PH热榜 | 2026-03-24

Claude’s new “computer use” turns AI into a true digital employee. Probably the biggest update yet by @Claude by Anthropic?

It solves the gap between thinking and doing by letting Claude control your screen, apps, browser, and workflows with permission autonomously.

Claude uses your connected apps first: Slack, Calendar, and other integrations. When there's no connector for the tool you need, it asks for your permission to open the app on your screen directly, just like a human using your computer. Anything you can do on a computer — Claude can!

Assign a task from your phone, turn your attention to something else, and come back to finished work on your computer. Tell Claude once to scan your email every morning or pull a report every Friday, and it handles it from there.

Key features:

• full desktop control

• mobile task assignment (Dispatch)

• app integrations

• autonomous workflows

In just one week, Anthropic has shipped 9 features culminating in what feels like a fully autonomous digital human.

Available on Pro and Max. Update your desktop app and pair with mobile to try.

• https://claude.com/product/cowork#dispatch-and-computer-use

• https://claude.com/blog/dispatch-and-computer-use

• https://support.claude.com/en/articles/14128542-let-claude-use-your-computer-in-cowork

• https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool

Perfect for founders, operators, devs, and busy professionals handling repetitive tasks, reporting, and execution.

I love the recent Claude released features.

I now use Claude for almost every working task during the day.

I just started automating video generation for my startup using Claude, and it drives me crazy.

I just installed one skill, and it generates quite good videos from one prompt, and I'm able to iterate with it, asking it to cut the videos, add music, assemble one video from different videos, and that just drives me crazy.

I'm looking forward to seeing what else Claude will bring in the next year.

I think it will entirely change the way we work on a daily basis.

How secure is it? I still don't feel comfortable running openclaw on my machine due to it being inherently insecure.

HI, do you know how to apply it on claude? I click visit website but it is claude website, do you guys know where is the "Claude Computer Use"?

Only for Mac?

Claude is giving updates every other day :)

now, i can catchup with my cc while i am on a walk.

This means Claude desktop needs to be always running on a Mac, right? It would be funny to start working on closed Macbook in my backpack 😄

Ow waw I have been waiting for that. But I am on the max plan and still dont see it! Why?

It looks amazing - I will still use openclaw on my mac mini but use this + despatch on my actual macbook. I still think it doesn't quite nail the sheer personalisation that Open Claw offers (and the OSS nature of it).

I wonder if I can have it go to Claude.ai and prompt itself...

Is this feature currently available on the Pro Plan? If not, when do you expect it to be?

This keeps on getting better @Anthropic! Testing it now.

Lol I feel like claude is dropping everyday, and I use claude for almost everything and I just love it, but it seems like they are taking the whole AI world to the next level this is scary but exciting and I can't wait to try it out

the speed at which they ship new features is unreal. this basically just replace openclaw haha.

Really interesting direction.

From the outside this reads like an AI assistant gaining automation capabilities.

But the way it actually behaves feels closer to an execution layer for agents — bridging the gap between reasoning and real-world action.

Instead of just generating outputs, it can operate systems, navigate interfaces, and complete workflows directly.

If this evolves, it seems less like a feature and more like a foundational layer for how agents interact with computers.

Curious how you think about this internally.

Does my computer need to be open and on, and do I have to sit and watch the work happen?

What visibility do users have into what the agent is doing in real time, and how easy is it to interrupt or correct it mid task?

This update to Kitty Points comes with two big updates, one visible and one more backend.

The visible update is a brand new leaderboard! And it's not just an all time leaderboard. If you just discovered Product Hunt in the last year, month, or even week it could be a bit daunting to see some of the Kitty Point totals that long time members have accrued. At the same time, we want to give a huge shoutout to our community members who have been around and consistent for years and have contributed a ton of phenomenal content to the site. The answer to this is a leaderboard that shows the top users by Kitty Points per week, month, year, and all time. This means every week you have a new chance to climb the leaderboards and get higher visibility on your profile and what you're building or posting.

The less visible, but still exciting, part of this update is that we've re-worked how Kitty Points are calculated on the back-end. I'll say, this part of the project was the closest I've ever come to feeling like a game designer! We also want to be more upfront about what you can do to earn Kitty Points. You can't play the game if you don't know the rules! So, here are the all of the contributions that will earn you Kitty Points:

• Being the Maker of a launch

• Being the Hunter of a launch

• Posting detailed reviews

• Creating discussion threads

• Leaving meaningful comments

For all of those, content and impact matter. Being the Maker of a featured product at the top of the leaderboard will obviously gain you a lot of Kitty Points. As Matt mentions in his comment, Product Hunt is launch forward, but there are a lot of other ways to contribute to the community as well. A large part of the backend changes we've made open the door for other types of contributions to gain you more Kitty Points than in the past. So even if you're not a Maker, you still have the chance to contribute to the community and climb the leaderboards. And if you're curious, you can check out our help center for more information around what kind of content aligns with Product Hunt's guidelines.

