Hey Product Hunt! 👋

Brian, DevRel from Kilo here. We built @Kilo Code Code Reviewer to kill PR bottlenecks.

It runs automatically when you open a PR, catching security issues, performance problems, and style inconsistencies before your teammates even look at it, and offers comments and inline suggestions.

It's completely free with models like MiniMax M2.1 and GLM 4.7 - or use the latest from @Claude by Anthropic, @Gemini, or whatever you prefer from over 500 supported models.

What's your biggest code review pain point?

@Kilo Code Code Reviewer has become such a big part of my workflow. Can't imagine coding without it.

As a data engineer, I’ve been using Kilo Code reviewer and it’s surprisingly data aware, not just code aware. It helps me catch pipeline changes that would only break dashboards later

The amount of time this product helped me out is wild. I spin up PRs daily and let code reviewer take care of it before asking a human dev to do a final check.

had a blast working with the team on this new launch, introducing Kilo Code Reviewers -- AI-powered code reviews that understand your codebase and catch bugs before merging.

First launched on @Product Hunt about a year ago, @Kilo Code is now the most popular open-source coding agent, trusted by 1M+ developers.

S/O to @sytses @scobreit and team 👏👏

Reading this made me think about how often my PRs wait longer than expected. I usually spot small issues only after opening them. Getting early feedback would help me feel more confident before asking teammates to review.

Love the open source angle of Kilo Code and the transparent pricing model (you can bring your own AI models too). How it handles larger codebases with multiple modules and complex dependencies compared with other AI coding assistants like Copilot or Cursor?

I've been using KC for code reviews at Resume Matcher, and the experience has been amazing so far.

You can see it in action here: https://github.com/srbhr/Resume-Matcher/pull/638

I wanted to ask how this works with legacy codebases. My projects often have older patterns mixed with new ones.

The automation is interesting but I do worry about over dependency. My concern is whether developers might rely on it too much and miss deeper review thinking.

Reading this made me think about how long my pull requests usually sit waiting. I often wonder if small issues could be caught earlier. This feels like it could reduce that waiting stress for me.

This could be huge for junior devs on my team. They'd get feedback instantly instead of waiting for me to free up. Congrats, exciting to see this!

Wow, this is actually game-changing!

Gave it a first try today and its a very convenient feature. Certainly will continue using it.

This is wonderful. We've been using this Code Review mode since early testing and given tonnes of tips/feedback, love where it's at today!

If this reduces review back-and-forth even 30%, that’s already a win. Wishing you all the best :)

Very interesting product. Out of supported 500+ models, is there any recommended coding models list you guys suggest, as only few of them good in coding and review.

How does the AI understand our team’s specific coding standards and architectural decisions over time? (And can it learn from accepted vs. rejected suggestions?)

So much better than other code reviewers! Love that there are so many models available to power it

Congrats on the launch. OSS AI coding assistants are the future. What’s the one feature users fall in love with first?

This works as a Cloud Agent I presume.
awesome expansion of this product. Congrats to the team

pr bottlenecks are real , especially on small teams where everyone's busy building.

quick question: does it learn from previous reviews? like if i keep ignoring certain suggestions, does it adapt over time?

also how it handles false positives. nothing worse than an ai reviewer flagging stuff that's actually fine.

This looks super cool! I've been battling with CodeRabbit and it doesn't seem very helpful. We're a small team and we need something that's actionable. But CR seems to mainly give me fluff than actionable insights. I'm curious - how does Kilo compare to other similar products out there? Would be great to get some insight on that.

Does Kilo limit its analysis strictly to the diff / selected changes, or can it reason about a wider scope (files, modules, usage elsewhere) in a way that adapts to the size of the repository?

From my recent experiments with AI code reviewers, most of them either:

• stay diff-only, or

• try to scan “everything” and quickly become impractical on larger repos due to long processing times.

What I’m missing is something more adaptive — e.g. scope expansion that’s proportional to repo size or dependency graph, rather than a fixed “deep analysis” mode. Curious how Kilo handles this today, and whether repo-aware scoping is part of the design.

If you are a solo-founder or are vibe coding, having this code review gives you another check on the quality and security of your code. This is far better than skipping them and is faster & less expensive than hiring a fractional worker to do it.

