PH热榜 | 2026-05-21

Hey Product Hunt, I’m Xiaoyin, founder of Tycoon.us

A year ago, I became the first human CEO to be replaced by an AI CEO named Astra.

At first, it sounded like a stunt. Then Astra helped run real companies: HeyBoss reached 100K+ users, and SkillBoss hit $1M ARR in 30 days.

Tycoon is the productized version of that experiment.

It gives one person an AI CEO and a full team of AI agents. You can text Astra your goals, ideas, or tasks. She turns them into structured work, assigns the right agents, reviews progress, and asks for approval when it matters.

We built Tycoon for solo founders, indie hackers, and builders who want the operating power of a company without hiring managers, sitting in meetings, or coordinating everything manually.

You can view the step-by-step walkthrough here: https://www.youtube.com/watch?v=tYejVbsRRec

This is really cool and terrifying at the same time. But I am curious how does the feedback loop works here. What I mean is that there is a hierarchy in real world company but for AI agents how does that work? Does it have designated AI agents for each role in company ?

Is there any team or collaboration feature planned, or is it strictly personal use?

Feels like a clean way to turn strategy into something actually fun and addictive. If the gameplay depth matches the concept, this could easily become a go-to for anyone who enjoys building and scaling systems. Excited to see how far it goes 🚀

An AI running the entire thing is wild. But I want to test it out - I have $53 worth of credit but nothing happens. Did I miss something that says a subscription must be purchased to make this do anything other than chat with it?

bugs are everywhere. cant view skills, the main bug is in the chat and company creation function. it doesnt work. it is broken. no matter how many agents or claude codes and hermeses you put inside it, it doesnt work right. astra made tasks, none of them even started. charging 50$ for a trial, up to 1500$ for a broken product? the only good thing i see here is the buy a domain on namecheap.com option. maybe they should start a product hunt launch instead. terrible first, and last experience

The “AI CEO” framing is bold, but the useful bit here seems to be the approval boundary around work that touches real customers.

For the CMO agent specifically, I’d love to see every public post/campaign carry a small source trail: what customer language, positioning notes, or prior winning content it used, plus which parts were invented. That would make founder review much faster because you’re judging both the output and whether the agent understood the business correctly.

Solo iOS founder here building a relationship app on the side, exactly the indie-hacker persona you’re describing. The HeyBoss + SkillBoss track record before Tycoon is the most impressive credibility on the board today, congrats on the launch!

What’s Astra’s actual signal for “this needs approval”: per-agent confidence, category of task (publish vs draft, spend vs research), or learned from user behavior over time? Feels like getting this right is crucial to not immediately reverting to doing everything manually.

Congratulations

Congrats Xiaoyin and team! really fascinating vision here. how does Astra decide when to escalate decisions to the founder versus letting agents execute autonomously?

Really interesting idea. Curious how Astra handles quality expectations, since every founder may define “good enough” differently. Does it learn each user’s satisfaction bar from approvals, edits, and rejections over time, or can users explicitly set the quality standard for different types of work?

Curious how well it actually retains context over time — or if I’d still end up re-explaining things every few steps. That’s usually where most “agentic” tools start to break down.

The branding around “Tycoon” is interesting—it subtly reinforces a mindset of building and growing wealth, which makes the product feel more motivational than purely functional.

Congrats on the launch
I am curious to is it possible to integrate the agent with meta mcp for directly running ADs and design creatives as well. Like is it capable of execution or only of strategy

Hello,

I am having a lot of trouble with you product/services. Worker pods are consistently failing with HTTP 402 at pod spawn despite active subscription, $99 balance, and $100 monthly spend limit. Happening across all tasks. Started after subscribing to the Tester plan. Multiple retries are wasting credits.

Also, the "tycoon.us/support" page is not working.

Please advice,

Thanks!

When Astra creates a plan and assigns agents, how does it handle conflicts between agents — like if the CMO and CTO have competing priorities on the same sprint?

API calls? Expensive? or different format?

This feels like selling the shovel in a gold rush. Fantastic idea if executed well. $49/mo is not much, but we dam well know it's gonna get real expensive really fast. How do you guys optimize for token usage efficiency?

Love the idea - however my group of first tasks say they are running but the threads show 0 sessions, 0 tool calls, 0 tokens used. Restarting them didn't help!

I'm curious how well does this work for a product idea that already is built. eg. A web app, marketing site are setup but all of the other functions besides AI Developer are needed? Will this trip up if it's not fully in your ecosystem?

It sounds exciting, but it also feels a bit frustrating. It might be better if we could see the actual results instead of just numerical promotion. At least, it's not yet clear what problems are being solved? Is it about fewer people? For example, what effective growth can it bring to independent developers?

Congratulations on the launch!
I created an account and I think it’s a very good product.

I did a brief test of the experience and I really loved Astra in general .

However, it would be interesting to be able to see the tasks in 'To Do' on the same screen and interact with each one, either to add more details or to ask follow-up questions.

Astra as a coordinator seems very good to me. It would be useful to be able to incorporate some 'skills' into the brainstorming process; for example, when I want to explore options, Astra sometimes seems a bit too expeditious.

A thought crossed my mind, if two founders used Tycoon for five years and then both disappeared, would the resulting companies become more similar over time because they’re run by the same system or more different because each AI CEO has absorbed a unique history of decisions?

It feels like the answer says something important about whether intelligence is coming from the model or from accumulated organisational memory.

The text your CEO interaction model feels very natural. Did you intentionally design Tycoon to reduce dashboard complexity for founders?

Really impressive to see AI agents handling the full business ops stack. Curious how does Astra handle decisions that need brand context or tone? Like if a CMO agent is writing content, what's feeding it the company's voice? Excited to see where this goes

The HeyBoss / SkillBoss numbers are the part of the pitch I keep getting stuck on. Were those companies originally human-run and then handed to Astra? It's a different credibility claim each way, and the cold-start version is the harder one to defend

Love the idea but how come you guys aren’t a one-person company?

Hi Product Hunt community!

We are excited to introduce Mintlify Workflows

What is it?

Automations that keep your knowledge base up-to-date without manual upkeep. Pick a workflow, choose when it runs, and let it maintain itself.