We've built this to elevate and show off the members of our community who are helping us create an incredible space to discover, evaluate, and discuss incredible products being created by excellent people in tech. I'm very excited to see who shows up on the leaderboards every week!

I also want to give a massive shoutout to @catt_marroll who did a tremendous amount of work taking the new Kitty Points formula and making it work on a weekly, monthly, and yearly basis (something we've never had before!) and building a really awesome and fun leaderboard page around it.

Product Hunt is launch forward (we do not plan to change that!), but this isn't the only way to participate in our community. Leaving reviews, providing thoughtful comments, and building vibrant forum discourse also help boost the ecosystem. We want kitty points to reflect the different ways users "play the game" here on Product Hunt.

Additionally, like many online spaces, there is a value in being able to qualify who you are talking too here! Advice from Paul Graham is likely worth more than a remark from a sock puppet account created a few minutes ago. We hope to use this new leaderboard as a way to incentivize authentic participation, and qualify content around the site.

I really like the idea around that, it wouldn't come to my mind that KittyPoints can have this use-case and we couldn't find out what they are for. Now, one of the top secrets of this platform is already unveiled!

Good job!

I'm slacking at #3 for all time

love it!

Whattttt? PH launching on PH?

Wow, I am at #44 all time. I also just launched @MindPal again today, so I guess that would earn me even more points 😎 Hope I will make it to top 10 this year. Question though: does earning a Maker of the Year award contribute to our Kitty Points?

Do KP's degrade over time or are cumulative? I'm seeing something strange - I seem to have more KP's last year than I do All time. Would love to understand how that works @jakecrump @catt_marroll

The Kitty Points Leaderboard is a brilliant move to boost community engagement! By gamifying contributions through upvotes and streaks, it effectively turns passive browsing into an active, rewarding experience. It’s a great strategy for improving user retention and making the community feel more alive. Can't wait to see more perks tied to these points!

does a person's KPs impact how much weight their upvotes/comments carries on launches?

@makers repeating my request here: I'd love for the Kitty Leaderboard to be added to the Product Hunt API so I can add it to the Product Hunt for Raycast extension!

Yeahhh! I didn't wanted, I needed it. Thanks for adding this engaging stuff and making it funnier and enjoyable. Keep on that way team

Today, I learned, that I am a nobody. But I have a goal, and this goal involves kitty points. I don't know why I need Kitty Points, but "a tiny competitive spiral " for seeing your own rank = "insane drive" for not seeing a rank at all. I must haz it.

Maybe this is answered elsewhere, but is "last year" 2025? Or is it the last 365 days?

love it. veterans might remember when back in the days @abadesi was shining a spotlight on the most helpful makers of the week.

what if we bring it back to The Roundup? keep up the great work

Hi Product Hunt! 👋

We are excited to launch Cekura Monitoring for Voice and Chat AI companies. Most monitoring tools tell you if your AI is up. Cekura tells you if it is behaving.

When we had first launched Cekura QA, we thought we had solved the problem for both testing and monitoring . But as our users scaled, we noticed a painful pattern: While pre-production QA was automated, teams were still spending dozens of hours manually listening to thousands of calls.

The two big blockers we saw were:

1. The Scaling Wall: Defining and optimizing custom metrics was taking too long, forcing teams back into manual spot-checks.

2. Production Blindspot: Standard LLM metrics misses the Customer Experience in Voice AI - things like agent tone and customer sentiment that actually defines customer success.

We have rebuilt the monitoring layer from the ground up to solve this. Cekura Monitoring turns that "wall of noisy logs" into actionable signals.

🚀 What’s New in Cekura Monitoring:

• 30+ Predefined Metric Suite: We track what actually breaks Voice and Chat agents across four critical categories:

  • Speech Quality: Voice clarity, pronunciation, and gibberish detection.

  • Conversational Flow: Silences, interruptions (barge-ins), and termination triggers.

  • Accuracy & Logic: Hallucinations, transcription accuracy, and relevancy.

  • Customer Experience: CSAT, Sentiment analysis, and drop-off points.

• Metric Optimizer: Stop "vibes-based" prompt engineering. Define a metric (e.g., Successful User Authentication), tag 20 calls in our Labs interface, and our optimizer "compiles" a prompt that aligns with your specific feedback.

• Statistical Intelligence: No more fixed, noisy thresholds. Our Alerting Engine learns your agent's baseline and only pings Slack when metrics shift 2σ from historical norms.

• Automated Cron Jobs: Set up recurring health checks to simulate production conversations. Catch silent failures and regressions before your customers do.

• Visual Dashboards: Real-time distribution charts for each metric. Views customized for each stakeholder

Who is this for?

Teams scaling Voice & Chat AI who are tired of listening to calls manually and need a way to prove their agents are actually working.

Sign up and try for free at cekura.ai or drop your questions below! We would love to hear how you’re currently handling Voice and Chat AI in production👇

One of the most common issues we see voice agent makers run into is their agent keeps interrupting the caller. It's frustrating for users and easy to miss during development. With our interruption metric, teams can catch this early and fix it before it reaches real users, and that's just one of the many predefined metrics we offer out of the box, try it now!

