give up.

Congrats on the launch, Abhishek! Huge fan of keeping everything official through Meta instead of relying on sketchy browser hacks. Connecting Claude directly to WhatsApp Business is going to save so many people from inbox chaos. Let's goooo!

Hey everyone, Abhishek here, Suresh's partner in crime building Peach 👋

Adding to what Suresh shared, you keep using WhatsApp Business exactly like today, but now Claude, ChatGPT, or any MCP-compatible tool can see and act on that account for you. Draft replies with real context, chase quiet leads, clear a backlog, send a file — without you doing the digging first.

Setup takes a few minutes, no code, and Lite is free forever for one number. Would genuinely love for you to try it and tell us what what you'd want it to do next.

👉 trypeach.ai/co-pilot

@itzsuresh Congrats on the launch. Was pulling my own WhatsApp exports into Claude last week, so this is timely.

When Claude or Cursor reads a thread through this, what actually leaves my machine and what do you keep? That's the bit that would decide it for me.

Turning WhatsApp into a workspace for AI assistants is a really interesting idea. Helping professionals manage messages without changing their daily habits feels like a practical use of AI.

I really appreciate the focus on the official Meta Business APIs. A lot of WhatsApp automation tools rely on unofficial workarounds that businesses are understandably nervous about.

Was building on the official APIs a deliberate product decision from day one, even if it meant shipping more slowly?

Love the focus on making better use of what you already own. The closet-first approach feels much more practical than just pushing more shopping.

Love the focus on getting more value from the clothes you already own instead of always buying something new. Making fashion inspiration more personal and practical feels like a great direction.

Hey Hunters 👋, I'm Tanmay, ML Engineer at Revy. My work at Revy spanned 3 major areas:

• Training computer vision models to detect and segment clothing items, and optimizing these models to run efficiently both on-device and in the cloud.

• Building data pipelines to load, embed, and store fashion products so that users can have access to a wide range of product recommendations.

• The GPU inference service that accelerates all of AI at Revy. This service is optimized for high throughput, memory efficiency, and, most importantly, cost 😅. (We all know how pricey renting GPUs can get.)

We'd love to hear your experience using Revy!

Product Hunt! Excited to be here. I'm Kaylin, Founder and CEO of Revy.

I started Revy because I was constantly saving outfit inspiration, but when it came time to actually get dressed... I’d still wear the same things over and over again. And somehow, despite having a closet full of clothes, I was always buying more that would then go unused in a month.

That was really the starting point for Revy.

We’ve spent a lot of time building, changing things, realizing we were wrong about things, and learning from the people actually using it. So getting to share it here feels pretty surreal.

More than anything, I’d love feedback. What makes sense, what doesn’t, what you’d change, all of it.

And I’m curious: how much of your closet do you think you actually wear? I have a feeling the answers are going to be pretty bad (lol).

Heyyaa Product Fam!

I'm Divya (Div - like the HTML ones haha), growth + eng lead @ Revy.

Something that always surprised me is that most affiliate platforms for creators are built around scale - the bigger your following, the more you can earn.
At Revy, we want to empower microinfluencers and crerators. Everyone is a creator at Revy. We built our model the opposite way on purpose, so a creator with a small but real audience earns the same way someone huge does.

Curious what people here think: does follower count actually deserve to be the gatekeeper for who gets to monetize their content, or is that just the default because it was easiest to build?

The scattered screenshots and vague "fine by me" replies have quietly worn me down for years, so feedback that lands right on the spot it's about feels so much saner. Nice work, @zach_friesen.

The feedback loop between building and improving prototypes is usually the painful part. Making comments more contextual and connecting them directly with AI workflows feels like a smart way to speed things up.

As a Product Owners moving from Figma to Claude Code and constantly looking for feedbacks from stakeholders and testing users, this is life saver for me!