PH热榜 | 2026-03-19

I had hunted Stitch by Google almost a month ago. The Stitch team is back with some major updates to Stitch, thereby making it your AI-native vibe design partner! :)

Here is a quick walkthrough of everything new in Stitch:

🎨 The AI-native canvas can hold and reason across images, code, and text simultaneously. The new agent manager helps you design in parallel. (PS … light mode!)

🧠 A smarter design agent now understands your entire canvas context. You can swap images, generate product briefs, or mix mobile and desktop screens on the same canvas.

🎙️ You can vibe design with your voice (in Preview). Stitch can ‘see’ your canvas and your selected screens. You can ask for design critiques, variations, or navigate your canvas.

⚡️ Instant prototypes. Just hit the play button to see a prototype or preview your app in seconds. Stitch can imagine the next screen based on your mouse click.

📐 DESIGN .md and consistency. Every new design automatically starts with a cohesive design system which vastly improves consistency. The new DESIGN .md file can be used to export or import your design rules.

Read more about the updates here. Stitch is perfect for designers exploring variations or founders shaping new products. If you’re into the future of AI + design, this is worth checking out!

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

Curious how Stitch handles the gap between generated UI and production code. With most vibe design tools I"ve tried, the output looks great in isolation but falls apart once you drop it into an existing design system - wrong tokens, hardcoded values, etc. Does Stitch have any awareness of existing component libraries or is it starting fresh each time? Also wondering about responsive behavior - generated from a single viewport or does it reason about breakpoints at generation time?

Looks pretty cool. Tho today I read that Apple started rejecting solutions that are vibecoded. 🫣

I’ve tried a few similar tools, and they look great at first but get messy when you try to integrate them into an existing system. Curious how well this handles that in practice.

I am often trying to marketing assets with product UI. Is this something that can help me with that?

Tested it without giving any UI hints, just described the core functionality, and Stitch inferred a layout I would have probably landed on myself after a few iterations. Impressive how it picks up context implicitly.

Curious: how does it handle design consistency when you iterate heavily and go back and forth with prompts? Does DESIGN.md help keep things stable or does it drift?

The token mismatch Mykola called out is the production blocker. Every vibe design tool I've tried generates clean code that ignores your design tokens and component API. If Stitch can ingest a token file and respect those constraints at generation time, that closes the prototype-to-production gap.

Thanks! I really enjoyed using Stitch — it helped me improve what I already felt was a promising UI.

That said, one thing I (and other iOS engineers I’ve spoken to) found frustrating is the lack of visibility into what Stitch is doing. For example, when I point out issues or explain what I don’t like in a design, Stitch starts making changes without confirming whether it actually understood my concerns. It would be really valuable to have more communication and feedback — especially asking clarifying questions before jumping straight into a solution.

Another issue is that it sometimes seems to hang indefinitely. The only way to recover is to refresh the page or rerun the prompt, but there’s no indication of whether it’s still working or stuck. Some kind of status feedback would make a big difference here.

Lastly, I often find it confusing to choose between Gemini 3.1 (Pro) and NanoBanana. When I’m making small refinements to an existing UI, it’s not clear which option is more appropriate. It can feel like Gemini 3.1 would give better results, but at the same time, NanoBanana seems more suited to iterative tweaks — making the choice unclear.

Congratulations!

Sounds like a lot of "inspirations" from https://www.wenderapp.com/ 🫣

This is crazy. I tried it and got amazing results.

Great Product honestly

MiniMax M2.7 is an AI agent model pushing toward self-evolving systems, not just assisting work, but actively improving how it works.

Current AI still needs heavy human orchestration across research, engineering, and workflows. M2.7 builds and optimizes its own agent harness, using memory, self-feedback, and iterative loops to improve performance over time.

What’s different is the self-evolution loop — it can analyze failures, modify its own setup, and re-run experiments autonomously. That’s a big shift from static models.

Key features:

• Agent Teams for multi-agent collaboration

• Complex skill execution with high adherence

• Strong performance across software engineering + office workflows

• End-to-end project delivery + real-world debugging

Benefits: Faster experimentation, reduced manual effort, and AI that acts more like a junior researcher/operator than just a tool.

Great for developers, researchers, and teams building AI-native workflows or automating complex tasks.

How far do you think self-evolving agents can go before humans are only setting goals and everything else runs autonomously?

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

until they do the wrong thing at scale.

