PH热榜 | 2026-06-07

Dreambeans is a new Google Labs experiment that turns your existing Google apps into a daily personal briefing.

You already generate a lot of signal across Gmail, Photos, Search, and YouTube. The problem is none of it connects. Dreambeans runs overnight, cross-references those sources, and brews a fresh set of illustrated stories by morning covering things you might otherwise miss.

🔹 Cross-app synthesis: Gmail, Calendar, Photos, YouTube, Search

🔹 AI-generated visuals personalised using your Google Photos

🔹 One-tap actions on stories (see ticket, watch trailer)

🔹 Feedback loop refines your next drop without altering the current one

If you live in Google's ecosystem and want a morning layer that surfaces personal context proactively, this is worth checking out.

Available now for US Google AI Ultra subscribers (18+). Waitlist open for everyone else at labs.google/dreambeans.

the proactive angle is the interesting bet here. risk with any daily digest is it surfaces noise and people learn to skim past it within a week. lives or dies on the filtering, 3 things that actually matter beats 30 that might. curious how youre tuning relevance over time.

Is it just me or does the layout feel veryyyy much like Pinterest?

A shame to gate this behind Ultra plans. How does something this paywalled - while still being a beta - hit #1 on PH? Crazy times.

Love the idea of turning cross-app signals into a morning briefing. How are you handling relevance so the digest stays useful over time, and do users get controls to tweak which sources are included?

I live in gmail and calendar all day so this is either going to be incredibly useful or incredibly unsettling. the cross-app synthesis idea is smart because all that data is already sitting there disconnected. but gating it behind Ultra while it's still an experiment is a weird move... feels like charging for a beta. also the "AI-generated stories from your photos" thing is going to be a very polarizing feature. some people will love the morning recap, others will feel like their phone is watching them journal

To be honest, I don't get it - you get a story every morning based on your images, search, and stuff? I'm assuming it's just for entertainment? Unless I've missed something, this one's a pass.

Open and free with local Whisper is the part that'd get me to actually install it. Interested in this - running fully local, which Whisper size are you defaulting to and how do you keep the release-to-text latency low enough that it still feels like it appears where you need it? Overall, well done!

The selected text as context feature is clever, hope this gets the attention it deserves!

The local Whisper option is what got my attention, most of these tools quietly push you to the cloud and hide that choice somewhere deep in the settings the floating answer while reading is the use case I keep coming back to, opening a new window just to look something up kills the focus every time, curious how it holds up on longer dictation though, does accuracy stay solid on a 2 minute ramble or is it more tuned for short commands?

hold-hotkey-speak-release is exactly how voice input should work. no app switching, no extra windows. the local whisper option is a nice touch too for anyone paranoid about sending audio to the cloud

the floating answer mode for when you're reading is the use case i keep thinking about. right now my workflow for looking something up while reading is cmd-tab to a browser or open a new Claude window which breaks the reading context completely. a floating answer that appears in place without switching apps is a different kind of interaction. curious how the floating UI actually works, does it appear near the cursor or in a fixed position, and what dismisses it

Local-OR-cloud is the right wedge - most speech-to-text tools quietly force the cloud and then bury that decision three menus deep. I build voice agents and the thing I keep relearning is that transcription is only half the job: where most tools lose users is what happens after the words land on screen. Editing friction, formatting, getting it into the next app. If Wave nails that handoff as cleanly as the privacy choice, it's a different product than the rest of the category.

I've tried many voice-to-text tools, but they usually break workflow. Wave feels like a native part of the OS. It’s particularly killer for coding and writing—being able to dictate complex logic or refine sentences without touching the keyboard is a massive productivity boost.

Hi Product Hunt, I built CabinLink because airline Wi-Fi portals often have a useful flight map, but the details are buried, slow, or disappear when the cabin connection gets flaky.

CabinLink reads supported onboard flight manifests over cabin Wi-Fi and turns them into a native Apple app experience. You get the route, current position, altitude, ground speed, heading, time remaining, aircraft details, and destination weather when an internet connection is available. If the Wi-Fi drops, the last known reading stays on screen instead of leaving you with a spinner.

It is built for iPhone, iPad, Mac, and Apple Vision Pro. There is no login, no ads, and supported cabin manifests usually do not require a paid Wi-Fi pass.