Why now?

With recent AI advancements, the gap between what your product does and what your knowledge base grows wider, and closing it becomes a project of its own. We built Workflows to save you time and automate the busywork of maintaining knowledge bases so you can stay focused on shipping.

What's included?

Each workflow is focused, tested, and ready to go. No prompts are required to write or maintain. We built them around the tasks teams want to automate the most:

- Codebase updates: sync content when PRs merge to your code repo
- Changelog: draft entries from recent product updates on a recurring schedule
- Translations: automatically translate your content to increase reach
- Broken links: find and fix links automatically
- SEO: audit titles, meta tags, headings, and canonical tags
- Grammar: catch typos and grammar errors
- Brand tone: enforce your style guide's voice and rules

never gonna have to bug someone to update the docs ever again

can't live without this product, workflows keep our docs totally updated. you're the best!!!

Congrats!! Docs are always the first thing to rot when the product ships fast.

this is a game-changer -- very excited to set it up with Slack and Notion!

curious what you've seen are the most popular integrations??

@hahnbeelee Congrats, for the launch.
The PR merge trigger for codebase sync is the right design decision. Documentation drift usually starts the moment code ships and nobody updates the corresponding page.
But what I would like to know is that.. is the changelog workflow specifically: is it generating entries from commit messages, PR descriptions, or something deeper like actual diff analysis? The quality of auto-generated changelogs depends entirely on what signal it's reading from.

Feels like an intersting long-term shift here. Curious if the hardest problem here eventually becomes technical or behavioural. Most knowledge bases don’t fail because information is missing.

They fail because teams stop trusting them. How do you think about preserving organisational trust once docs become AI-maintained instead of human-maintained?

This is awesome!

Super interesting addition to the Mintlify ecosystem. It’s tough to keep up with documentation when you’re shipping fast, so pre-built automations make a ton of sense. How tightly does this integrate with version control systems like GitHub or GitLab triggers? Can we customize the automated changelog generation based on specific commit tags?

This is a good example of automation being useful because it removes a recurring ownership problem, not just a one-time task. Docs usually break because no one owns the boring update loop, so tying workflows to product changes makes a lot of sense.

Self-updating knowledge bases is the right framing — docs rot has always been an attention problem, not a tooling one. Curious how you handle conflicts when the codebase update workflow and the brand-tone workflow want different things in the same paragraph — does the user arbitrate, or does one always win?

Really like the direction of self-maintaining docs—feels like a natural evolution as teams ship faster.

One thing I’m curious about: over time, as multiple workflows keep updating the same docs, how do you prevent “automation drift” where tone, structure, or even intent slowly diverges from the original?

Is there a way to enforce a kind of long-term consistency beyond just individual workflow rules?

Docs going stale the moment your product ships is such a silent killer for dev teams. The PR merge trigger is genuinely the right hook cause that's exactly when things break. One thing I'm curious about though, can it detect when a feature gets fully removed and proactively flag the outdated docs before users hit a dead end?

When the codebase changes significantly, how does Mintlify handle keeping existing docs in sync — does it flag drift automatically or is that still a manual process?

Self-updating docs is one of those things you don't realize you need desperately until you've spent an afternoon hunting for the one endpoint that changed three PRs ago. As a solo founder I always deprioritize documentation and then pay for it later. How does it detect what changed — does it watch the codebase directly or does it work from commit diffs?

The codebase update workflow is actually intriguing. Keeping docs aligned with merged PRs is something that every developer team struggles eventually. Just wondering if the system is capable of detecting when code change actually impacts user-facing behavior versus internal refactors.

Love the idea of self-maintaining knowledge bases. One thing I’m curious about is how do you balance the accuracy with the speed of automation, especially for technical products where a small hallucination can create major implementation issues?

It would probably also be great to use something like this for user-facing release updates — automatically generating feature announcements, changelogs, and product update posts whenever something changes.☺️

Google just separated the agent manager from the IDE and shipped it as its own desktop app.

What it is: Antigravity 2.0 is a standalone desktop application built entirely around orchestrating multiple AI agents in parallel, scheduling background tasks, and managing subagent workflows across projects.

Most AI coding tools still make you sit in the loop: prompt, wait, respond, repeat. Antigravity 2.0 breaks that pattern by letting agents run in the background on cron-like schedules, work in parallel across subagents, and carry full project context from AI Studio to your local environment in one click.

• Run multiple agents simultaneously across parallelized subagent workflows

• Schedule tasks that trigger agents automatically in the background

• Export full projects from Google AI Studio to local development with one click

• Connect natively with Firebase and Android

• Issue voice commands instead of typing prompts

• Use the CLI for terminal-native work or the SDK to deploy custom agents on your own infrastructure

If you're a software developer or engineering team that has outgrown one-shot prompting and wants agents running across your build loop without babysitting them, this is built for that workflow.

P.S. I hunt the latest and greatest launches in tech, SaaS and AI, follow to be notified → @rohanrecommends

Never used the IDE version much either. The subagent workflow is where it gets interesting: running parallel agents without babysitting each one is the shift that makes this feel different from Cursor or Claude Code. The background scheduling is a nice touch too. Excited to see how far they push the Firebase and Android integrations.

how to make this bunch of ai agents run in the background so as not to allow burn all tokens my tokens at all?

Scheduled agent work gets interesting the moment a team comes back later and has to decide whether a run is safe to accept or needs review. I would want every scheduled task to leave a small receipt: owner, allowed capability set, files or services touched, stop reason, and whether it completed, paused, or hit a guardrail. Without that, background autonomy can create more context debt than leverage.

Most software abstractions succeed when they hide complexity. Multi-agent systems seem to be doing the opposite, exposing planning, delegation, coordination, and supervision as first-class concepts.

Do you think the future interface is actually a visible org chart of agents, or does that disappear entirely once the system becomes reliable enough?

google shipping a standalone agent orchestrator separate from the IDE says a lot about where this space is heading. agents aren't a feature inside dev tools anymore they're becoming their own category

I have used Antigravity IDE Version before to make projects during hackathons, academic projects. What I like about using it, is it's way of interpreting the user's written prompts or instructions in a structured way as it's finishing a to do list one at a time. With the upgrade of 2.0, I am sure the ability of achieving multi tasking through multi agent feature would be a significant update.