How do you control that?

This direction feels inevitable.
Once agents start improving their own workflows, it stops being just a tool and becomes more like a system you’re managing.

The part I keep thinking about is control.
If the system keeps evolving its own setup, how do you keep things predictable in production?

Especially for real workflows, stability often matters more than raw capability.

Self-evolving AI is the right direction for any prediction system where the underlying distribution changes continuously. Our football analytics model faces exactly this — features that predicted match outcomes well last season (possession stats, pressing intensity) need reweighting as teams adapt tactically. A static model doesn't flag when its feature importance has drifted, so you only discover the problem in retrospect.

The 'analyze failures, modify setup, re-run' loop you describe is essentially formalizing what good data scientists do manually between seasons. The self-feedback mechanism is what's interesting — the system needs to know not just that it failed, but why it failed in a way that suggests a structural fix vs a data quality issue.

The hard tradeoff in real-time prediction contexts: how does M2.7 balance exploration (trying new configurations) vs exploitation (keeping outputs stable while a process is live)? In a sports context, you can't be A/B testing model architectures mid-match. Curious if the self-evolution loop has a 'freeze' mode for production stability.

The long-term memory feature is what makes this interesting to me. Most AI agents today are essentially stateless – you start fresh every session and lose all the context you've built up. An agent that actually remembers your preferences and past tasks over weeks could be a real productivity unlock.

How does the memory work in practice? Is there a way to review or edit what the agent has stored about you, or is it a black box? Being able to curate that memory layer would make a big difference for trust, especially when connecting it to workplace tools.

@MiniMax is cooking. They launched M2.5 last month, with SOTA performance at coding (SWE-Bench Verified 80.2%), and they're pushing it forward (again) with M2.7, with an 88% win-rate vs M2.5.

Mind-blowing.

Oh and pro tip: you can give it a spin for free in @Kilo Code and @KiloClaw ✌️

One concern is predictability. If the system keeps evolving its own setup, it might become harder to track why certain decisions are being made.

Big believer in this direction, the real shift is not better models, it’s systems that can execute and improve over time.

What’s interesting here is the move toward agent teams and real task execution. That’s where most things still break today.

Curious how you’re thinking about consistency, memory, and control as the system evolves, especially for enterprise use.

If this works as described, this is not just another AI product, it’s a step toward autonomous execution. @mazula95 @rohanrecommends

I've been using MiniMax 2.5 in my product and the bar is really high already - can't wait to try 2.7

Congratulations on the launch 🎉, we've seen great results with 2.5, and we added 2.7 to agenhq.com already 🚀

Ok this is awesome...just gave it a try on a specific use case and totally worked as intended. Well done. I'm going to keep using this.

"Self-evolving" is doing a lot of work in the tagline - curious what that actually means in practice. Is M2.7 updating weights from deployment feedback, or is it more like improved fine-tuning pipelines between releases? The autonomous agents use case is where I keep hitting model limitations - mostly around tool use consistency across long sessions. Does this address that specifically or is it more general capability improvement?

Hey Product Hunt! 👋

I'm KP from the team at Netlify.

We kept seeing the same gap: AI tools are incredible at generating code, but there's still a wall between "here's your code" and "here's your live product."

You generate something, then you're on your own to host it, wire up auth, configure builds, and figure out deployment. That's where momentum dies.

So we built netlify.new.

Go there, describe what you want, pick your AI agent (Claude, Gemini, or Codex), and you get a working app deployed to a live URL — not a preview, not a sandbox, a real production site on Netlify.

The part we're most excited about is what happens next.

You don't start over when you want to change something. You run the agent again and iterate on the same app.

Add a contact form, swap the layout, wire up authentication — it all happens in place. No regenerating, no migrating, no re-platforming.

We designed this for the prompt-first builders who want to ship real things fast — but also for the developers who want to invite their designer or PM to spin up an internal tool without handing them a CLI.

Try it at netlify.new and let us know what you build.

We're especially curious: what's the first thing you'd prompt?

How many AI-generated apps have you had to rebuild once they needed real infrastructure? Netlify now lets you start from a prompt and deploy directly to production - so the thing you prototype is the thing you scale.

The curated prompt templates for common Netlify workflows - like accessibility audits and A/B testing with Edge Functions - are super useful. Pick a template, hand it to your agent in Netlify, and keep iterating. Couldn't be more excited about this launch! 🎉

What I really liked is that you can run the agent again on the same app. Most tools make you start from scratch every time. You let people keep building. That feels real.