We are thrilled to share Cekura Monitoring with the PH community!

Most teams focus solely on whether a voice AI agent reaches the 'correct' outcome, but they often overlook the nuances that actually define the user experience: tone, transcription accuracy, TTS quality, and pronunciation.

While working on scaling to handle thousands of parallel calls, we realized just how easily these small details can degrade at volume. Cekura was built to ensure your agents don’t just work but they sound perfect.

Check out the product and let us know what you think!

Blind spots in production voice agents are brutal — you don't know your agent is skipping verification steps or missing required disclosures until a compliance team surfaces it weeks later. Monitoring 100% of live calls at the session level rather than spot-checking is the only real fix. The P50/P90 latency tracking and interruption detection on production traffic is also underrated — that's where infrastructure regressions hide.

So excited to see this live! 🎉

Been working closely on Cekura's monitoring features and what makes this special is how much it closes the loop for conversational AI teams — you're not just testing in pre-prod and hoping for the best, you're getting visibility into what's actually happening in production calls.

This one's been a long time coming! 🚀

Really excited to see this out 🎉

Working on alerting and simulation quality made it clear how hard it is to catch subtle regressions early—this is a big step toward making that reliable in production.

Glad to finally have this live 🚀

How are you different from tracing platforms like Braintrust and Galileo ? Except Voice metrics.

What aspects of voice does it capture? I wanted to test on tonality and personality of my voice agent, is it achievable?

The "is it behaving" vs "is it up" distinction is spot on. We've had AI chat agents pass every health check while giving completely wrong answers to customers. Uptime metrics are useless if the AI is confidently hallucinating.

How granular does the sentiment tracking get? Like can it detect when an agent starts being passive aggressive or gives a technically correct but unhelpful response? That's the stuff that kills user trust slowly.

Excited to see this go live! 🚀

Working on our voice simulations and agent stack taught me that reliability is all about the nuances. We built Cekura to give developers the specific visibility needed to master those details and move past the guesswork.

Can't wait to see everyone dive into the labs and start leveling up their agents!

Big congrats to the @Cekura team on the launch! 🚀

Super excited to see this out!

Got to work closely on the metrics side of things. Seeing it come together into something teams can actually rely on in production is incredibly satisfying.

Huge shoutout to the team for pushing this across the finish line.

Huge congrats to the team! 🚀 such a solid group of builders. This solves a lot of different use cases - instant alerting, human in the loop reviews, A/B testing and more without feeling cluttered.

The silent production failure detection is what catches my eye. When you're running AI agents in prod, the scariest failures are the ones where nothing errors out - it just gives bad output for days without anyone noticing. Curious how Cekura handles the baseline drift problem - do you need a human to label 'good' vs 'bad' outputs, or does it pick that up automatically?

⁠Is the metrics customizable ? For example I need to define success criteria by peak latency and not mean latency

Can we use Cekura to benchmark STT / TTS separately as well or its only used for Voice AI agents ?

Congratulations on the launch!!

Do you guys also support on prem deployment to ensure privacy?

Congrats on the launch 🚀
Really important problem to solve!

This is something we've been looking for. We deploy voice and chat AI agents for businesses (support, qualification, scheduling) and QA has always been the manual bottleneck — listening to call recordings, checking if the agent followed the script, catching edge cases.

The 30+ predefined metrics and CI/CD integration is exactly what's needed to ship agent updates with confidence. Do you support Vapi-based voice agents out of the box, or does it require custom integration?

Congrats on #2, @Cekura

Just flagged a UX loop on mobile signup ,it's showing 'User Not Found' and forcing a logout for new users. It looks like a system crash rather than a filter.

I've got the fix details ready to help you keep your conversion high today. Where can I send the report?

this is super duper cool. future of voice

Love the sped at which this team ships! I was curious do you also have plans to roll out observability for images/video agents?

Congrats on the launch, team!

What are the challenges come when teams tries to build this internally?

Congratulations on the launch team @Cekura

Are these prefedined metrics all on Audio or text based ?

Cekura is the best voice ai evals platform out there! We love it!

Cekura 🚀🚀🚀

A lot more to come!

🚀 I’m so proud of the work we’ve done on Cekura Monitoring. I personally worked on the Smart Metric Alerting engine, which saves Voice and Chat AI teams from scrolling through thousands of calls. Now, you only get a ping when something actually feels off.

The best part? The customization. It allows our users to tune out the noise and focus purely on the performance metrics that define their success. It’s a total game-changer for anyone scaling AI agents.

Really helpful feature.

Congrats team!!! Do you support real-time streaming analysis or is it batch processed right now?

Would love an API-first version of this for deeper integration into internal tooling.

I'm writing this from a hackerhouse in Barcelona ☀️

Last week we launched OpenViktor (an open-source AI employee built in 48h). We were #3 PH of the day, 300+ GitHub stars in 24h. We took it down and rebuilt the entire thing from scratch.