Parallel agents are powerful, but the monitoring layer is what makes this actually usable. Once multiple agents work at the same time, the hard part becomes catching conflicts, knowing what changed, and deciding what needs human review.

Google announced so many Antigravity updates lately SDK, CLI, etc.

Scheduled background tasks is the part I'd actually use. Quick question though, what happens when an agent task runs longer than the context window? Does the subagent remember where it left off between runs or start fresh each time? That's the wall I keep hitting building with agents lately.

I’ve used Antigravity IDE for hackathons and projects—I like how it turns prompts into structured, step-by-step tasks. Excited to see how the multi-agent feature in 2.0 improves multitasking.

Hey Product Hunters 👋

Raphael, WeWeb CEO & co-founder here.

WeWeb is THE AI app builder for non-technical users: it lets you generate full applications with AI, then customize every part visually with a drag-and-drop editor.

Over the past year, we've all watched AI transform how fast you can build web apps.

But what happens after the prompt?

For most non-technical builders, AI generation hits the same wall: you create something impressive in minutes… then get stuck at 80%.

You can't really edit the output. You don't understand what's under the hood. One wrong click and everything breaks.

We are fixing that.

WeWeb is the AI app builder for people who want to understand what they're creating: where AI speed meets no-code control.

Generate full apps with UIs, workflows, and data structures in minutes.

Then use the drag & drop editor to inspect exactly what the AI produced, modify any part visually, and publish your app in one-click.

What can you build with WeWeb?

• SaaS products that scale to millions of users

• AI-native applications

• Internal tools and operations platforms

• Client portals, ERPs, and CRMs

Backed by YC, WeWeb is used by solo founders, startups, scaleups, and companies like PwC, L'Oréal or Décathlon to ship production-grade apps at scale.

I'd love to hear your feedback - ask me anything, I'll be here all day 🙂

🎁 Exclusive for the Product Hunt community: Use code PH2026 for 20% lifetime discount on any plan. You have one week to take advantage of this exclusive PH discount, don't waste time!

Huge congrats on the launch 🎉

We built gamercoin.co / gamerperks.co on WeWeb to test an idea fast, with little coding experience between us. The pitch in this launch is exactly what made it work for us. We could get something real off the ground, then actually go in and edit the UI, logic, and data without staring at code we didn't understand. WeWeb is the reason we got past it.

Rooting for the team today 🚀

We have been working hard on this and we will continue to work hard to provide the best AI experience possible for our users :)

I see in your website that the app can be in Github. Does that mean I have the full app in Github and I can host it somewhere else ?
If that the case, is it your own framework or an existing one ?

A lot of AI tools can generate apps now, but the frustrating part comes after: editing things, understanding what was built, making changes without breaking everything 😅

Definitely going to try it!

Congrats on your launch WeWeb's team!

Congratulations to the whole WeWeb team for the launch ! Go WeWeb🔥🔥🔥

Hey! AI guy here, here are some technical details about the AI capabilities in WeWeb to give you a better picture of how it works in practice :)

The AI is fullstack, meaning it can work across your entire project. Ask questions, get advice, or have it design, build, edit, or fix anything in your app!

Backend: It helps you design and create your database, set up security access rules, and handle all the backend logic (APIs, internal flows).

Frontend: It can build responsive pages from scratch while keeping a consistent design throughout your app, and it wires up the frontend logic (no-code workflows) linked to your backend (API calls, fetching data from the DB).

Of course, everything the AI generates can be visualized and edited in no-code!

If you have any questions on the topic, don't hesitate to ask! :)

Hey team,

Amazing work on the v3! The backend part is truly impressive.

Is there any plans in the future for an MCP? Importing our skills or own models?

Congrats on the launch!

PS: so cool to see the product evolution since I left the company. You guys rock 🙌

Hi everyone! I'm Florian Briou Rolle, Co-founder & CPO at WeWeb.

I am incredibly proud of our team for the hard work they put into bringing this new version to life, and I absolutely can't wait to see what you all build with it!

Please drop your thoughts, questions, and feedback in the comments below. I’ll be hanging out here answering them today, and we are ready to iterate fast based on what you tell us. Let us know what you think! 🚀

Hi all 👋,
I'm Matthew, Education Lead at WeWeb.

It's amazing to finally get the new full-stack AI experience live!
We'll be prepping a new academy over the coming months. It'll cover how to get the most out of WeWeb and build following web development best practices.

If there is anything specific you'd like to see, just let me know :)

Launch day! 🚀 DevOps here, servers are warmed up, autoscaling is armed, and I've got coffee within arm's reach. Bring on the traffic, I've been waiting for this all sprint.

love the idea of being able to actually go in and tweak yourself! When the AI generates the app, can you also go back and re-prompt it later to make changes, or is it more of a 1-shot generation and then everything else happens in the visual editor?

Congrats on the launch!!
I've been freelancing on WeWeb for a few years now. The tool already provided with an amazing visual editor that made building webapps so pleasant. Now coupled with AI capabilities, it brings even more power to help developers build reliable applications! Speed up build time without sacrificing control over the what you create. I'm truly excited to craft in WeWeb 🔥
Keep up the amazing work WeWeb team ❤️

I've been burned by too many AI builders that create something cool in 30 seconds and then completely fall apart the moment you try to change anything. like yes the demo looks great but then you're stuck because you have no idea what's actually happening under the hood and one wrong move breaks everything.

As a previous fullstack developer I was never against other AI builders, I just kept getting let down by it. the promise was always there but the reality was always the same: clunky output, limited customization, and this constant feeling that you were one edge case away from hitting a wall you couldn't get past

What got me with WeWeb is that it actually lets you go in and edit things yourself after the AI does its thing. not in a 'here's the raw code good luck' way but visually, so you can actually understand what was built and tweak it without everything falling apart.

I've been using WeWeb for years now and been watching this one come together and it's great to see it ship. Congrats to the team, this one's worth the wait 🎉

WeWeb is made for those who believe understanding what you're building still matters.

If you too can’t let go of the thrill of figuring things out and learning by doing, but still want to benefit from the AI speed boost, you're in the right place!