Anyway, just wanted to say this is cool. Hope it goes well, @thisiskp_!

Finally! Watching the team work on it and using the early versions I couldn't wait anymore for the final one. Such an awesome addition to the platform

This idea of going straight from prompt to a live URL is interesting to me. I usually spend more time setting up than building, so I can see myself using this just to test ideas quickly without overthinking the setup part.

The prompt-to-deploy flow is something I keep wanting to work seamlessly. Been building a few small tools recently where the bottleneck is actually the setup - getting env vars, build config, and domain wired up correctly takes longer than writing the code. Does netlify.new handle that config layer from the prompt, or is the prompt mostly for the app code and you still do infra setup manually after?

Netlify has always been the place you bring your project after you've built it and you were ready to host it on the best infra with the best DX. Now you can also create new projects on Netlify and it's already set up on the best infra, ready for iteration. Prototypes, internal tools, and iterating on real world sites, this is amazing!!

Also don't forget you can have your full team on Netlify with roles that allow granular access to only the areas that your team agrees on. True full team empowerment and governance.

I like the simplicity here. I can imagine opening it, typing an idea, and just seeing something real instead of a mock. That alone makes it feel more usable compared to tools that stop at code generation.

Is the goal to start with Netlify and then export to the IDE for further complex work or is the goal that I should be able to stay within Netlify throughout my whole development process?

Can you bring your own context or instructions to the agent, or does it start fresh each time? For anything beyond a landing page the agent needs project context to make good decisions, and that's where most of these tools break down. Building Ritemark app - context and instructions are piling up as fast as code.

Hey Product Hunt 👋

I'm Naor, co-founder and CEO of InfrOS.

The standard cloud workflow is broken — and everyone's normalized it:

Deploy → watch things break → optimize reactively. Repeat forever.

We flipped it.

InfrOS takes your requirements upfront — business, technical, compliance — and proactively designs the right architecture before anything gets built. We then emulate it in a real environment to validate performance before a single resource is provisioned. What you deploy is already optimized. Real optimized.

And when your codebase, requirements, cloud provider offering, environment or price changes? InfrOS reoptimizes at the design level. Not a patch. A controlled redesign.

We call it shift-left optimization. We don't predict how your cloud will perform — we prove it.

Early customers are seeing 43% infrastructure cost reductions and 63% faster deployments. We just closed our first Fortune 500 deals through the Ignite DeepTech accelerator, and we're opening InfrOS to the broader community today.

🎁 Product Hunt exclusive: Use code PHLAUNCH for 20% off our paid plans— valid 7 days.

Two things I'd love your feedback on:

Does the shift-left framing resonate with how your team thinks about infrastructure?

What's the hardest part of your current architecture workflow?

We'll be here all day — ask us anything.

🙏

— Naor

Hey Product Hunt:)
A lot of thought went into making this practical, not just theoretical. Really excited to launch it!!

This is a strong take on a problem many teams have simply accepted as normal. The idea of validating and optimizing infrastructure before deployment feels far more practical than constantly reacting after things break. I am especially curious about the emulation layer you mentioned. How closely does it match real-world production behavior across different cloud environments?

really interesting angle on cloud optimization this feels like solving the problem at the right stage instead of waiting for infra issues to show up later. the “prove it before deploy” part especially stood out. how are you handling cases where requirements change very frequently across teams?

Really excited to finally share this and grateful to everyone taking a look today! I truly believe this product will reshape the way teams design their infrastructure. We’re here all day, come say hi!

Thrilled to be launching InfrOS today.
We built it around a belief that feels obvious in hindsight: infrastructure decisions should be validated before anything gets provisioned, not after issues show up in production. That shift from reactive fixes to proactive design is what excited us most, and we’re happy to finally share it.

The idea of proving architectures through emulation before deployment is really compelling. How detailed is the emulation — does it simulate real traffic patterns and failure scenarios?

Really exciting day for us!
This has been a long journey from idea to something people can actually use. We built this to solve problems we kept running into ourselves, so it means a lot to finally share it here and see how others experience it.

Good Morning America!
Let's see how good Product Hunt's autoscalers are - we know ours are optimal :)
Looking forward to seeing everyone on our website and giving our platform a try.
It's FREE, and you have 20% off the first month

Excited to launch InfrOS today. We built it to help teams make better infrastructure decisions before anything is deployed, not after problems appear in production. We believe planning and validation should come first, and we’re proud to share a product built around that idea.