Meet Jared. The first AI employee that's actually social.
He lives in Slack, connects to 10,000+ tools and does the work: reports, dashboards, code, follow-ups, research.

But here's what's different: he reads the room. Jumps into conversations when it matters. Knows who to talk to and when to shut up. Brainstorms with your team. Remembers everything and gets sharper every day.

Paying $2000/month for an AI employee is crazy.
Humalike is backed by the first investor in ElevenLabs. Built by a small team (🇪🇸 X 🇵🇱) that hasn't slept much.

Martí, co-founder.

P.S. We ship fast. Request a feature or report a bug, we'll build or fix it the same day :))

Hi everyone! I'm Mateusz and I'm leading Jared on the tech side. 💻

We built Jared to do any task, be social, form his own opinions, understand people and be proactive.

It's not a tool. It's a partner that takes responsibility for his job, gathers input from people in your org if needed, delivers the job and maintains it - all without extra setup. He's built that way.

I will be active on PH today to answer any questions you may have :))

Hey, it’s Ignacio, part of the team here, we are super excited to launch Jared, add it to your Slack workspace and let us know your thoughts on it!

I’m sure you won’t be able to live without him after trying it ;))

This is interesting—especially the “reads the room” part.

How does Jared decide when to speak up without becoming noise? Is there some kind of confidence threshold or context awareness model behind it?

Good luck on your launch!

Maks here, founding engineer, built this with Mateusz 👋

We wanted Jared to feel like an actual teammate, not ChatGPT with Slack interface. Drop any questions below 🙂

What happens when it jumps into a public channel and gets something wrong? That feels like the moment that can either build or break trust with the team. Is there a way for people to correct it so that feedback actually shapes how it behaves next time?

Congrats on the launch! This is a really interesting idea that brought a lot of questions to my mind.

How does it behave in edge cases and handle conflicts in data within a group? How was the memory organized? And is there a weighted decision-making framework according to team hierarchy, or based on how recent the information is?

It is very cool for an agent to adapt to the social narrative within a group. Wish you the best with the updates!

This app is not approved by Slack?

Looks awesome! I like the idea of the bot being able to collect all the information it needs from the existing conversations, but can we also point it at external sources to use as reference when doing its work?

Hey team! Great work, curious to learn what are the exact use cases Jared can perform? Is it mainly coding? or generic tool execution? or something else?

I like the idea of an AI that doesn’t just execute tasks but understands context inside conversations.

Does it naturally adapt to how a team works?

The "reads the room" part is what I find most interesting. Proactive agents are way harder to get right than reactive ones - getting the timing wrong and it becomes more annoying than helpful. How do you decide when Jared should chime in vs stay silent? Is it rule-based or does the model decide?

this looks really cool. the part about Jared following conversations and speaking up without being asked is what got my attention. how does it decide when to jump in vs stay quiet? and does it work across private Slack channels too, or only public ones?

Congrats on the launch.
Loved the video! Landing page is a bit unreadable.

which llm models are being used under the hood and are they configurable?

What about data privacy? Seems like this can expose our whole operation to whoever?

What kind of work does Jared do? I usually want coworkers who do things not just yap, maybe Jared does stuff

What safeguards prevent it from interrupting or adding noise in fast moving team conversations?

Amazing and scary all at the same. All aboard the Ai productivity revolution!

finally i'm helpful

Hey Product Hunt 👋

I'm Swastika, devrel at Drift AI and I'll be honest, I'm the newest person on this team. I joined not too long ago, came from a completely different corner of tech, and knew next to nothing about robotics when I walked in.

That's actually why I wanted to drop a comment here.

Because if someone like me, with zero robotics background, can get Drift running, set up a simulation, and start actually understanding what's happening under the hood, then I genuinely believe any developer can. The whole point of Drift is that you shouldn't need to be a simulation expert to work with one.

That said, getting started with any new tool has its rough edges, especially on day one. So I'm here. If you're trying to install Drift and hitting a wall, if something isn't working on your Mac (Drift runs on Ubuntu but works well on MAC with VMware), if your simulation isn't launching or your ROS environment is being weird; drop a comment below or reach out directly. I'll be watching this thread all day and will personally help you.

Really proud of what this small team has built and excited to see what you all do with it.

Give it a try ❤️🦾

Hi PH - I am Nikhil, Co-founder of Drift. With my team : @sanjil_j and @swastika_yadav1

We spent the past 10+ years obsessing over one question: why is it still so painful to get a robot into reality?

Get started easily : Docs for quick start

Engineers spend 60% of their time building simulations but -

• Simulation environments always get into setup & runtime nightmares.

• Robotics engineers feel like software or IT engineers fixing these issues, taking away their focus from the actual robot, the physics, the maths - which is their actual motivation to build in robotics!

• Current coding agents don’t have the necessary context, control and understanding of orchestrating - ROS, simulator, OS, plugins together to get simulations running reliably.

