PH热榜 | 2026-04-11

After seeing this, it feels like an essential.

Hi everyone!

Claude is definitely entering every interface knowledge workers actually live in.

With Claude for Word, you no longer switch to a separate chat window. You stay inside your document: highlight text, leave a comment, or describe the edit, and Claude revises it as tracked changes while keeping all your formatting and styles intact.

It also shares context with Claude for Excel and PowerPoint, so one conversation can span your whole Office workflow.

P.S. Spoke with a founder building plugins in the M365 ecosystem today. He’s feeling the pressure but is still pretty optimistic—the game is still very much on!

"Finally! As someone who spends 3 hours a week just resolving comment threads and formatting inconsistencies between stakeholders, this is a lifesaver. The fact that it shares context with Excel is huge – I can't count how many times I've had to manually re-explain a sales figure from a spreadsheet in a Word doc. Excited to test if it can handle my messy 200-page spec docs."

Claude Code just added a new slash command,/ultraplan, that moves the planning phase of your coding workflow out of the terminal and into the browser.

The problem with planning in a CLI is that you're stuck in a linear, terminal-bound loop.

You run /plan, read a wall of text, approve or reject the whole thing, and move on.

There's no way to comment on a specific section, annotate what you agree with, or flag what you want changed without rewriting the entire prompt.

/ultraplan fixes this.

You run the command from your local CLI, and Claude generates the implementation plan remotely on Anthropic's cloud infrastructure.

Your terminal stays free while that's happening.

When the plan is ready, you open it in a browser on claude.ai and interact with it like a document: inline comments on specific passages, emoji reactions per section, and an outline sidebar to jump around.

You can ask Claude to revise, iterate, and re-draft until the plan looks right.

Once approved, you choose: execute the plan in the same cloud session and get a PR, or teleport it back to your terminal with full local environment access.

What makes it interesting is the split between where planning happens and where execution happens.

Planning is mostly reading and reasoning, so the cloud is fine.

Execution often needs your local environment. /ultraplan respects that distinction.

Who it's for: engineers and teams already using Claude Code who need more control over the planning phase before code gets written.

Particularly useful for larger refactors, migrations, or anything where approving a plan blindly feels risky.

Worth noting: requires Claude Code v2.1.91 or later, Claude Code on the Web enabled, and a GitHub repo. Not available on Bedrock, Vertex, or Foundry.

Been using Claude Code daily building 9 AI products

solo in 4 months. The planning phase is genuinely

the bottleneck — /ultraplan moving it to browser

is the right call.

The terminal loop (plan → approve → forget context)

kills flow. Visual annotation of specific sections

would be a game changer.

planning in the cloud while keeping execution local is the right call. curious if teams can share the plan for async review before execution - that changes the migration use case.

We are use to terminal. So is there a migration path for existing users?

free, open source, native Swift, OCR included - that's a rare combo. most tools compromise somewhere on the annotation layer.

One question - why we need this if mac already have this feature inbuild?

Based on the screenshots and videos it seems to have a very nice look and feel. Amazing that you provide that as open source and free! Despite the look and feel are there more upsides in comparison to the native mac os screenshot and video capture functionality?

Finally a free alternative to CleanShot X that doesn't feel cheap. The fact that it's built natively in Swift makes a real difference on Apple Silicon.

Building a native Mac screen recorder is no easy feat given all the strict capture permissions Apple requires. I can see myself using this daily to grab quick bug repros for my GitHub repositories. I would love to know if you built this with pure Swift or used a cross-platform framework like Tauri under the hood.

running code review through an AI agent is solid. whether it blocks or just suggests matters a lot - one is a tool, the other is actually an authority in the workflow.

what all platforms are supported for this?

Two likes after posting on Xwitter is painfully specific and that's why your product lands. Most marketing advice for people who build stuff is either too abstract or too startup-bro to actually use. The interactive messaging workshop is the part I'd go to first. Like your product!

"@fmerian is a key hunter and Product Hunt optimization expert."

thanks for the mention, @j1ngg! keep up the great work

the fact that each skill reads your product brief first is a cool approach. most marketing tools just generate generic output because they don't know what you actually built. i'm having trouble with (and learning lots about) distribution at the moment, will give this a shot :) congrats on launching!

The '2 likes' opening hit too close to home, literally went through this yesterday with my first ever X post about my app. The messaging workshop part is what I need most right now, I can build the product but translating it into words that actually land is a completely different skill.

is this framework designed for solo makers with zero marketing budget, or does it assume you have some paid channels to work with?

@farzatv just dropped something very interesting...

Clicky is an AI buddy that lives right next to your cursor on Mac. It sees your screen, hears your question, and walks you through whatever you’re working on like a friend who’s insanely good at everything.

AI lives in chat tabs. You explain context. It guesses. Slow and broken. Clicky works inside your workflow, no context switching. It sees exactly what you see and guides you in real time.