Congrats on the launch! I like the concept of optimizing your cloud infra. Was this designed more as a configuration optimization or migration/setup of your entire cloud infra? I'm also interested in how the "emulation" works since this is a very proactive approach. Without deploying and observing traffic patterns, how are you are able to determine the optimal cloud configs?

Using it now to define my HIPAA complaint infrastructure! Great step by step discovery of my needs and easy generation of useable code! Very thoughtful and thorough!

This feels extremely relevant for solo devs or small teams that are price sensitive and are actively thinking of ways to keep costs down. Are there ways to be able to isolate for what's driving cloud cost after getting set up?

I’ve seen how much time goes into fixing infrastructure issues after deployment. Catching those decisions earlier could save a lot of effort.

How accurate does the emulation get compared to real production behavior, especially under unpredictable traffic patterns?

To answer your question, the shift-left framing resonates immediately, especially for anyone who's lived through a painful post-launch infrastructure fire. For most teams I've spoken to, the most challenging aspect of the current workflow is the discrepancy between the design and the actual deployment. Congrats on the launch!

Hey Product Hunt 👋

We built OctoClaw because we kept running into the same problem: there were too many repetitive tasks across marketing, sales, and support, but hiring for every role early was unrealistic.

Generic AI tools helped a bit, but they still required constant prompting, supervision, and context every single time.

So we asked ourselves: what if AI felt less like a chatbot, and more like hiring a specialist?

That became OctoClaw - a system where you hire domain-trained AI specialists, each focused on one area of work.

Today that means specialists for:
• marketing
• sales
• support

The goal is simple: give founders and lean teams leverage without adding headcount too early.

We’re launching early because we want honest feedback from people who actually build and ship.

Very curious:
Which specialist would you hire first?

By the makers behind award-winning @Octomind (#3 Product of the Day in Feb 2024, #1 in Oct 2024), @OctoClaw helps you get AI specialists in marketing, sales, and support, that run 24/7 in the cloud.

No coding. No terminal. No Docker. Just tell them what to do and they get the job done.

Give it a spin: octoclaw.ai

Interesting concept! Having AI agents connected to real workflows instead of just chatbots is a solid approach. How does the handoff work when an agent can't handle a task — does it escalate to a human?

Very cool! congrats on the launch!

This is a smart and relatable way to position AI for growing teams. Framing it as hiring a specialist rather than using a generic assistant makes the value much easier to understand. I can see this being especially useful for founders trying to scale without expanding headcount too quickly. Which specialist are users finding most valuable so far?

Love the concept @daniel_rodler @marc_mengler 👏

Quick question on pricing: it looks like fixed per agent/month, but for heavier agents (e.g. PR review with high token usage), how does pricing adapt?

Also, big congrats on the launch 🚀

Curious to hear and witness some success stories by other hunters who have used Octoclaw.

Congrats on the launch! Looking forward to what's possible with it.

Hi @daniel_rodler , congrats on getting OctoClaw out there!

Well, this line got me: hire specialists, not chatbots. I visited the homepage and like its dashboard.

Also gotta say, the persona names made me smile. Patience Worth for support. Sterling Hype for marketing. Small thing but it shows you guys actually thought about the human side.

Anyway, excited to see where this goes. Hope it takes off.

love this positioning making ai feel like a specialist instead of just another chatbot is a much clearer way to think about it. this feels especially useful for small teams trying to do more without hiring too early. which part is the most mature right now marketing, sales, or support?

Looks pretty interesting. I keep wondering about specific agents vs. the general agents. I didn't try it as trial requires a credit card. Might consider a no credit card trial at this early stage just as some feedback. Congrats on the launch!

Are there ways to set guardrails for each specialist? For instance, when it comes to marketing, are there ways to focus in on specific topics, avoid certain words, or adopt a certain writing style?

Watched the demo and the multi-agent coordination looks genuinely promising. One thing that gave me pause though: the LinkedIn auto-posting flow. Is there a human approval step before anything goes live, or does the agent publish directly? For content that touches a personal or company brand, I'd want at least a one-click confirmation before it goes out.

Would love to see a "supervised mode" as an option for those of us who aren't ready to go fully autonomous yet.