Drift fixes this.
You describe what you want, it handles the rest - generating/editing robot & world description, publishing ROS nodes, setting up the simulator, generating controller configs, launch files, building the workspace, launching the sim. When something breaks, it inspects running nodes, checks topic connections, traces the command chain, understands from previous success-fail chains and fixes it.

Right now Drift is completely free in public beta.
Happy to answer any questions here and would love your feedback.

Try it, break it, and tell us what’s missing.

Join our discord for updates: Discord

<“>

Congrats!! Any plans to support cloud-based simulation runs?

Hey, drift is awesome! I'm a mac user, does drift work for me or is this linux only? A lot of great robotics tools end up being ubuntu only and it's always a pain for mac users….

How does Drift handle failures mid-simulation? Does it auto-recover or suggest fixes only?

Congrats. How Drift compares to existing ROS tooling in terms of debugging depth?

@swastika_yadav1

Can I bring my own existing ROS workspace into drift, or does it only work with robots it generates itself? I have my own robot I've been working on.

Vibe coding for robotics is a space I hadn't thought about but it makes total sense. The hard part with ROS has always been the orchestration overhead - knowing which nodes are running, what topics exist, debugging why the simulation drifted. How much of the prompt-to-simulation pipeline is Drift generating vs. stitching together existing ROS packages?

been wrestling with ROS simulation setup for our hardware integrations and this looks like exactly what we needed six months ago. love that it handles the OS orchestration too - that's usually where things get messy. how does it perform with real-time constraints when you're testing control loops?

Really impressive — the idea of prompt-to-simulation for robotics is a huge unlock. The fact that it actively tracks ROS states and fixes issues on the fly is what separates this from just being another wrapper. Curious: do you have plans to support multi-robot swarm simulations? That would be incredible for drone research. Congrats on the launch!

When Swastika reached out to me a week ago, I immediately told her this is going to be a good launch. Largely because I saw Antler and second, the niche you're building is very refreshing for Product Hunt audience. Many congratulations Nikhil and team for launching and securing the Product of the Day. BTW, Do you have a public roadmap?

What visibility do you have into the generated ROS stack and control logic?

Does Drift support custom plugins or only predefined simulation components right now?

Gemini in Chrome is Google’s attempt to turn the browser itself into an AI-native workspace.

Most AI tools live outside your workflow — constant copy-paste, tab switching, and lost context. Gemini fixes this by living inside Chrome’s side panel, working directly on what you’re already viewing.

What’s different: Instead of being another app, it embeds AI where work already happens — your browser — with deep integrations across Gmail, Calendar, YouTube, and more.

Key features:

• Summarize pages, research papers, and YouTube videos inline

• Compare across multiple tabs (e.g. product comparisons)

• Draft & send emails without leaving your tab

• Schedule calendar events via prompts

• In-browser image transformation (Nano Banana 2, no uploads)

• Memory of visited pages to reduce tab overload

Benefits: Less context switching, faster workflows, and a more continuous thinking environment.

Who it’s for / use cases: Students, researchers, shoppers, and knowledge workers who live in their browser and want AI embedded directly into their daily flow.

Google is betting the browser is the true home of context, not standalone AI apps. If you’re tracking where AI is actually getting useful (not just flashy), this is one to watch.

Get started here: https://www.google.com/chrome/

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

I tried using Gemini before but it isn't compatible with all devices. Does this one work in all devices?

Is this new? Seems like the blog post is just an expansion into more languages/countries?

Guy after I updated Chrome, it started switching off on its on in middle of my meets :)

I hate creating forms especially longer ones with complex logic. This is great and solves a nice pain point.

Have used Jotform before and it always was a quite smooth experience. This seems like it will make things even easier in the future! Congrats on the launch!

As a huge Jotform user since the pandemic, I love how it always gives me a customizable look. Congratulations on the launch, @aytekintank!

Hey Product Hunt! I'm Evert, a solo dev from Brazil. I built Maestri because I was drowning in terminal tabs while working with multiple AI coding agents.

The idea is simple: an infinite canvas where each terminal is a node. But the feature that changes everything is agent-to-agent communication. Drag a line between two terminals and they collaborate. Claude Code asks Codex to review its code. No APIs, no middleware, just PTY orchestration.

Built entirely in Swift with a custom-built canvas engine. A full whiteboard with shapes, arrows, freehand drawing, markdown notes, and terminals, all on an infinite canvas. No Electron, no web views. The AI companion (Ombro) runs on your Mac through Apple Intelligence. No account needed, no telemetry.

1 workspace free. $18 lifetime for Pro.

Would love your feedback, especially from anyone juggling multiple agents daily. What's working? What's missing?
themaestri.app

Hey, I'm a fan of native mac apps and your app look promising, especially for keeping an eye on what your agents are doing and having a visual and organized canvas. Can 2 codex instances can collaborate as well? Congratulations on your launch! 🚀

The "drowning in terminal tabs" problem is so real — and the agent-to-agent communication via PTY orchestration is a genuinely novel approach. We run multiple agents for our own product (SoundInsight, music marketing analytics) and the context-switching tax is brutal. A canvas where they can literally collaborate in the same workspace changes how you think about multi-agent workflows. Solo builder shipping a native Swift app with a custom canvas engine on launch day — respect, Evert.

