Connect with the Product Hunt community

Bond secured #1 Product of the Week on Product Hunt earlier this summer and we ve decided to leak our exact playbook.
Join us on August 13th for a workshop where the Bond team will reveal exactly how we earned more than 800 upvotes in 24 hours without ever dropping below the top spot.
You ll leave with:
- A step-by-step launch timeline, even if you re on a time crunch
- The exact assets to prepare
- A launch-day checklist and distribution plan
- Access to a private community of founders preparing for and sharing advice on Product Hunt launches
We ll also give direct feedback on your upcoming launches and end with an open Q&A.
Sign up HERE
PS. Want research-backed productivity lessons from the world s highest performers, behind-the-scenes Bond updates, and first access to playbooks like this one? Join our newsletter and learn how to win back 10 hours every week.

Hey Product Hunt community,

With hundreds of products launching every single day, I ve been trying to understand the underlying architecture of homepage visibility and traffic distribution.

As a maker, when you launch a product, you often notice it doesn't always appear directly on the main featured feed for every visitor or on every device/browser.

This got me curious about how the platform balances fairness vs. algorithmic curation:

Take a look at my own numbers from the last three months: 167,820 words dictated! But what is the impact? An average person types around 45 words a minute and speaks around 150. So an hour of typing produces about 2,700 words, and an hour of talking produces about 9,000. Same hour, three times the output. I spend the difference on longer prompts and fuller answers to my team, and I didn't add another hour to my day to get there :)

My own numbers are the small story. Worldwide, people have run 683,838 phrases through Lispr and skipped about seven months of typing. Seven months! Counted in working hours that is 1,232 of them, one person sitting at a keyboard eight hours a day from January through to the end of July, and nobody had to be that person. Those hours went straight back into people's lives.

Existing HTML can look complete in a browser and still be difficult to edit. The challenge becomes clearer when one file combines standard HTML with SVG, Canvas, IFrame content, positioned elements, and nested tables.

This week I rewrote our feature descriptions. Again. Third version.

What I didn't do was write to the eight people who said "ping me when it's ready" three weeks ago.

And I guess that editing copy gives me a finished thing at the end of the day. Sending those eight emails gives me eight chances to be ignored, and no way to tell if I did it well or badly.

Building always feels like progress. Selling only feels like progress in hindsight. So I hope it's not laziness.
What was yours this week? 

Hey everyone Brjan here, solo founder of Screenify Studio (live today!).

The real story, not the polished version: I kept shipping features nobody noticed because my demo videos made them look flat and boring. I'm a builder, not a video editor, and every demo ate half my afternoon in a heavy editor.

OpenAI's own app guidelines say commerce is physical goods only, and name subscriptions, credits and freemium upsells as not allowed, directly or indirectly. The same surface already ships an ads pixel and a Conversions API. That ordering tells you what it's for.

There's a distribution channel a lot of makers are quietly building for right now, and the rulebook for it is a page almost nobody reads. I read it this week because I was working out whether Murror belongs there, and one paragraph changed how I'd plan around it.

octoscope 0.28.0 is out, and the integrity scan can now see an attack that lives in two files instead of one.

Here's the shape it used to miss. You read a workflow, it triggers on pull_request_target so anyone can start it, and it holds nothing worth stealing. Fine. Except the last line is uses: ./.github/workflows/build.yml with secrets: inherit and that file reads your deploy token. Neither file is a finding on its own. Together they're a stranger's pull request opening a path to your secrets.

It was a week of refactoring: one response shape across every surface, the structure behind it pulled into one place, and the bugs that turned up while I was in there.

The shape. /ocr/fields, /ocr/markdown and /ocr/text now return the same four keys. values is your data, exactly the schema you asked for and nothing else. cells is a flat map keyed by path, one entry per value, carrying box and quad (both always present, normalized 0 1000), verified, review, and the raw evidence behind that verdict. review is the summary: declared, returned, boxed, verified, flagged[], by_reason. image is the source size, for turning the normalized boxes back into pixels. Full shape with a worked example: https://space-ocr.com/docs#post-...

In the last 48 hours we've seen big growth coming from ProductHunt, X and YouTube! I would like to say thank you to all of you for the support. It has been really fun so far.
Visitors 2.5k+
Yesterday s listings:
$100 286 clicks #1 in Global ($0.35/click)
$50 59 clicks live 24h ($0.84/click so far)
$24 104 clicks #2 in USA ($0.23/click)
$11 20 clicks live 12h ($0.55/click so far)
We still have some free countries/spots left, just saying
https://worldmap.lol

I ve been thinking a lot about the difference between personalization and real adaptation in fitness software.

A profile with age, goal and experience level is useful, but it still treats each workout request almost like a new session.

A more interesting model is longitudinal: remember what the person actually completed, which movements and muscle groups were trained recently, how difficult the exercises felt, what restrictions exist, and how nutrition has looked over the previous days.