When you dev on weweb you don't really need AI to speed it up. Weweb is quick enough :D
+1 Good luck team!

Congrats on the launch @raphael_goldsztejn and team! WeWeb is the obvious choice for non-technical builders who actually want to confidently ship something with the help of AI. Y'all continue to impress and I can't wait to see what people build with it.

First time using an app builder AI. I went in expecting it to all fall apart after second prompt but WeWeb surprised. The app was very well put together. Loved the product and the AI.

Been burned by black-box AI builders. This "full control" angle hits home. If the AI can scaffold screens + data cleanly and I can tweak workflows without weird hacks, that’s legit. Gonna try it on a small client tool this week.

Developer at WeWeb here — really glad to see the interest! Happy to answer any questions if you have any!

@raphael_goldsztejn Congrats on the launch. One question: is there an integrated design system to keep design consistent over time?

Cheers

Looks amazing, I'm working on a much simpler landing page HTML builder project and can't imagine how much extra work is needed for all the rest, dynamic API database binding.

I've been using WeWeb for projects lately and IMO they are the best for non-technical users. AI integrated, easy drag and drop, easy to build things and customize components. I use Supabase on the backend and it integrates really intuitively. Absolutely recommend trying WeWeb. I'm genuinely impressed by what they've accomplished and how intuitive they are.

This is a really smart positioning — "the safety net of a no-code editor" for vibe-coded apps. I've been building my SaaS (YTubViral) mostly in code, but there are parts where a visual layer on top would have saved me days of iteration. Curious: does it handle dynamic data-heavy pages well? Like dashboards with charts, tables, and real-time updates? That's usually where no-code tools start struggling. Congrats on the launch!

I’ve created lots of apps with WeWeb, but one of them is https://www.linkeflow.io/es/ – I built it to test the idea of automating everything on LinkedIn, and that idea led to many others.

WeWeb is undoubtedly one of the best platforms for creating apps without writing a single line of code, and it also lets you customise and tweak everything to your liking.

¡Let's go, WeWeb!

Recently I read that venture investors have started refusing to invest in startups built with AI because they’re impossible to maintain in the future and everything will eventually need to be rewritten from scratch. How are you solving this problem? Or does this not concern you and is it the user’s problem?

Really interesting approach with AI → JSON → App as a guardrail layer 👀

One thing I’m curious about: how do you handle state complexity and edge cases as apps grow?
For example, when workflows become highly conditional (role-based access, async flows, partial failures), does the JSON abstraction still stay manageable, or does it become harder for non-coders to reason about what’s actually happening?

Feels like that’s where many tools start simple but hit a ceiling.

Hey everyone! 👋

Excited to launch Slideshot on Product Hunt today!

Huge thanks to @fmerian for hunting! As always, extremely appreciate your support 🙏

I built Slideshot from a simple realization. AI agents made it so much faster to ship features. At the same time, I felt the speed of how fast you can market these new features is lacking behind. Especially when it comes to video demo preparation. I'd need to manually walk through the flow, record it, then edit it, adding some nice zoom effects, then generate a GIF. It takes a while...

Slideshot is my attempt to make that workflow "agentic".

Instead of opening another screen recorder, you connect Slideshot to the agent you already use through MCP. It supports Claude, Codex, Cursor, ChatGPT, or any other MCP-capable agent. Then you describe the product flow you want to show.

The agent drives the browser, walks through your web app autonomously, records the demo, and Slideshot returns a polished video with built-in zoom effects, cursor motion, and an intro animation.

Here are the main use cases I'm optimizing for:

✻ Product marketers creating launch videos, changelog assets, in-app educational videos, and feature announcements

✻ Product teams keep demo assets up to date as the product changes. Think of a Help Centre that automatically shows how a video of the feature works in the latest version.

✻ Customer success and support teams creating walkthroughs for help docs and customer education

What I’m most interested in is whether demo creation can become part of the same AI-assisted workflow as product building itself.

If an agent can help ship the feature, it should also be able to help market that feature.

It's is still early, so I’d love feedback from teams that regularly need product demos but do not want to spend time recording and editing them by hand.

---

Because I'm building Slidehsot as an agent-first tool, the pricing model is also different. Usually, video recording/editing tools charge a monthly subscription per seat. For Slidehsot, I opted for usage-based pricing. You only pay for the videos you generate.

For Product Hunt, the first 25 people who use the PH21SLIDE code at checkout will also receive $5 worth of credits for free.

absolute fan of Ahmed's work.

the maker of @Katalog and @Arcmark keeps cooking, introducing @Slideshot, a simple and elegant agent-driven screen recorder.

if you use @useloom or @Screen Studio and want to automate this work, give it a spin! you'll enjoy it.

Do teams ever push back wanting truer pacing for onboarding or sales videos? Or it it more like everyone just wants the cleaned-up version? Congrats on a well-executed product

congrats on the launch @geek_1001
Is the editing of video is ai driven or need manual intervention

Product demo videos are the one thing I keep putting off. Recording, editing, re-recording when the UI changes... it never ends. Having an AI agent handle this would remove a real bottleneck. Question: can it handle multi-step flows? Like "user signs up → connects YouTube → sees their first dashboard"? That kind of walkthrough is what I need for my landing page but I've been too lazy to produce manually.

This is a strong wedge. One place I’d watch is narrative intent: if the agent records exactly the flow you ask for, the demo can still miss the “why this matters” moment.

It would be useful to attach a tiny script/storyboard to each recording: audience, promise, steps to emphasize, parts to skip, CTA. Then the agent isn’t just capturing UI motion; it’s preserving the GTM/story context around the feature as the product changes.

Using MCP to let an agent drive the app rather than relying on brittle scripted shortcuts is clever. It's a guarantee that demos stay reproducible when the product UI shifts. How does it handle apps that need auth flows or complex stateful setups before the walkthrough can begin?

Okay this is actually so cool — the fact that it drives your app by itself and just hands you back a finished demo video is wild to me. I'm always putting off recording demos because it takes forever to get right, and this just… does it for you. Really curious how it handles things like popups or slow loading screens though. Definitely upvoting this one.

this would've saved me so many hours. recording product demos manually and then adding zoom effects and cursor animations in post is the most tedious part of marketing. does it handle multi-step flows where you need to show different pages or is it single page only

Nice, great work @geek_1001 🙌

Wow! Can’t wait to try it out ✨

Thanks for launching..