What stands out:

• Screen-aware (on hotkey) 👁️

• Voice-first with spoken responses 🎙️

• Points directly at UI elements 🖱️

• Real-time streaming (AssemblyAI + Claude) ⚡

• Privacy-first (no passive recording) 🔒

• Free & open-source 🆓

Use cases: learning tools like Blender, DaVinci, After Effects, designing in Figma, writing, debugging, even analyzing stock charts. Possibilities are endless.

This isn’t just another AI chat. It’s a new interface for interacting with AI. If you’re a creator, builder, or someone who learns by doing, this is worth checking out.

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

Very interesting idea, is it able to control the mouse? That would elevate the abilities intensively.
Might be more difficult to convince people that it won't do something weird than to actually achieve it ahhaha

does it supports local AI integration?

@farzatv Well done Farza. Clicky feels like a thoughtful idea, and I like how naturally it brings AI into the workflow. The key will be keeping it helpful without adding distraction or friction.

The constant visibility is the double-edged sword here. I'd want a timed context window, not always-on capture.

Before i try it, somehow I have i feeling that it would be too much of a distruction? or im too old 🫣

Great but i want to know one thing how it is different from tool like micro and other tools for ai interview?

Ai interview prep is great, now how does it compare to JacknJill or Juicebox? What are you the best in the world at?

@justonedev Nice work Aleks. Voicr has a clear and useful concept, with a strong focus on speed, convenience, and low friction.

Hey Product Hunters 👋🏼

I'm Jugal, maker of 1% Better app.

This came out of a personal need. I tried 20+ habit tracking apps but eventually gave up on all of them in under a week. Main problem was that they feel like data entry job. So I made a clean minimal non-intrusive version of a habit tracker.

And here's the kicker: What most people don't realise about streaks is that it compounds. And compounding isn't intuitive. If you got just 1% better every day, in a year you'd be 37x better than today.

That's the math: 1.01^365 = 37.78x 🚀

This app helps you visualise the compounding effect of your progress.

Practice your music, go to the gym, have a calorie deficit day or an alcohol free day, you now have a simple way to track whether you were just 1% better today and watch those tiny wins stack into exponential results over time.

In Premium version you can:
✅ Add multiple trackers and answer "Were you 1% better today?" for each habit you want to track
✅ Set custom notifications at pre-defined time of the day for each of your trackers
✅ iCloud backup & sync
✅ Export your data to CSV (in bulk and for individual trackers)

I wanted a simple app that takes <3 seconds a day, is clutter-free and works on the concept of compounding your habits rather than a static streak.

I'm planning to do more with the data-crunching to help you understand your progress better, will be releasing cool new updates soon. Would love to hear your feedback after you've tried it for yourself.

Thank you.

App link: https://apps.apple.com/app/apple-store/id6761054951?pt=128677803&ct=producthunt&mt=8
Website: https://1better.today

Love the concept.

I’ve been using this for over a week and I can say this has been the only one that doesn't overwhelm me.

I track my workouts and meditation with it and the compound math thing actually keeps me coming back.

Most tracking apps I download, use for a week, and forget. This one stuck because the daily check in takes like 3 seconds and watching the curve grow is oddly satisfying.

Someone a James Clear fan huh? Love the build here. The dark + neon UI is refreshing. The compounding line graphs here are a powerful visual! Will be following this one

Simple idea, I like the focus. Are people actually sticking with the yes/no approach, or do they start wanting more detail over time?

Love the app! The UI, the muted pastels, the adjustable fadeout. Chefs kiss.

Looks like a great idea, but the mics on my iPhone don't work properly so I can't test it :(

Hi everyone!

MolmoWeb is one of those rare open releases that really stands out.

It is one of the strongest open web agents out there, able to actually see the screen and act on it. But what is even more impressive is that Ai2 did not stop at open-sourcing the weights — they released the whole stack: training code, eval harness, annotation tooling, synthetic data generation pipeline, client-side demo code, and the full MolmoWebMix dataset.

I honestly do not have much to add beyond respect. Open-sourcing is not an obligation, but doing it at this depth and quality, consistently, is extremely hard. The Ai2 team continues to set a high bar.

This is the kind of foundation the open agent ecosystem actually needs.

screenshots-only is a bold constraint. how does it handle dynamic content that changes between screenshot and click?

Hey! Looks interesting! How much of Buildermark's code is AI-generated? :)

What counts as AI-generated though? Pure agent output is easy to measure. A dev who wrote the spec, accepted 90% of agent suggestions, and debugged the rest - how does your tool categorize that?

Developers and companies are claiming to have X% of their code written by AI, but most are just guessing and not actually measuring. Buildermark matches coding agent diffs with commits to calculate how much is by agents.

It also generates a convenient archive of all conversations from all of your agents, like Claude Code, Codex, Gemini, and Cursor.