Then the next recommendation is based on that history instead of starting from zero.

Netfox 0.16.5 is out and the bug I fixed wasn't the bug I filed.

Yesterday I logged this against my own app: sometimes the main window can't be reopened. Window Netfox clicks through and does nothing.

We just launched @Dograh , the open-source alternative to Retell and Vapi for building voice agents.

Self-hostable, so you own your stack and your data.

We started it out of rage - every tool was too expensive, too closed, or scammy (addon after addons).
So we open-sourced every line from day 0, on one belief: no company should own voice AI.
Big labs are lobbying with govts to ban open source.

More open source is how we fight back.

A few years ago, I read the book The Four by Scott Galloway.

The book, published in 2017, focused on the four companies that dominated the market at the time: Amazon, Apple, Facebook (now Meta), and Google.

This is a packed one!

Safari users, your wait is over. The Recall extension is now live on Safari and Edge, with a smoother experience for everyone, including adding connections right from the browser. If you have not tried it yet, install it on your browser of choice today: https://www.recall.it/download

We haven't posted much lately. Most of that time went into rebuilding Meku from the inside out.

This is not a UI refresh. We added new features, integrations, improved the core, and the new interface is just the part you see first.

Launch week retention is easy to get. I want to hear about week 6, when the novelty is gone and your tool is competing with the spreadsheet it replaced.

If you've shipped something that took over from a Google Sheet or a Notion table, what kept people in it past the first month? And what did the ones who left go back to?

1/ (hook)

Most people don't realize this: you no longer need to know JavaScript to build a Chrome extension.

With PlugThis, you describe what you want your browser to do and it builds a real, installable extension for you.

Here's what that actually unlocks

The one where OpenAI's smartest models finally answer. Turns out we'd been calling one JSON field by a name they retired back when o1 shipped, and they were not shy about it.

Fixed

• Reasoning models on a custom OpenAI endpoint work again. Point the OpenAI-compatible provider at OpenAI itself, pick anything from the o-series or the GPT-5 family, and every single request died before the first token: Unsupported parameter: 'max_tokens' is not supported with this model. Use 'max_completion_tokens' instead. Which is a wonderfully clear error message to receive zero times, because you're in a chat window and all you see is that nothing happened.

  The lazy fix would be to rename the field everywhere and move on. Except only about half of the OpenAI-compatible world speaks the new name: Mistral answers 422, and Ollama and llama-cpp-python ignore it in complete silence which doesn't fail your request, it just removes the ceiling on the answer. That's strictly worse than an error, because nobody ever finds out.

  So Skim now opens with the name every server understands, and switches only for an endpoint that asks for the other one by name, in its own error and only when that error names both keys, which is what separates "use the other name" from "that number is too big". Cost: one rejected request the first time you talk to such an endpoint. No new settings, no model list to maintain, nothing for you to choose. Exactly the amount of thinking about this you should have to do: none.

  Huge thanks to @ondrahracek for this one not just the fix, but a writeup that walked through what LangChain and goose each broke trying the obvious thing first, and two tests verified by mutation. Reviewing that was a genuine pleasure.

Housekeeping

I m pretty new to launching products, and I m a bit confused about the right order of things.

If I have an idea and a basic MVP, should I:

• Buy the domain and set up the website, branding, etc. first?

We wanted to come back and say thank you, properly.

Since launching here, Kopai crossed 150 users, and a lot of that traces directly back to this community. The upvotes mattered, but the comments mattered more: several of you asked sharp, specific questions that are already shaping what we build next.

A production incident can consume six hours of investigation and end with a ten-minute fix. An engineer can spend an afternoon reviewing an AI-generated implementation that took 20 minutes to produce, experimenting with five approaches before finding the one that works, or helping a teammate make a critical architectural decision. Yet the system may record only a ticket moving from In Progress to Done.

I saw this repeatedly while building AI-native systems for enterprises at my previous organization, leading a team of talented developers.

The consequence is bigger than poor reporting: companies misjudge capacity, overlook contributions, and reward visible output over difficult work. AI coding agents are making that mismatch impossible to ignore.

I pay a lot of attention to how a product handles me before I ever hand over a credit card. How generous is the free tier? Is the onboarding smooth, or does it make you feel like a nuisance for not upgrading immediately?

Some apps treat free accounts like a favor they are reluctantly doing for you, while others just let you use the tool and trust that you will pay if it provides value. I have walked away from solid apps simply because the free experience felt pushy, and I have stuck with limited tools just because they were respectful of my time.

Does the attitude of a free tier actually impact your choice, or do you only care about what a product offers after you pay?

I was in my car today and tried to use Siri to send an email.

And somehow, I ended up spending more time fighting with Siri than actually sending the email

Try this