Feature requests: verified emails, sample copies for outreach with desired intent and ability to run campaigns inside the platform.

This is cool, can be useful for prospecting or seeking jobs.

Congrats. Just curious as a builder, what are you using to match contacts? Is there an API for this??

When I entered my profile and potential company to reach out to, it showed me an error. Please advise. (screenshot below)

Curious whether anyone's used it to break into public sector or education. Does it work just as well when your target is a school, a nonprofit, or a government office? Congrats on the launch!

Where do you pull the employee background data from, is it mostly LinkedIn, and how do you keep it current?

While I like the idea behind it, there is an error that needs to be fixed. When I put my LinkedIn url and the company's name, it shows Drop your LinkedIn Url again. I tried it quite a few times. Please check this out.

I tried it out and results were quite good. One thing I would love to see is the ability to bulk upload a list of companies and get all the matches exported in a file. Would save a lot of time.

How about creator / personal introductions? You put in your profile URL and the person (creator, investor, talent, etc.) you want to directly reach out to. Then it matches based on shared interests and experiences. Could be extended to X and Instagram too.

Hey Product Hunt,

13 years building the same belief: your browser should adapt to you, not the other way around.

Vivaldi 8.0 is the biggest thing we've shipped in years, and it starts with how the browser feels.

We call it Unified.

Previously, Vivaldi's interface was a collection of layers (tabs, toolbars, panels, content), each subtly separated. Useful, but fragmented. With Unified, those boundaries are gone. Everything now lives on one continuous surface that wraps the entire browser. Alignment becomes more precise, spacing more intentional, and interaction more direct, because elements no longer sit in isolated layers.

If you love the look you've already got, it's still right there in Settings.

What's new in 8.0:

• Your theme flows across the whole window, edge to edge 

• Six layouts to kickstart customizing the browser to your needs

• 7,000+ community themes to explore

No tracking. No AI sitting between you and the web deciding what you see. Just you and the web. On desktop, mobile, and yes, your car.

Your browser. Your rules.  

Tell us what you think! We read every comment here.

Hi, Product Hunt! 👋 I’m Darius, the co-founder of Attention Insight.

We built a new Figma plugin specifically for UX and product designers.
We spoke with hundreds of UX designers and product teams and found a clear need for a data-driven way to validate design decisions.

Because today, design decisions still look like this:
🤔 “Make the CTA bigger.”
🤷 “This layout feels better.”
🔁 “Let’s try one more version…”

And suddenly you’re moving elements around instead of actually knowing what works.

We wanted a way to validate design decisions earlier, before user testing, before development, and without leaving Figma.

So we built Visual Usability Checker — an AI-powered Figma plugin trained on millions of eye-tracking data points to help you validate design decisions earlier with instant recommendations, cognitive load insights, and predicted attention.

With it, you can:
➡️ Get AI recommendations on how to improve your design instantly
➡️ Detect high cognitive load areas that may confuse users
➡️ See where users are likely to look in the first seconds
➡️ Spot weak visual hierarchy and usability issues early
➡️ Compare multiple design variations side by side

All without leaving your workflow.

Instead of guessing, you get clear, data-driven feedback you can act on immediately.

How it works:
Run the plugin → Log in or Sign up → Choose your testing workflow → Select a Figma frame → Get attention maps, scores, and insights in seconds.

You can iterate immediately and see how changes affect user attention.

Our goal is not to replace user research.
It’s to help designers catch problems earlier, reduce guesswork, and make decisions with more confidence.

We’re especially curious:
- What part of design validation still feels too subjective in your workflow?
- Where do you spend the most time debating instead of deciding?

- Who wants to get a discount?
Would love your feedback 🙌

Playground file : https://www.figma.com/community/file/1575839690110028010/visual-usability-checker-playground-file

How does this work technically? Do you analyze frames/images, identify which elements are shown in them, and then pull statistics from a database? For example, we all know that in advertising people look at faces - if you determine that the image contains a human face, a dog, and a tree, then it’s easy to understand that a person will look at the face first, then the dog, and then the tree. How did you implement this?

Hey Product Hunt! 👋

I’m Tiffany from Mixpanel, and we’re excited to introduce Mixpanel Headless, a Python SDK that makes the entire product surface area programmable and composable.

Builders can now access all the power of Mixpanel—from complex funnel analysis to attribution modeling—without leaving their IDE.

But what we’re most excited about?

You can now use agents to answer anything about your product. Ask a question in plain language, get working code in seconds, and execute it against Mixpanel data. Agents compress hours-long analysis into durable code you can run every week.

You can check out the docs and try out Mixpanel Headless today: https://docs.mixpanel.com/docs/m...

We’d love to know your feedback or thoughts!

Hi. Do users get to review the generated query/code before it runs against production analytics data?

So pumped to see this live today! 🚀

As someone who's worked closely with how teams actually use Mixpanel, one of the biggest pain points has always been the gap between asking a question and getting an answer. You'd have to build a dashboard, wait for a data pull, or loop in an analyst.

Mixpanel Headless flips that. Now, whether you're building an AI agent or just want to script your own weekly retention report, you have direct programmatic access to the full analytics surface: funnels, cohorts, attribution, all from Python.

The agent use case is what really excites me. Instead of one-off answers, you get durable code that keeps running. This of course isn't for everyone, but for those that can leverage it, I've already heard some amazing ways that this makes life easier for Mixpanel users.

I'm happy to answer any questions about how this works under the hood or how to get started! What workflows are YOU most excited to automate?

Hey Product Hunt! I'm Charlotte, CEO at Axite.

We are so proud of this launch and genuinely excited to share Tacet with you. Perfect for anyone who has always wanted a deeper look into their brain activity, without the price tag or the clinic visit!🧠

It turns out the way you tap your phone tells you something about your brain. Not what you type, not which apps you open, but the timing between your taps. That rhythm shifts when you are cognitively sharp, when you are fatigued, when your sleep has been poor. It is a signal that has always been there. We just built the software to read it.
Brain monitoring has always been locked away in expensive hardware and clinical settings. Tacet changes that. It runs quietly in the background, no difficult setup and no expensive headset, and turns that tap signal into something you can actually use.