The framing of "specialists that execute" rather than "automation" is doing a lot of work here, and it's the right call. The mental model of hiring rather than configuring lowers the activation energy significantly. Curious how you handle cases where the AI specialist makes a wrong call. Is there a review layer before it takes action, or does it execute autonomously by default?

Thrilled to hunt Comet for iOS... the first agentic AI browser built for mobile!

Comet is an AI-native browser from Perplexity that brings a full agentic assistant into your pocket. It doesn’t just answer questions, it takes actions across your tabs, apps and searches while keeping you fully in control.

Comet for iOS turns your browser into an actionable workspace.

Key features → benefits:

• Chat with your tabs → No more jumping screens; get answers from everything you already have open.

• Cross-tab summaries → Synthesized insights instantly, not page-by-page.

• Actionable AI assistant → Research, plan, write, shop and follow up hands-free.

• Ad blocker → Cleaner, faster browsing.

• Voice control → Truly hands-free mobile research.

• Agentic workflows → Ask it to handle tasks as you would on Comet for desktop. See exactly what actions the assistant is taking while you remain in full control.

Download on Apple App Store and experience agentic browsing on mobile: https://apps.apple.com/us/app/comet-ai-personal-assistant/id6748622471

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

It's ABSURD that this isn't in first place on PH for the day, lol. Comet is one of those "how did I live without this?" tools, in so many ways, especially now that Perplexity Computer is off the ground. The fact the team has been able to refactor the Comet experience to a mobile format is nothing short of amazing.

Great work, team! I've spent so much time training up my Perplexity agent with personality paramaters, skills, a portable identity manifest, deep connections to all of my tools and messaging, and so much more. I don't know what I'd do without her, honestly!

finally. excited to use!

Tried a few AI browser tools before, but they always felt a bit clunky , This looks more polished and actually useful for everyday workflows.

Comet’s AI browser could totally transform the way content creators research and storyboard ideas. I’d love to explore how short, AI-powered videos could showcase these workflows in action.

the evernote migration being painless is huge honestly. that's the biggest barrier for most people switching note apps, you've got years of stuff in there and nobody wants to spend a weekend cleaning up broken formatting. the no device limits thing is also a great move, locking free users to one device felt like such a cash grab by evernote. how does the rich editor handle things like tables and code blocks?

I’ve stayed with older tools longer than I should just because switching felt painful. A clean Evernote import alone removes a huge barrier.

Wow, and you import Evernote attachments too. I don't think anyone else does that. Finally, something that I can actually switch to. I attach PDFs, sometimes audio files to my Evernote, and that's what's kept me stuck in it forever. Will you be launching desktop/iOS apps?

As a former big fan of Evernote, Cimanote is a welcome sigh of relief. It has a great UX and just works. It feels like it picks up the torch of being my extended brain. I love using it now and am excited to see how it grows.

I've trialled a number of notes apps over the years and was a long time Evernote user but eventually went back to my default mac app and Google Keep.

Been using Cimanote and I've been loving the UX. It's simple, clean and efficient.

Dictation would be a key feature for me.

What you've built so far is fantastic and I'm looking forward to how the app evolves!

There's something quietly powerful about "get out of your way." The best tools don't demand attention — they just hold the space. The frustration with Evernote wasn't just price, it was the grief of watching something you trusted turn into something that needed managing. What are the two or three things you're absolutely refusing to compromise on as you grow this?

This one hit me. I was an Evernote user too. Felt that same frustration when they jacked up prices and locked everything down. You built the thing I just complained about.

As a homepage positioning expert, I spent some time on the site. The import feature got me. Most tools make you start from scratch. You let people bring everything with them. This strategy makes someone trust you right away.

Anyway, just wanted to say I'm rooting for you, @blagoja. Hope this takes off.

Congrats @blagoja. Excited to give this a shot. Similar to others I have abandoned Evernote as it wasn't living up to my expectations and reverted back to my old Notes app on my macbook. With these big name products you always find lack of support, lack of new feature etc. I have done a similar thing with a fitness/food tracker app, just tired of all the ones that showed promise but never lived up to expectations.

I used the early beta and found it useful. It was so easy to share my note with others.

Interesting space — how do you see Cimanote positioned vs Notion (flexibility) and Apple Notes (simplicity)? Where do you want to win long-term?

OMG finally, someone fixed all the reasons why I ditched evernote.
This looks so good.
Does it work on Android, IOS, Windows and MacOS?