Thanks to Rudrank Riyam (@ rudrank on X.com), creator of ASCCLI, and Paul Hudson (@ twostraws on X.com), creator of Kickstart. ASCCLI helped with App Store Connect launch work, and Kickstart helped keep the launch plan grounded in actual distribution and Product Hunt best practices.

I would love feedback on airline coverage, the first-run explanation, and what flight detail would be most useful to add next.

Reading the in-flight Wi-Fi manifest directly is a clever approach, most flight tracker apps need you to input your flight manually.

I know so many flight enthusiasts who will love this. Also on shorter flights you don’t have an onboard entertainment system that shows you the flight stats. And even if you had it is convenient not to leave the movie you are watching or the game you are playing for the stats! Great work!

This is such a great launch, especially when holiday/vacay session is coming. BTW, I am on my travels too, so may use it later. :)

Sorry I am sure you get this all the time, any plans for an Android version?

If not I might whip something up for myself in the meantime

I like how specific this is. Airline Wi-Fi pages are usually slow, ugly, and annoying to keep open, so turning that hidden flight data into a clean Apple-style dashboard is a neat idea.

The part I’d care about most is knowing before the flight whether my airline is supported. Even a simple “likely works / not sure yet” list would make the app easier to trust before boarding.

The "supported manifests" caveat is where the first-run experience is going to make or break it — most users won't know if their airline is supported until they're already in the air with expectations set. Is there a way to tell someone before they board whether their flight will work, or does that require knowing the carrier in advance?

Every other tracker either makes me type a flight number or leans on a GPS trick that dies the second the screen locks. My one question is the offline behavior - when the cabin wifi cuts out and you hold the last reading, are you dead-reckoning position from the last speed and heading?

no login no ads no paid wifi pass is the entire trust proposition for a travel utility and it's the right one. travel apps that require an account or upsell you mid-flight are the worst category of software to deal with at altitude. the constraint of working with what's already available without asking for anything is a good design philosophy for this specific contex

Which airlines or wifi providers are actually supported via manifests right now?

the use case i'd want to validate before building on this is freshness. job postings have a short shelf life and a listing that was active three days ago might already be filled or pulled. what does the update frequency look like across the 1.8M listings and is there a way to query only postings updated within a specific window so i'm not surfacing stale data to end users. that latency question is probably the thing that determines whether this works for real-time job search applications or only for market analysis

This is like a Wayback Machine for job listings. Maybe a data analyst can use this to see the number of times a company posts the same job type in a given time period and gauge overall turnover rate.

The normalization layer is where I'd want to understand the depth of work done. Thirty-plus ATSs means thirty-plus ways to express job title, seniority, and employment type — and raw aggregation without opinionated normalization just moves the cleanup problem to whoever's consuming the API. How standardized are fields like seniority level or work location across sources, or is that still left to the builder?

Historical job data is super useful for trend analysis. Do you expose fields like salary ranges and remote policy consistently, or is that only available when the ATS includes it?

Unified ATS access is the part people underestimate. Anyone who built a jobs-data side project knows you start scraping LinkedIn/Indeed and rebuild it monthly when something breaks. 30+ ATS in one schema removes the worst part of the stack. Free tier is a strong play for early devs.

I would like to see this include a verification badge. Verified that the job is active, the company is real and even show how many active postings a company has, how many they post and repost

怎么更新信息的时效性？

The File Inbox is the feature I'd keep a close eye on — letting anyone send files to your link without an account is genuinely useful for freelancers collecting client assets, but it also tends to attract spam and storage abuse at scale. Curious if you've built any guardrails there (size limits, rate limiting per inbox) or whether that's a problem you're deciding to solve later.

Single-purpose tools win for me every time. I build Whatsapp/voice agents and the same trap exists in every category: pile features until the original use case feels heavy. "Upload from phone, get a beautiful link, share" is a clean wedge. Following.

the custom domain and branded link previews is the part that sold me. every time I share client files through google drive the link looks like garbage and has zero brand presence. also the file inbox where anyone can send you files without creating an account is such a simple idea but solves a real problem... I spend way too much time asking clients to "just upload it to the shared folder" and explaining how

Congrats
I usually use Localsend to share files across ipad, laptop. does your application has any advantage over Localsend?