It looks great! What new features you guys are planning to add on?

Congrats on the launch! Kilo Code has really outdone itself with planning and outlining complex tasks as well as bug fixing and error analysis. Will you develop a system for integration to make it more efficient for teams to setup & separate windows or workflows?

Hi everyone! Happy coding!,

I definitely recommend trying kilocode's new "review" mode feature.

It successfully reviewed my entire spaghetti code website initially generated with a free deepseek ai model that I was unaware of it's lack of html/css coding abilities.

It successfully referenced 12 out of 13 critical 🚨errors/⚠️warnings which saved me hours of debugging.

The only warning recommendation from kilocode's review mode I could do without is a warning: "oncentextmenu=return false;" "prevents right click menu".

I purposely turned off the right click context window for an image, but it's not a big deal at all.

Overall I rate the new review code mode a 9/10, would give a 10/10 but it just released today and I never experienced perfect software.

Enjoy the ai revolution in 2026 and thank you team Kilocode for the hard work I notice you guys put in! 💯

Yesterday, I was trying to install it. But also saw how many people started complaining about attacks and hack attempts. This opens doors to new business ideas to patch the blind spots.

This is what people have been waiting for: an open-source agent that can turn ideas into actionable plans and execution. It’s quite similar to Minara, which we launched yesterday. But we’re particularly focused on closing the loop from analysis to decision to on-chain execution in digital finance.

I love that it’s hackable. This feels less like a tool and more like a platform waiting for crazy ideas.

As a heavy power user of Poke for proactive, messaging-native assistance on top of my email and calendar, I’m curious: what are some workflows where you’ve seen Clawdbot actually outperform Poke in the real world.

I suspect operating systems will eventually become headless, driven by context and intent rather than windows and clicks. This feels like a step in that direction. Congrats.

Agents with full shell + file access hit scale issues fast when permissions and secrets are too coarse, and chat ingress gets probed as soon as it is public.

Best practice is capability based allowlists enforced by a policy engine like OPA, plus sandboxed execution (namespaces seccomp or gVisor) and append only audit logs per action.

How do you authenticate each chat channel and map users to scoped capabilities, and is per tool trace replay or human approval on the roadmap?

This is really exciting. I'd love to get in touch with the team to integrate with our OSS lightweight protocol for making agent actions safer by requiring verifiable intent + provenance + evidence before high-impact tool calls (fail-closed by default). It fits super well with Moltbot's agent/tool workflow (ie. wrapping tool calls in Pi sessions with a quick verification step via SDK/JS port). Could add proactive security for things like browser, nodes, camera, or system exec without much overhead. Would love to explore an integration and happy to prototype a skill/wrapper or submit a PR if you're open.
Either way, supported and wish you all the best.

This is a very “finally” product - assistants that can’t actually do things are basically autocomplete in a tab. The “chat anywhere” idea is especially compelling too: WhatsApp/Telegram as the UI feels like the most natural interface for real life.

I’m curious how you’re thinking about trust when the agent can run commands and touch files - what does the default “safe” experience look like, and how do users review or undo actions when something goes a bit off?

Congrats on the launch - if you nail safety without killing the flow, this becomes something people keep running quietly in the background.

Congrats on the launch! @steipete - Will be tinkering with Clawdbot over the weekend.

The best open source product till now!

this looks really great @steipete. excited to try it out :)

A new era of working flow.

I really like the combo of self-hosted + chat anywhere. Being able to talk to an agent from WhatsApp or Telegram while knowing everything stays local is a strong trust signal, especially for power users and builders.

Except running it locally, what other advantage it has? it is going to hit the same LLMs living on the cloud and sharing my data with them right?

giving a bot "full system access" is both awesome and terrifying. does it run in a sandbox/container so it can't accidentally rm -rf my home folder if i give a vague command?

starting using it a week ago, after I saw someone run it on Raspberry Pi and damn! :)

This is a powerful tool. I installed it on a separate computer that I was not using for personal used. The tool is very powerful and could delete or modify your files.

“AI that actually does things” is a bold promise. Exposing real system control + chat-first access is powerful, and also where UX, safety, and trust really matter. Curious how you’re thinking about guardrails as people automate more of their daily workflows.

Hi how can I prevent this from bankrupting me?

Hey PH, I'm Arvin, founder of LobeHub.