It’s really refreshing to see user experience as a top priority for Cimanote.

Evernote lost its focus years ago when it decided to optimize for value extraction instead of utility.

I’ve been using (and enjoying) early versions of this app. I’m excited to see what comes next. Keep up the great work!

I wish this had launched four years ago, when I decided to abandon Evernote. It will be interesting to fire up Evernote again and see how the export goes!

Simple and effective. The essential element done well at a fair price.

Any plans to support dictation / transcription integration?

Really appreciate the honesty and transparency here, @blagoja Blagoja. It’s rare to see a founder lay out the pain points that motivated the build so clearly. The Evernote import and device-unlimited approach are subtle but huge quality-of-life wins that often get overlooked in note taking apps.

Curious how did you balance speed and feature completeness during the soft launch? So often a fast, clean experience gets compromised by trying to do too much too soon.

Evernote lost the plot years ago. Curious how you handle sync conflicts when editing the same note on two devices — that's where most note apps fall apart.

Built PixelClaw because I got bored waiting for Claude Code.

Modeled after the Claude Code mascot, it’s a tiny pixel crab that lives on your Dock, sleeps, wakes on click, and chases apples you drop.

Started as a dumb side project between prompts, and turned into a surprisingly fun little companion.

Would love to hear what features or behaviors you’d want in a tiny Dock companion.

This is so cool!! Make it pettable

Waiting for builds or AI responses can feel surprisingly long. Having something small and interactive like this actually makes that timer nice.

A new friend for Dinoki!

ok this is adorable. would be cool if you could feed it and watch it grow like a tamagotchi 🦀

I love little computer people, thank you

This is awesome, a full little background minigame, might have to adjust my screens to make room for this little guy between runs

This brings back memories! The first time I saw a Mac as a kid, it had those eyes in the menubar that followed the mouse pointer.

Hey Product Hunt! I'm Alisa, founder of Lucent (backed by YC and Ryan Hoover :)

We built Lucent because we kept seeing the same problem at every company we worked at - bugs and UX issues slipping into production and nobody catching them until users complained. The data to find these issues was already there in session replays, but nobody has time to actually watch thousands of sessions.

So we built an AI that does it for you. Lucent watches your session replays automatically and surfaces bugs and UX issues in real time, with full reproduction context sent straight to Slack and Linear.

How it works:

1. Connect your PostHog account

2. AI watches session replays 24/7

3. Bugs get sent to your Slack or Linear as users encounter them

To get started, you can get 400 sessions processed for free and see what issues your users are coming across - most teams find issues in the first hour they didn't know existed. Sign up here: lucenthq.com

Trusted by teams obsessed with product quality, like Reducto, Browser Use, Productlane, Julius, Finta, Happenstance, and Evidence.

Would love for you to try it out and let us know what you think - happy to answer any questions!

Find us on socials:
- https://x.com/lucent_ai
- https://www.linkedin.com/company/lucenthq

Good job team. The pain of watching dozens of replays is real. If AI can do it and summarise the findings, that's huge. Do you work with tools like clarity etc. or you have own session replay tool?

I wish you good luck today!

Does Lucent only detect bugs or also UX issues

Early user here, Lucent is so powerful and simple to use. 10/10 would recommend!

All I have to say is unbelievable. This will be so useful for people to find bugs and inefficiencies inside their product. I am one-shotted

Congrats on the launch, @alisarae. I spent almost five minutes on the site. The quote that hit me hardest was from Justin Lee: "We installed Lucent on a Friday, by Monday morning it had filed 7 bugs we'd never seen."

That one line does more selling than any feature list. It captures the speed and the value.

One small observation while poking around. That quote is mixed in with a wall of other testimonials. It deserves its own space, maybe right under the hero. Let it breathe.

Anyway, it was just something I noticed. Hope the launch goes well.

Super interesting—does Lucent only detect technical bugs, or can it also flag UX friction (like rage clicks, dead ends, confusing flows)?
The latter would be insanely powerful for product teams.

The session replay angle for issue detection is interesting - we"ve had Datadog and LogRocket running for a while and the gap I keep running into is that replays tell you what happened but not why it matters. Is Lucent doing its own severity scoring, or is it more like "here are the unusual patterns, you decide"? Wondering how it handles the noise problem - low-signal rage clicks vs. something that actually broke the flow.