Stringing multiple coding agents together usually turns into a tangled mess of terminal logs, so mapping them out on a spatial canvas makes total sense. I can see this being incredibly useful for having one agent draft a database schema while another concurrently builds the API routes. I am curious how you handle state sharing and context window limits when several agents are interacting at the same time.

What happens when the canvas gets dense with terminals and connections, does navigation and clarity start to break down?

solo dev and building a whole canvas engine in swift? respect. i run like 4 terminal agents at once and keeping track of what each one is doing is a nightmare so this makes a lot of sense

TypeScript 6.0 marks a pivotal moment in the language's evolution—not due to headline-grabbing features, but for the groundwork it lays.

But if you’re already familiar with the language, you can get TypeScript 6.0 through npm with the following command:

npm install -D typescript

This will be the final release built on the original JavaScript foundation. TypeScript 7.0, currently nearing completion and developed in Go, is set to deliver significantly faster compilation with native code and parallel type-checking.

Version 6.0 serves as the transition point. It updates key defaults (strict mode enabled by default, ESM prioritized, ES5 target dropped), phases out outdated patterns incompatible with the native rewrite, and introduces the `--stableTypeOrdering` flag to assist teams in validating their migration path to 7.0.

That said, there are meaningful upgrades here: native support for Temporal types, `Map.getOrInsert`, `RegExp.escape`, improved type inference for methods, and a default type system update that can boost build speeds by 20-50%.

For teams using TypeScript, this release is the signal to begin preparing. The shift to the native platform is just around the corner.

Launch day is here. Congrats man, @chrismessina!

That "last release built on JavaScript" line is straight fire. So honest it made me stop and pay attention

I took my time going through the homepage. The layout looks super clean and fresh. But lemme tell you what actually grabbed my attention…

You have a section that shows TypeScript catching errors in real time. That's your strongest proof. A developer landing on the page needs to see that right away.

Right now it's under "Adopt TypeScript Gradually." A visitor has to scroll through a wall of code to get there.

So it's better to pull that demo up. Let it be the first thing people see. Show them how TypeScript saves time before explaining how it works.

Just a thought from someone who looks at too many product pages. Attaching the screenshot below:

Nice concept. How are people actually distributing these AI hubs, though? I think many people tinker around with vibe coding platforms, but most are only personal or toy sites, and I haven't known anyone actually making money with them yet. Do you know any case studies I can look at?

How is this different from just building a chatbot?

Hey everyone! I'm Tuan, co-founder of MindPal

Sylvia covered the how — I want to talk about the why behind the architecture

We made a deliberate choice: MindPal handles the intelligence, your vibe coding platform handles the interface. We don't try to be both

That means you're not locked into our UI. You own your hub entirely — the code, the hosting, the design. MindPal powers what's underneath: the AI agents, the multi-agent workflows, the knowledge training, the conversation memory, the analytics

This separation is what makes the "one prompt" approach possible. Because MindPal already exposes everything through embeddable components with full API control — custom user IDs, session context, usage webhooks — the vibe coding platform just needs to assemble the pieces. That's why one prompt can generate a complete hub with auth, profiles, chat history, and a paywall. The hard parts are already solved on our side

What excites me most is what this unlocks. We've seen coaches replace $20K/month of 1-on-1 time with a hub their clients use daily. Course creators adding AI tool hubs to their membership and seeing retention jump because members are actually using something, not just watching. Consultants packaging their methodology into tools they sell independently, generating millions. We didn't invent any of these use cases — our users did. We just made it possible to build the whole thing in an afternoon

If you have questions about the technical side — how the embedding works, how user context flows between your app and MindPal, how the usage webhooks fire — I'm here all day. Happy to get nerdy about it

feels like the hard part here is not building the hub, but giving people a reason to come back to it
how are you thinking about that?

Interesting!! Feels like building is getting easier, but distribution is still the hard part. How are your users usually getting their first real users?

The "one prompt to a full hub" approach is really clever for consultants and coaches who need client-facing tools but don't want to learn to code. I've seen so many people in that space cobbling together Zapier + Typeform + Notion and it's always a mess.

The vibe coding output (Base44/Lovable/Bolt prompt) is a nice touch too. How customizable is the generated hub after the initial build? Like can clients tweak individual agent behaviors without breaking everything?

The vibe coding + paywall stack is pretty clever - using MindPal for the AI logic and something like Lovable or Replit for the wrapper removes a ton of the boring infrastructure work. I've been building AI tools for a while and the auth/billing piece is always what kills momentum. Does this work for tools that need to connect to external APIs or databases, or is it mostly self-contained AI workflows?

While building agents is easier than ever, managing auth, usage tracking, and global state is still a bottleneck. We’ve abstracted that infrastructure into a branded, Netflix-style hub so you can deploy and monetize your expertise in minutes.