Here is what you get:
📱 A cognitive fitness score that shows how sharp your brain is performing, whenever you like
💤 Sleep quality tracking based on your behavioural patterns
📈 Longitudinal trends so you can see how your brain changes over days and weeks
☕ A log to add context like stress, caffeine, or a bad night, so your data actually means something.

📏 The perfect tool to measure whether a lifestyle change is having a real effect on your brain's performance

One thing worth knowing: the scores get more accurate after about two weeks of use. The algorithm needs time to learn your personal baseline. So download it today, let it run, and see your cognitive scores appear over time. That is when things start to get really interesting!

We are building towards a full remote brain monitoring platform, but we wanted to get something real into people's hands now. A first glimpse into your own brain, accessible to anyone with an Android smartphone (we are still working on the IOS release!!)

🎉 Product Hunt exclusive: we are offering a special discount during the first week of launch: A one month free trial! Grab it while it lasts.

How to use: Download the app, make an premium account trough settings and add the code FIRSTTAP for one month free!

Download the app and tell us what you think. What would you want to learn about your own brain most? We would love to hear your thoughts!

Hey Product Hunt! I'm Ruchella CTO of Axite.

🧠📱We built this app because we kept asking ourselves: why is it so hard to actually see your own brain health patterns? Tacet changes that. It passively monitors your smartphone behavior and shows you, your cognitive fitness and sleep patterns so you can make better decisions about your health.

During my PhD I worked with a group of talented scientists at Leiden University and University of Zurich who were researching how your smartphone behavior can be used to monitor your brain health. The algorithms behind this app has been published in multiple scientific publications.

Since then our team at Axite has been working hard to share the insights from these algorithms to you. We are excited to finally be able to share it today.

👩🏽‍💻Happy to answer any technical questions

Huge congrats @charlotte_franenberg 🎉 and everyone at Axite for the launch!

I really enjoyed building @Tacet . It’s been super exciting to work on something that takes everyday phone usage and turns it into meaningful insights about sleep and cognitive performance.

A lot of work went into making the experience feel simple and effortless for users, while doing some pretty complex things behind the scenes 😄

Can’t wait to see what people think and how the product evolves from here. Proud to be part of it 🙌

Hey Product Hunt! I'm Dirk, Data Scientist at Axite.

What excites me most about Tacet is that it makes something real and measurable out of a signal that has always been there. Every time you tap your phone, your brain is leaving a trace. Most people have no idea. The science behind this has been quietly building for years in research labs, and now it is finally in people's hands.

As someone who works with brain data every day, I find that genuinely thrilling. Your phone already knows more about your cognitive patterns than you do. Tacet just shows you what it knows.

Can't wait to hear what you think!

Hello Product Hunt - We are so excited to finally be sharing Tacet with you! It has been a long journey to get here and I hope you love it as much as we do! Happy tapping!

Can't wait to use the app! IOS coming soon?

So happy we finally launched and that people can now use this amazing product! For me the coolest thing is seeing your cognitive scores change over time and figure out what lifestyle works best for you and your brain🧠 Gongrats team!✨

Hi Product Hunt! I'm Nynke, Data Scientist at Axite. It's really exciting to see Tacet go out in the world.

Working at Axite, I’ve learned a lot about the brain and found it fascinating how much subtle stuff in how our brain functions can actually show up in data, often without us even realising it ourselves.

That’s also what I find so cool about Tacet; seeing those kinds of patterns become something people can actually see and explore for themselves from just their phone is pretty amazing.

Very excited to see what people think of it 📱

How can those behavioural patterns be spotted? I understand that activity (mainly that physical one can be tracked by external devices - rings, watches), but what if I forgot to set my smartphone to track my brain activity?

Hi Product Hunt! I am Vasiliki, UX researcher at Axite. Happy to see Tacet officially live on Product Hunt today!

A lot of thoughtful work went into shaping both the product experience and the science behind it. Excited to see where the team takes it next. ✨

I’m curious how you’ve worked through the wellness-scoring tension on the user side.

Cognitive scores are the single most anxiety-loaded thing to quantify, because brain decline is one of people’s deepest fears. What does Tacet do for a user whose score is trending the wrong way: stay observational, suggest action, surface professional help? And from early data, does the score motivate behavior change or mostly generate anxiety without corresponding levers the user can pull? Big congrats on launching!

Heya, this looks cool. Would love to try a TestFlight build once on iOS! Consider posting that build to departures.to (free, I have no relationship with them, just have posted there and gotten lots of good beta feedback users, haha)

Such a beautiful website you have, guys! 😍

支持自然语言查询，不需要复杂的语法。还能设置自定义参数（时间范围、语言、地区、域名过滤），我用它专门追踪日本政府官网的新能源政策，精准获取一手信息，排除第三方解读干扰

Hey Product Hunt! Artem here, co-founder of NewsCatcher.

Back in 2020, Maksym and I were data engineers who couldn't find a reliable way to get clean, structured news data — so we built our own infrastructure. Five years later, it powers intelligence workflows at banks, hedge funds, and risk platforms, continuously indexing billions of web pages.

Today we're launching CatchAll — a web search API that builds structured datasets from the open web.

The web is full of real-world events that never get assembled into usable data: which fintechs raised Series A rounds last quarter, which crypto exchanges faced regulatory action this month, which AI companies were acquired this week. CatchAll finds them all, validates every result, and returns a clean deduplicated dataset — not a list of links.

Submit a natural language query and CatchAll retrieves a massive candidate set, filters out noise, and returns structured records ready to pipe into an AI agent, a monitoring workflow, or an analytics pipeline. You can also set up a monitor to re-run any query on a schedule and push fresh results to a webhook automatically.

We're in early days and genuinely here for feedback. Sign up and you'll get 2,000 free credits to start. Share your use case in the comments and we'll 5x them.

Whoa, I was browsing through some of your datasets. Fantastic!

Love it!

This is interesting but how do we make sure that extracted is legit?

For me, the thing that breaks the dictation habit is the cleanup tax after the words appear.