Congrats on launching this product !

LFG!

Been following their journey for a very long time and I’m really excited to see where this goes!

Congrats on launching. The GEO / AI-citation angle is the most distinctive part of your pitch. How do you actually measure whether your content gets cited by ChatGPT, Perplexity, or AI Overviews? Is there a feedback loop where Alfa learns which structural patterns get picked up by which engines, or are you optimizing for heuristics you believe matter? That's so much harder than traditional SEO ranking measurement and curious how you're tackling it.

Strong wedge for lean SaaS teams — the slow part usually isn’t just drafting, it’s knowing whether the article is grounded in the right proof.

One thing I’d be curious to see in the workflow: do you separate the SERP/keyword brief from the company-specific source packet? For SaaS content, a post can be perfectly optimized and still feel generic if it misses the product’s actual use cases, customer language, current positioning, or proof points.

The most useful CMS-ready output might show a small “source map”: which claims came from SERP research, which came from the company/product docs, and which parts still need a founder or marketer to approve. That would make it easier to scale content without slowly drifting into interchangeable AI SEO copy.

We built WeKraft because we were living the problem ourselves.

As developers and founders, we were constantly jumping between GitHub, Jira, Slack, Notion, and a dozen notifications — just to answer "what's actually happening in this project?" The real work lived in the codebase. The updates lived somewhere else entirely.

We kept asking one question: if the codebase is the source of truth, why isn't project management built around it?

So we built WeKraft — an AI-first workspace for startups and fast-moving engineering teams. GitHub is at the core. Commits, PRs, and code activity automatically feed into your project intelligence. No manual updates. No drift between what's shipped and what's tracked.

A few things we're proud of:

• KAYA, AI Project Manager — keeps your team aligned, flags risks, and moves projects forward without micromanagement

• HARRY, AI Senior Developer — reviews code, spots issues, and gives teams a senior dev brain on every PR

• GitHub-native — your codebase is the source of truth, and WeKraft treats it that way

• Built for speed — powerful enough for complex projects, simple enough that you don't need a 3-day onboarding

Hey Wekraft, what about your activation milestone here: is it first repo connected, first task created, or something else? I mean what moment does the user realize it's worth paying for?
And did you pick it from data or gut feel?

I like the idea of a PM tool that starts from GitHub instead of forcing everyone to duplicate work.

Pretty cool idea. How does it handle syncing with multiple repositories? Seems like that could get tricky.

Building automate Task/issues directly from PRD/SRS , Auto review code -> assign bugs / issues auto to team members.
WeKraft is not a weekend build. we are students of CS and wants to change how software's are build. We truly Love product hunt for giving such platform to showcase our talent to whole world . @product thank you.
@Jira @Linear @asana

as someone who builds html emails regularly this is nice. the export-clean-html part is what matters most because half the builders out there give you bloated code that breaks in outlook. saving as JSON to reopen later is a smart touch too, most free tools lose your work the moment you close the tab

How does Narrow connect to LinkedIn without risking accounts, like is it a browser extension or something else?

@yashgupta417 - “no automations” is an interesting point to tout in the marketing; is this to reinforce your compliance with LinkedIn, or to position yourselves as different from the sea of LinkedIn automation tools? Cool product, would love a native mobile version - that’s where I find LI messaging to be the most onerous & unwieldy. Congrats on the launch!!

P.S. As a 15-year-old solo developer, setting up international payment verification takes a bit of time. If you try to upgrade to the 'Paket Sultan' and experience any issues with the payment gateway, please note that it is currently undergoing final review. Drop a comment here if you want to test the premium features, and I’ll gladly grant you access for free!

Hey everyone! I’m back! Just finished my final school exams this morning 📝 algorithm is running smoothly, and the gateway review is almost there. As promised, leave any questions, feedback, or just say hi below, and I will personally grant you FREE premium access right away! Thank you for the support so far! 🛡️✨

Congrats on the launch. The Firebase Dynamic Links shutdown left a real gap, and indie pricing for this category is overdue. Curious about the technical side since iOS made deferred deep linking notoriously hard. Fingerprinting got killed, pasteboard match destroys UX with the toast, and SKAdNetwork gives you a campaign ID but no deep route. What's your actual match signal in 2026, and what install-attribution match rate are you seeing in production? That feels like the single number that decides whether a Branch or AppsFlyer refugee trusts the switch.