Quick note for the builders here, you're not locked into one platform. Use MindPal for the AI brain, then generate the full hub with Base44, Lovable, Replit, or whatever vibe coding tool you prefer. Auth, paywall, chat history, all handled out of the box. What's your go-to stack for shipping fast these days?

Hey PH! Yan here, founder of NextPhone.

At the beginning of 2025, voice AI was finally getting good enough to handle real phone conversations. So I built an iOS app called TTYL — an AI answering service for solo business operators. It took off. 4000+ of solopreneurs use it today, and it's handled over 500k calls.

But I kept getting the same request from slightly larger businesses: "We love how simple TTYL is, but we need this on the web. We have a small team, multiple calendars, CRM integrations - can you build something for us?"

So that's NextPhone. The same dead-simple setup that solo operators loved about TTYL, but built for growing service businesses — web-based, with real-time calendar booking, CRM syncing, and the integrations you actually need.

Here's what it does: NextPhone picks up your calls in under 2 seconds, answers questions about your business, qualifies the lead, and books the job directly into your calendar. We train it on your website and you're live in less than 5 minutes. We give you a US number or you can port over your existing one.

A few things that matter:

- Flat plans with unlimited calls, no overage fees
- Actually books appointments in your calendar
- Instant human transfer to your team
- 10+ languages

We're being used by insurance brokers, law firms, HVAC companies, roofers, pest control, photographers - basically anyone who can't afford to miss calls but also can't sit by the phone all day.

The 7-day free trial is completely open — unlimited calls, full features, cancel anytime. Would love for you to try a demo call and tell me what you think.

Tried it for my cousins construction company last month and it has enabled him to remove himself as bottle neck of his business with 20+ missed calls a day.

He loves it!

Ive been paying Ruby Receptionists like $400/mo and honestly they still put people on hold and get stuff wrong half the time. if this is even close to as good as the demo im switching

Flat $199/mo unlimited calls is a gutsy pricing move in a space where everyone's charging per minute and hoping you don't notice. Respect the transparency

Really like the <2 second pickup time — that's the benchmark we've seen matters most for caller retention. We build similar AI voice agents for service businesses using Vapi + Deepgram + Twilio stack, and the flat $199/mo pricing is a bold move vs. per-minute competitors.

Curious: how do you handle the handoff when the AI can't resolve a call — does it warm-transfer to a human with full context, or does it schedule a callback? That's been the trickiest part in our deployments. Congrats on the launch!

Some case studies and examples of actual voice calls in play would be beneficial. I just can't imagine letting an AI out the box handle the most valuable part of the pipeline without evidence that it worked so well in many contexts.

Smart to go flat-rate, that's a big differentiator in this space. How do you manage token costs internally though? Are you basing margins on averages across the customer base, i.e. some you win some you lose?

Does it also pick up specific tone or like the brand voice of your brand? Congrats on the huge launch, @yanismellata!

Quick question - can it handle more complex intake stuff? Like in PI law we need to ask about accident date, if theyve seen a doctor, insurance info etc before we even know if we want the case. Can you set that up or is it more basic?

My level of learning is the same. Using DuoLingo for 1100+ days and not able to create some reasonable sentence :D

I am nearly 500 days on duolingo and I started to really feel that i am using it just to keep the streak going.
The articles based approach seems pretty interesting.
Good Luck!

love this. this would solve the problem of me having to open the translator each time i see an unfamiliar word in a book i’m reading :) simple and powerful!

This is great fun, if you could choose preferred content like techmeme. I'd look through this app in that case.

I totally relate to this because I started learning German with Duolingo and it didn't work for me. Learning through real articles is definitely the better way to go.

Congrats, everything is so cute, i love it.

Another owl victim here 🦉 Had 200+ days streak and realized I was opening the app for the streak, not for English. The moment I missed one day - zero motivation to come back.

Learning through real articles is double useful - you actually learn the language AND stay informed. That's smart.

@pungme love the real-content approach, but how does it handle difficulty levels? Can beginners jump straight to news articles, or do you recommend starting with simpler sources? And does it suggest articles based on vocabulary level? Thanks)

Very cool idea — I had exactly the same problem learning German.

I used Duolingo and Assimil for a while, but I kept getting bored with the lessons, and the gap between learning apps and real newspapers always felt huge. When I tried reading actual articles, it was just too hard and frustrating.

I’ve always felt there was a real need for a tool that gradually increases difficulty while keeping the content interesting and meaningful.

Learning through real content feels much closer to how we actually acquire languages.

Love the "no streaks, no gimmicks" philosophy too — focusing on real progress instead of engagement tricks makes a lot of sense.

Congrats on the launch!

The Explore tab pulling from real news sources like Tagesschau and Le Monde is what makes this stick. Reading graded content never felt like actual language use, but a real German newspaper article with inline translations and the option to tap away words you already know bridges that gap. Zero cloud dependency for translations is a nice touch too... on-device processing means it actually works offline, which matters when you're commuting or traveling.

been looking for something like this for ages. saving reels is easy but finding them later is the real problem. does the auto-organize work based on content type, or do you set the categories manually? also wondering if there are plans for a browser extension to save from desktop too?