Dictation is great for raw thought, but different surfaces need different levels of cleanup: an LLM prompt can stay loose, a commit message needs precision, and an email needs just enough polish without losing the spoken cadence. The post-processing rules feel like the right place to solve that.

One thing I’d want is per-context presets plus a quick raw transcript / cleaned text comparison. That would make it easier to trust the tool because you can see whether it’s preserving the thought or silently over-smoothing it.

local-only and no subscription, rare combo! how do you handle the punctuation problem. say "comma" and it adds punctuation? how would "period" be decided as word or punctuation? congrats on your launch!

Congrats. Where do I see pro and free features list?

I'm a solo dev and built this because I needed this and everything was subscription based or had very little control over the output.

I just want to make this the best app ever so if anyone has a suggestion please let me know!

I appreciate you downloading this so much!!

P.S you get 5 watermark free exports, during preview you'll see a watermark though. This is to prevent screenshots otherwise the app would just be easily exploitable lol.

Been on Mockuuups for a while. What's actually different here?

Works for mobile recordings too?

One-time purchase instead of yet another subscription - that alone sells it for me haha. How much time does it actually save compared to just doing this in Figma? Congrats on the launch!

The AI Skills feature is the part that caught me. Turning a repeated prompt into a slash command that deploys to Claude Code, Codex, or Gemini from one place is a real workflow fix. Curious how skill versioning works: if you iterate on a /weekly-review prompt, does the old version get overwritten everywhere, or is there any history? Also interested in whether the MCP server exposes a 'list all skills' tool so agents can discover what commands are available without you telling them first.

Can you share a skill with teammates, or is everything intentionally local?

Congrats on the iteration! As an iOS dev keeping a CLAUDE.md file in my repo right alongside the markdown chaos you described, the missing piece for me is mobile. Daily notes on Mac all day, but the moments I most want to dump thoughts into the memory layer are in transit. Is an iOS companion in the depth-vs-breadth tradeoff zone, or does the Mac-only constraint stay deliberate because cross-device sync would compromise the local-first promise?

Hey again, PH! 👋 I'm Hojong, the solo dev behind Novi Notes — back with launch #2.

Where 1.0 left off

My first launch fixed one mess: my notes — and all my `CLAUDE.md` files, skills, and agent configs — were scattered across dozens of repos, and version-controlling them was a nightmare. So I built one local place where Claude could read and write everything over MCP. No cloud, no API keys.

That solved reading. But two itches stuck around: I was still retyping the same long prompts every session, and my Claude setup was still spread across project folders. So 1.1 closes both loops — let you build your own AI commands, and pull all your Claude assets into one place.

What's new in 1.1:

- 🤖 AI Skills — Turn a repeated prompt into a one-click slash command. "Write this week's review" becomes `/weekly-review`. Write it once in Markdown, deploy the same skill to Claude Code, Codex, or Gemini.

- 🗂️ Workspace — Every Claude and Codex project on your Mac in one tree. Read and edit `CLAUDE.md`, skills, and hooks right inside Novi Notes — no more digging through old folders.

- 🎙️ Meeting recording — Capture mic and system audio together; WhisperKit transcribes it on-device, then your AI shapes it into notes. Nothing leaves your Mac.

- 🔌 Zero-config MCP — Node is bundled in, so connecting is genuinely one click. 1.0 started with Claude; 1.1 adds Codex and Gemini.

- 💾 GitHub backup — Your notes, version-controlled in your own private repo. Your data stays yours.

And the promises haven't changed: one-time purchase, no subscription ever, everything stays on your Mac.

The funny part — making one skill deploy cleanly to three different AI clients in one click turned into its own rabbit hole 😅. But it's the thing I reach for most now.

I built 1.1 for the same person 1.0 was for — someone who lives in the terminal and IDE, leans on AI all day, and just wants one private, local place where their notes and their AI finally meet. Would love your thoughts: if you could turn one repeated prompt into a slash command today, what would it be?

Hey PH,👋

While building our photos app, we noticed something: we obsessively back up our photos, but our most important documents, the ones that actually matter in emergencies, end up getting overlooked.

That's why we built Locker. It's an end-to-end encrypted vault for the few documents and records that really matter, giving your family access to them when you're not around.

Here's what's inside:

• End-to-end encrypted document storage, not even Ente can read your files

• Trusted contacts: add your spouse, sibling, or lawyer, so they can access documents when they need them

• Organize documents into collections (Property, Medical, Financial, however you want to think about it)

• Available on iOS, Android, macOS, Windows, Linux, and web

• Fully open source, your data is yours to leave with, anytime. You can run it on your own server.

• Built on the same encryption foundation that protects 500+ million photos for Ente users today

Do check out the product, we'd be grateful for any feedback!

Hi Nathan! Claude in the sidebar feels like the right shape for browser-AI, congrats on your launch! Quick one: when i switch tabs mid-conversation, does the sidebar follow me to the new tab or stay anchored to where the thread started?

Hey great for safari, you can also add a auto form filling feature to this. I think a good auto form filling feature is still awaited.

Hey Manish, congrats on shipping 🎉

The line that stands out: "secrets cannot live on the same plane as the VM due to prompt injection risk." That is exactly the right instinct, and it is the part most agent infra gets wrong by treating secrets as just another env var inside the sandbox.

Building in an adjacent layer myself (security and permission control between agents and the apps they touch), so the question I keep circling: once secrets live off-plane, what does the agent actually hold at runtime? A short-lived scoped token per call, a broker reference it has to ask through, or does the VM get the real secret injected at execution time and the isolation is purely network-level? The distinction matters because controlled egress stops the agent from calling domains it should not, but it does not stop a prompt-injected agent from misusing a credential it legitimately holds for a domain it is allowed to reach.

Also curious where the skills system sits in the trust boundary, since npx skills add instavm/skills means third-party skill code runs inside that same isolated compute. Who vets what a skill can reach?

Regarding resource allocation how granular can we get with CPU/memory limits per microVM via the control plane, and does it support auto scaling metrics if an agent gets caught in an expensive recursive loop?