Looks amazing! Have always hated hunting through PostHog for bugs, absolute game changer!

great product that keeps us from shipping bugs

Congrats on the launch!!

Can it also go through logs and detect potential issues for apps where interactivity is not necessarily in the UI on a web based dashboard?

I’ve set up alerts before and eventually stopped checking them entirely. If this actually reduces noise, it solves a real problem.

Always-on web monitoring is something I keep wanting but the noise problem kills every tool I try. Most of them alert on everything and you tune it out within a week. Curious how Scouts handles the signal/noise tradeoff - is it user-defined triggers, or does it learn what you actually care about over time? The iOS angle makes sense for this use case, monitoring notifications feel more natural on mobile than a dashboard you have to remember to check.

I have Google Alerts running for a handful of competitor terms and I stopped checking them months ago. If Scouts pushes only when something matters instead of every tangential mention, the iOS app turns passive web monitoring into something you'd rely on daily.

Power user here- congrats!

If Yutori could connect with my email, calendar and task apps that'd be amazing- any plans for that yet?

Congrats on launch! I have been using Yutori recently. Now, a dedicated app on phone would be cool with alerts.

Filtering what actually matters is the hard part.

this is one of the best products i have ever seen today

Hey all, @abhshkdz here from Yutori! 👋

Super excited to share that Scouts is now on iOS! Download it here.

Scouts is the most powerful way to research and monitor anything that matters to you — a flight deal, whether your brand is being talked about, a job listing you've been waiting on, a sold-out item back in stock, news from around the world, how a market is moving, etc.

The kind of things you'd otherwise have to check manually, set up clunky alerts for, or just... miss.

Scouts handles it all. Quietly in the background. So you're always up to date.

A phone app has been one of the top requests since we launched Scouts. Now it's here.

Monitor the web for anything, and get push notifications when something changes.

Would love to hear how you use it! Please do reply here and let us know.

This is the way. When the job can be done on local hardware you already own, it feels wasteful to rent offsite tokens.

I've used it for a few client calls and standups: transcription accuracy is solid (better than expected for local), custom LLM prompts let you tweak summaries exactly how you want, and the Obsidian export plus webhooks feel genuinely useful for power users.

This is a very thoughtful take on AI meeting notes. The fact that everything stays on the Mac makes the product immediately stand out, especially for users who care about privacy and control. I also like that you are positioning it as something that can work alongside existing tools instead of forcing people to switch. Do you see talat becoming more transcript-first over time, or is improving summary quality still a major focus?

The privacy angle here is underrated. Most notetakers treat "your data" like a byproduct. You're treating it like it belongs to you — because it does. What I'm curious about: do you think local-first transcription changes how people actually speak in meetings? Like, does knowing nothing leaves your machine shift the quality of what gets said?

really like the privacy-first direction here. having realtime meeting notes that stay fully on-device feels like a big win, especially for people who are not comfortable sending sensitive conversations to external servers. the obsidian export and custom prompts are a nice touch too. how has the response been so far from people already using granola or similar tools?

I run meetings in two languages — some fully in Czech, some in English. Does the transcription handle both well, or is it optimized mainly for English?

Does it support English only?

Excited you guys are going to get this 🔥

Is this open source?

“AI agents that run operations” sounds great.
In reality, most workflows break on edge cases.

Similar to how there could be internal tool sprawl, could there be an issue with AI agent sprawl? For instance, AI agents spun up that do very similar tasks being used within the same team when there should really only be one?

Open source AI agents for ops is an interesting wedge - the self-hosted angle matters a lot for anything touching customer data or internal workflows. Most teams I know are cautious about letting a SaaS agent touch their CRM or support queue. How does Budibase handle the agent permissions model - is it role-based on what the agent can read/write, or more of a "here is the whole app, go" setup?

Congrats on the launch! 🎉 The safety-first angle on autonomous agents is the right call — I’ve seen several threads here on Product Hunt lately where AI safety keeps coming up as the #1 concern for builders. Most teams are still cobbling guardrails together from scratch, so having this baked into the stack is a real unlock. Curious how NemoClaw handles edge cases where agent decisions conflict with safety boundaries — graceful degradation or developer-facing alerts? Building OceanMind, an AI-powered breathwork app, and exploring this kind of infrastructure for personalized agent flows.