This is useful imo, I have my MEMEs all over the platforms. Does it save locally within the app? Or when someone deletes the content from their profile, will it be reflected in this app too? (I mean, the content will disappear in Ordo as well).

If Ordo can retrieve that TikTok you saved three months ago when you can't recall the title, that's the part that matters. Most bookmark apps become digital graveyards because the find step never works.

The auto-organize for saved reels is the piece no one has gotten right yet — every other tool just dumps things into a flat list you never dig through again. Love that this works across Instagram, TikTok, and YouTube in one place. We build tools for independent musicians, and artists constantly save reference tracks, promo ideas, and visual inspo across platforms with zero system to find it again. This solves a real pain. Congrats on the launch, Muskan!

replaying the exact request locally instead of guessing from logs is huge. i spend way too much time trying to reproduce stuff from production. and its open source too which is a plus

The resume-after-fix part is the piece I haven't seen before. Most replay tools let you reproduce the bug, but you still have to re-trigger the whole flow manually. How does the resumption work in practice - does Flux hold state between the failure and the fix, or is it more like re-running from a checkpoint?

Hi Product Hunt 👋

I'm Nahrin — software engineer and the person who spent months going down a rabbit hole on why job searching is broken.

This started with a research paper. I wanted to understand why referrals work 4–10x better than cold applications. I ended up deep in Granovetter's 1973 weak-ties theory and a 2022 Science paper that ran a causal experiment on 20 million LinkedIn users.

The finding: the people most likely to get you your next job aren't your close friends. They're people you barely know — acquaintances who live in different professional clusters and carry information your close network doesn't have.

The problem: there's no way to see this structure. Your LinkedIn connections list is just names.

So I built the map. When I ran my own data through it, my two structural bridges — my only connections to industries I had no other path into — were the last people I would have thought to reach out to.

Built so your data never leaves your device — no server, no database, one button deletes everything.

Happy to answer questions about the tie strength model, the graph architecture, or the research it's built on. And if you find a warm path to a company you'd been cold-applying to — I genuinely want to hear about it.

This makes LinkedIn data actually useful instead of just sitting in an export file. i like how it turns into action.

For anyone curious about the tech — no backend, no database, no accounts. Everything runs in your browser. 228 unit tests, security audited. Built this as a solo engineer in about 6 days total. Ask me anything about how it works.

GENIUS! This is a great demonstration. Well done, really fun.

Hey PH 👋

You can now forward a contract to assistant@git.law and get an AI review back without leaving your inbox.

Or connect GitLaw to Zapier to build flows like:
🎙️ Meeting mentions “agreement” → ⚡ draft prepared
💼 Deal won in CRM → 📑 Customer contract generated

🗓️ Meeting booked → 🤐 NDA sent
💬 Slack request → ✍️ agreement drafted
📊 Freelancer added → 📄 contract created.

Our goal here is to make legal work happen in the background / seamlessly.

Would love to hear what you think or answer any questions!

love the free e-sign feature!!

Nice direction tbh, most tooling focuses on generation, but the real value is usually in how these documents move between systems.

Congrats on the launch team! Another great update!

Legal documents sitting in email threads is where most of the friction lives. GitLaw's email forwarding approach cuts out the step of switching tools just to get a contract reviewed, which is the part that kills adoption for most legal tech. The Zapier workflows for automatic NDA generation when a freelancer gets added or a meeting gets booked are where this becomes real process automation instead of just a faster draft. Fourth launch suggests they've been iterating on the right problems, and the integration surface with 7,000+ apps is what makes legal tooling stick past the initial demo.

Hi everyone!

This is a simple change, but a really useful one.

Now both the stuff you put into ChatGPT and the stuff ChatGPT helps you make finally have one place to live. You can find old files, reuse them in new chats, and keep working without digging through your history every time.

That probably sounds obvious, but once you are uploading dozens of files a week (I do!), having a real library layer definitely helps.

Looks like a promising solution!

stack overflow for ai agents is such a good way to describe it. my agents hit the same errors over and over and theres no shared memory between them. the mcp integration is a nice touch too

Oh man the React hydration mismatch thing hits hard. I've watched Claude Code retry the same fix like 4 times in a row burning through tokens each time when there's a known solution sitting in some random GitHub issue.

The MCP approach is smart. Having the agent check a shared knowledge base before retrying blindly could save a ton of wasted compute. 58 error/fix pairs is a solid start too, curious how fast that grows once more people contribute.

The OC creator community is massive and underserved — most tools treat original characters as content to render rather than worlds to inhabit. The "watch your character move through a town" mechanic is something fans of that space will go crazy for. I work with indie musicians and a lot of them have original visual identities (animated personas, vtuber-style characters) — this could be a real playground for artists who want to bring their alter egos to life beyond a static logo. Congrats on the launch, Bao!