How do you deal with edge case transactions and make sure the classification is explainable to the accountant?

I’m Vahin, co-founder of NAADI.

Three months ago, NAADI was just an idea.

Tris had experienced first-hand how difficult corporation tax was to learn. What started as conversations about making tax concepts easier to understand slowly turned into a much bigger question:

Why are accountants spending so much time on work that software should already be doing?

The more people we spoke to, the clearer the problem became. Before any real tax planning or advisory work happens, accountants spend hours cleaning data, reconciling records, reviewing transactions, and preparing information. Important work, but repetitive work.

So we started building.

Since March, most of my evenings, weekends, and early mornings have gone into NAADI. Thousands of lines of code, hundreds of conversations with accountants and business owners, countless bugs, rebuilds, and moments where we questioned whether we were solving the right problem.

Today, NAADI can take bookkeeping records, classify transactions, estimate a corporation tax position, identify potential risks and reliefs, and prepare the information needed to draft a CT600 in minutes.

But what I’m most proud of isn’t the technology.

It’s that we stayed focused on a simple goal:

Give accountants more time to be accountants.

Less time spent on preparation.
More time spent on advice.
More time spent helping businesses make better decisions.

Today we’re officially live.

If you’re an accountant, bookkeeper, or SME owner, we’d genuinely love your feedback.

And if you know someone running a small accountancy practice, an introduction would mean a lot.

We’re just getting started.

Check out the demo (no signup)

Hey Product Hunt 👋 I'm Kareem, the solo founder of Brisa.

I spent the last two years building this because wealth management felt broken to me: you either pay a fortune for a human advisor (who you talk to twice a year), or you get a generic robo-advisor that treats you like a risk-tolerance slider. Neither actually understands you.

So I built the thing I wanted to exist: an AI that sees your whole financial picture, explains every recommendation in plain language, and adapts as your life changes. It's early and I'm building in the open. I'd love your honest feedback, especially what would make you actually trust an AI with this. Roast it, check it out, question it, ask me anything. 🙏

🎁 Become a founding member: you'll get a 50% discount for 3 months!

I'm Sohazur, one of the Co-Founder of Caddie.

Every sales and discovery call has that moment, a pricing question you half-remember, a competitor you can't quite counter, a feature you forgot shipped. You stall, you say "let me get back to you," and the momentum dies.

Caddie fixes that.

It's the notetaker that talks back.

Most call tools just listen and summarize after the call. Caddie is live:

  👉 it joins your calls automatically,

  👉 listens in real time,

  👉 and feeds you the exact answer in your ear — the moment you need it.

 What Caddie does today:

 Auto-joins your calls

  No links to paste, no bot to invite. Caddie pulls from your calendar and shows

  up to the meetings that matter, Zoom, Meet, Teams.

 Answers live, from your knowledge

  Product questions, pricing, objections, competitor comparisons — answered in

  real time, grounded in your docs, emails, and past calls. Not generic AI

  guesses. Your actual source of truth.

 Remembers what you tell it (NEW 🚀)

  This is the big unlock. Correct Caddie once — "actually, our Enterprise plan

  includes that" — and it sticks. Every call makes it sharper. It learns your

  product the way a great teammate would.

 Chat with any past call

  Ask questions about any previous conversation, "what did they say about

  budget?", and get an instant answer from the transcript. Your call history

  becomes searchable memory.

  My goal is simple:

  Make sure no founder or rep ever loses a deal to "let me get back to you"

  again, no scrambling, no after-call regret, no fumbling the one question that

  mattered.

  If you've ever hung up a call thinking:

  "Damn, I should've had that answer ready."

  That's exactly what Caddie is built to solve.

  👉 Try it here: https://app.heycaddie.live

  I'd genuinely love your feedback. Caddie is evolving fast, and early

  validation from the PH community shapes what we build next.

  If it resonates, an upvote goes a long way ❤️ 

  Thanks for checking it out!