The persistent volumes angle is underrated in this thread — most agent infra discussions focus on isolation and secrets but gloss over the memory problem. If an agent's volume can reattach to a new run, that raises an interesting question: how do you handle volume integrity after a compromised run? If a prompt-injected agent writes malicious state to a volume that then gets reattached to a clean run, the isolation guarantee at the VM level doesn't help. Is there snapshotting or rollback on volumes, or is the assumption that the orchestration layer handles what gets reattached and when?

The origin of this was pretty unglamorous. We kept watching people paste the same definition of activation or churn into chat, then into a chart prompt, then into an automation, and every time they'd phrase it slightly differently and get slightly different numbers. The model was doing exactly what we asked but we were just asking five versions of the same question lol.

Skills came out of fixing that for ourselves. You write the definition down once, in plain language, and from then on any agent in the workspace reaches for it when the topic comes up. A new person can ask "how's activation trending" on day one and get the same answer the rest of the team would get, because the agent is reading the same playbook everyone else's agent reads.

If you've got a metric that means something specific at your company and you're tired of explaining it to the AI every single time, give it a try and tell us where it falls short!

Moving definitions out of one-off prompts and into shared durable context is exactly the lightweight semantic layer most teams skip until they're already drowning in inconsistent metric defs. How do you handle drift when someone updates a Skill that's been silently feeding 12 different surfaces — version pin, broadcast, or both?

this is a very handy tool for founders, but I have a question on how are you handling large dataset like clickstream locally and does the data site locally even after ETL?

Congrats on the launch! Super clean concept! Subtitles are practically mandatory for short-form content now, so automation like this is incredibly useful. How does the AI handle audio with heavy background noise or multiple speakers overlapping?

The animated captions and viral preset styles are such a time saver — usually you'd spend hours getting this right in Premiere or CapCut. Really like that it's all browser based so there's no software to download. Would love to see it handle auto-placement too so captions don't cover faces or subtitles.

Portrait captions — do any presets keep text above the TikTok bottom UI, or is it manual every time?

Are we able to also edit the videos from it's transcript like @Descript or this product specialises in subtitles and captioning only?

First of all congratulations to the Autosubtitles team. I just went through your landing page and the feature breakdown looks incredibly clean. I noticed in your FAQ that the video processing stays entirely local on the device and only the audio track goes to the server for transcription. Like this fantastic approach to keep data private. A question came to my mind about the export process. Do you handle the local video burning directly in the browser using WebAssembly or FFmpeg.wasm.... or does the browser canvas handle the video stitching during export?

And definitely dropping an upvote for this. Wishing you the best on the launch today, @davidee.

👋 Hey Product Hunt, I'm Ugur, co-founder of Rigyd.

For the past 4 years, we've built a platform that generates 3D visuals and powers AR try-on experiences for some of the largest footwear and retail brands in the world. We've processed 30K+ 3D assets and delivered immersive experiences to 10M+ users in the last 12 months at artlabs.ai. The goal was to give shoppers a photorealistic look, so we obsessed over mesh fidelity and texture quality.

Then robotics companies started knocking. They wanted millions of 3D assets to populate simulation environments. Sim-to-real transfer works through domain randomization: train your humanoid in 1M different kitchens and a real one isn't a surprise when it's deployed.

That's when we realized that looking right and behaving right are completely different problems. These teams needed proper collision meshes, physically accurate mass, realistic friction, and more. Not just pretty geometry and texture.

Most 3D assets weren't built for physics. A robot trained on them learns nothing useful. It falls through floors, grips through objects, and generalizes terribly to the real world.

And it's not the simulators holding things back. Isaac Sim, MuJoCo, and Gazebo are widely deployed. Newer engines like Genesis have arrived with even faster runtimes. The physics engines are mature. The 3D content going into them isn't.

The current fixes fall short. Teams either manually create or annotate every asset (doesn't scale) or accept platform defaults and hope the policy generalizes. Neither works.

So we built Rigyd. It ingests raw 3D, images, or text and produces validated OpenUSD (with USDPhysics schemas) and MJCF, with collision meshes, mass, friction, restitution, material properties, and semantic labels baked in. Works in Isaac Sim, MuJoCo, and Unreal out of the box.

No manual annotation. Weeks → minutes. From any input.

Our early users are already using it for
→ Warehouse automation
→ Simulating surgical rooms
→ Testing robotic arms before deployment

The UI is for exploration. The real surface is programmatic. Rigyd exposes tools so agents (or your own pipelines) can generate millions of physics-accurate sims directly inside NVIDIA Isaac Sim, MuJoCo, or wherever you train.

If you're building anything in robotics, AV, industrial automation, or embodied AI: claim the free credits on signup. All we'd ask is honest feedback. What works, what doesn't, what you wish it did differently.

If you've hit the sim-data wall and want enterprise/bulk API access, DM me. We're onboarding a small group of design partners in June.

If you just want to nerd out about OpenUSD, domain randomization, or sim-to-real transfer, we're in the comments all day.

Robotics is going to be the largest industry in history. We think the data infrastructure for it should be as easy to use as Stripe is for payments. That's what we're building.

Check it out 👉 rigyd.com

Ugur

Congrats on the launch! Looks great. Curious what asset formats you ingest (OBJ, FBX, GLB, STEP, USD)?

How do you validate that the physics actually behaves correctly post-generation? Do you run automated sim rollouts, compare against ground-truth measurements, or rely on schema conformance only?

Wow, quality of the generated assets are impressive. With all the materials and physics. 3 credits for trial is not enough though to test all the features.

Congrats on the launch. Also love the OpenUSD angle. I believe like the ecosystem is converging around more standardized pipelines instead of every company building tooling around meshes && annotations :)

One of the biggest bottlenecks in robotics today is not the model, it’s the data. Rigyd tackles a massive problem by making physics-accurate synthetic training environments scalable and practical.

Training across millions of variations is exactly what humanoid and robotic manipulation systems need to generalize in the real world. Excited to see where this goes.

This is cool. Curious how are you generating collision meshes under the hood? VHACD-style decomposition, SDFs, or custom approximations?

Congrats and good luck on the launch! Looks amazing

Congrats on the launch! Rooting for Rigyd today.

I love it!

Do you randomize visuals too, or is Rigyd mainly focused on physics and geometry?