Online Demo: Browse all 364 agent conversations that wrote 94% of Buildermark's code

https://demo.buildermark.dev/projects/u020uhEFtuWwPei6z6nbN

Hi again PH,

When we launched Tidy on Product Hunt a month ago, we found people were really excited to use it with their friends in their iMessage group chats! Today, we are launching tidy for group chats, which we've rebuilt to be more fun and helpful. What did we change?

• We added image generation and the ability to nickname Tidy for a chat (which we've found makes Tidy fun to use)

• Tidy only wakes up when mentioned, and stays awake for a short period. It still has access to your messages in between.

• You can set reminders, make reservations, search & autonomously browse the web and more in group chats

• A load of new apps (a quick note - certain apps aren't available in group chats for security reasons; we don't want other people to write emails for you!)

It's built to be shared with friends, so you just need 1 person with a Tidy Group Chats membership ($10/mo) for everyone to be able to use Tidy, unlimited in the group chat.

We'd love you to try it out at withtidy.chat 🫶

Onwards!

Aagam

Hey Product Hunt,

I wanted to share a bit on why I personally love using Tidy in group chats:

• Organizing meetups- as people discuss dinner preferences or meetup times, Tidy finds restaurants or places that make sense for everyone, and can ultimately book something the group.

• Planning trips- Tidy can check regularly for cheap flights, research places, and write a travel planning doc for the whole group.

• Family/group admin- one day my family decided to add Tidy to our group chat, which turned out to be great for organizing all our schedules together and answering questions like "what's the wifi password again?"

If you have something exciting you've done with Tidy (especially in group chats), please feel free to share!

--Brian

Best feature yet!! Super excited to use it in my college group chats

Wow what an awesome feature! Always wondered if something like this existed…

Being able to share Tidy in a groupchat is huge!!

Having Tidy be in my groupchats makes organizing events so much more fun!

Using this to settle up on Venmo after an outing? Yes please.

super fun!! congrats on the launch guys

This is awesome! In addition to the leap from personal to group/organizational assistant, adding Tidy to some group chats has let me see how others use Tidy and exposed me to several new use cases. Huge update, congrats!

I founded Osintir after realizing a hard truth: social media platforms don’t actually protect the content you upload. Once something about you enters the internet, it can be copied, manipulated, and redistributed without your control.

I’ve experienced this firsthand — my content has been altered and misused without consent. And now, with the rapid advancement of deepfake technology and AI-generated media, the scale of this problem has grown exponentially, impacting millions of individuals and brands worldwide.

Osintir was built to change that. It continuously monitors open sources across the web, using multi-layered protection to detect where your images and videos appear. You receive real-time alerts the moment your content is reused, giving you the ability to act immediately — whether that means removing the content, responding to misuse, or escalating the issue to authorities.

Osintir isn’t just a tool — it’s peace of mind. It means knowing you’re no longer powerless in a digital world that moves too fast.

Because your identity belongs to you.

LLMs don't 404 - they drift. are you tracking output quality degradation, or just API availability?

Hello Producthunt Community,

Thanks for checking out our launch!

LLM Ops Toolkit started as an internal dashboard our AI team used to monitor API uptime and cost drift across models. We open‑sourced it because every AI engineer ends up building something like this eventually.

It combines monitoring, cost analysis, routing simulation, and a maturity audit - all running locally in a static page.

Would love your thoughts on what metrics or integrations you’d want next!

🟢 Provider Status Monitor — Aggregate uptime across 18+ AI API providers with 90-day history and live response-time sparklines
💰 Cost Calculator — True cost analysis including hidden engineering overhead and routing savings
🔀 Routing Simulator — Visualise intelligent request distribution across models with cost/quality trade-offs
📊 Diversity Audit — 10-question maturity assessment with radar chart and personalised recommendations
🌙 Dark/Light mode, auto-refresh, browser notifications for outages, and a dynamic favicon that mirrors overall provider health

Filtering DM noise is a real problem — especially on Instagram where 90% of DMs are spam or cold pitches. How are you training the 'importance' signal? Is it based on engagement history, follower count, or something more contextual like message content?

👋 Hey everyone — I’m Sree, a maker who loves building internet products.

InboxJoy is born out of a personal problem, "too many DMs"! It's a simple product to help people cut through noise and focus on the messages that matter.

Before this, I built Superpage, which exited to a YC-backed company. I’m also currently building Replyless.ai, exploring how AI can make email inboxes less chaotic and more useful.

Happy to answer any questions, and I’d really value your feedback.

I like the idea of keeping everything inside Gmail !! How does it deal with messy threads where several people reply at the same time?

This will save us time, but will it make the replies repetitive after some time?

action item extraction is the real value. 'let's revisit this' won't surface as a clean item though - those implicit commitments are the ones that actually get dropped.