Agent safety is one of those things that sounds obvious until you try to implement it without crippling the agent usefulness. I keep running into the same tension when building with AI: the guardrails that prevent bad outputs also kill the edge cases that make agents actually useful. Does NemoClaw approach this as policy rules (static constraints) or is there something more dynamic going on - like runtime monitoring that can distinguish "risky but intended" from "risky and wrong"?

• The always on UGC is super interesting . Brands could turn their audience into creators instead of just consumers .

Hey, congrats on the launch. Is there meaningful utility for Doodle holders?

I remember that this started as NFTs, then formed into a Media, now is it AI?

I’d love to see how it manages my crazy week. I’m always juggling different time zones and back-to-back meetings. If the AI can draft responses that sound like me and actually get people to agree on times, that’s a huge win.

I think I could finally stop stressing about double-booking. The idea of an AI reading my emails and figuring out the best time for meetings is so appealing. I want something that feels like it’s really looking out for me, not just blindly scheduling.

I juggle two calendars — personal and work — and the overlap is where scheduling gets messy. Does it check both when suggesting times, or do I need to pick one?

Hi everyone!

For decades, the HTTP 402 Payment Required status code has essentially been a web developer meme. Today, @Stripe and Tempo actually made it the foundation of the agent economy.

If you're building autonomous agents, you already know the biggest roadblock isn't reasoning but purchasing. Agents get stuck on human-optimized checkout forms, 2FA, and visual captchas. MPP solves this at the protocol level.

When an agent hits a gated API or requests a resource, the server kicks back a 402 with payment details. The agent fulfills it programmatically (either via on-chain stablecoins or fiat Shared Payment Tokens), retries the request with a credential in the header, and gets the data.

Because it's Stripe, developers can monetize an API per-call or let an agent order a physical sandwich, and the funds just settle into their existing Stripe fiat balance.

We are officially moving from humans clicking "Buy Now" to agents negotiating microtransactions in milliseconds.

Following with interest. So far it is working well and communication from the devs is timely and useful. It also shifts the tech focus from US centrism - which seems to be the flavour of the day lately, all for it.

Have been using it since the web app. Glad to see the app release. The Spaces feature is definitely cool and unique. The customisation helps to make it feel my own, esp for my phone.

The framing matters here. "Private by default" isn't a feature — it's a value statement. Most AI tools make privacy a premium. You've made it the foundation. The planet angle is quietly bold too. Curious how users respond to that — do people actually feel the difference, or does it stay invisible until the moment they need it most?

Congratulations on the launch, going to be keeping an eye on this. It's great to see a UK business taking on this and giving options in the market.

Is it just a Chat interface or do you have platform API?

Definitely going to have to play about with this when I get a moment, data like this is really valuable for multiple reasons with the work I do - if it means I don't have to fight with Crunchbase I will definitely be an avid user!

Love the 'pay as you go' credit model. It's a game-changer for indie hackers who don't have venture backing yet. Definitely going to test this out for my next project. Great work, team

Vu, a 50MB native app built with Swift + Metal is music to my ears. 🎶 As a solo builder developing a 'Local-first' studio for writers, I’m obsessed with keeping things lightweight and efficient—'no fans spinning up' is the ultimate flex. How does the Metal rendering pipeline handle recording complex, interactive UI graphs without dropping frames?

Congrats on the launch, Vu! 🎉 This is exactly the tool I’ve been looking for — building OceanMind, a premium iOS breathwork app, and the demo video phase is always the most painful part. The 3D device frames with iPhone/iPad USB recording is a killer combo. Loved that it’s native Swift + Metal too, that matters. The AI editing roadmap for Q2 is what really caught my eye — filler word removal and edit-by-transcript would be a game changer for solo indie devs doing everything themselves. Rooting for this one!

The insight that people are exhausted from duct-taping tools together is real. But I'm curious about the human side of it — when "Ally" manages everything from WhatsApp, what happens to the intention behind the message? The person sending it still needs to feel heard, not just processed. How do you think about preserving that layer?

Hello @hikevindiaz, happy launch day. Two products at once is impressive.

I like The Puerto Rico Convention Center case study. A government client using RAG to give event planners real info. That's real work.

However, one thing I kept thinking while reading. You have a lot of products. Calendar. Orders. Phone. Forms. Tickets. Workflows. That's a full stack. But on the homepage, it feels like a list. Isn't that so? A person landing there might not know where to start. Just shared what I noticed.

Hope people like it.