Two years ago, I created LobeChat, an open-source, multi-model interface. LobeChat now has 70k GitHub stars, 14k forks, 2 million lines of code, and serves over 6 million users worldwide — yet most of it was built by me and my agent teammates. Today, I'm excited to productize the way I work into a new product: LobeHub.

LobeHub is the next generation of agent harness. LobeHub matches what Manus and Claude Cowork offer and goes beyond with capabilities today's agents can't support:

• Anyone can effortlessly build and team up with agent coworkers to deliver complex, systematic work — even assembling a quant team to execute trades. If token usage reflects how much AI power someone can leverage, today that power largely belongs to engineers. My mission is to democratize token consumption, empowering everyday users to craft their own agent teammates.

• Agent teammates are always-on and evolve with you. Today's agents are mostly disposable tools — task-driven with shallow, impersonal memory. At LobeHub, agents are true teammates. Each agent has persistent memory, editable by you, allowing humans and agents to co-evolve over time. We're building long-term agent teammates that grow alongside you — not just agents that complete tasks.

• A fundamentally new, agent-first experience. Spin up agents or agent teams while writing, chatting, brainstorming — from ideation to execution to delivery — across your entire workflow. Here, agents aren't just tools, they're units of work.

• Community-driven intelligence. My philosophy starts with people. Humans matter. AI intelligence and shared human intelligence are equally important. Through the LobeHub community, anyone can discover, reuse, and remix agent teammates and teams — customizing them to fit their own workflows and needs.

• Multi-model orchestration from day one. Our vision began with LobeChat's multi-model support, leveraging each model's unique strengths. By orchestrating multiple models, LobeHub delivers better cost efficiency and enables capabilities that single-model approaches simply can't match.

LobeHub is the ultimate space for work and life: find, build, and collaborate with agent teammates that grow with you. We're building the world's largest human–agent co-evolving network.

Ask me anything! 🚀

I see that the landing page directly takes to the sign-up page. Is there a home page I can read more about the product? Or you recommend, I signup and explore? :)

Maker here 👋

We didn’t start LobeHub because we wanted “better AI chat.”

We started it because after building and using AI tools daily, it became obvious that synchronous, UI-bound AI breaks down fast.

Agents need time, memory, and a place to live — not just a text box.

I started to use the open sourced version 2 years ago, in both the infrastructure of our AI cloud, collaborating across our team, and play around with our customized agents.

It used to be only a Chat app, now the new version looks so much interesting and amazing! I started to ask it write another project for me now.

Please share your use case down below for sharing your unique way of using LobeHub 😍

It's always an exciting thing to build LobeHub. We are shaping how people interact with agent to boost their work and life.

While focusing on the interface, I'm happy to hear from you: do you like the UI of LobeHub? Is it easy to navigate and use? Or any words you'd like to tell the designer in LobeHub?

This doesn’t feel like “another AI chat app.” Lobehub is clearly aiming to be infrastructure, and that’s way more interesting long-term!

The ability to annotate and refine agent outputs inline speeds up my review process.

Watching agents collaborate in parallel on complex problems is pretty cool!!!

Most “agents” today are chat sessions pretending to be coworkers: stateless loops, siloed context, and brittle hand-offs. You end up doing the real orchestration yourself—copy/paste between tabs, re-explaining intent, paying tokens to reconstruct state, and losing the thread of why the work mattered in the first place.

The human part matters as much as the system part. Memory should not be a black box that quietly profiles people; it should be legible and editable. LobeHub’s approach is “white-box” memory: keep what’s useful, discard what isn’t, and let agents adapt to how someone works without taking away agency. The goal isn’t more AI output—it’s less cognitive load, fewer context switches, and work that stays coherent over time.

I tried the stock trading agent group and it provided clear, structured reports with risk analysis. Well done guys!

How do LobeHub agents build long-term memory and context without becoming stale, biased, or overly confident over time?

Hey can you please tell me how its work properly ?

Amazing product, congrats! LobeHub is a wonderful open-source agent project that can help people collaborate with an agent team. That's an interesting idea!

I see LobeHub becoming an essential part of remote team workflows; asynchronous AI collaboration is powerful.

How do users decide what an agent should remember vs forget?

What kinds of work become possible with agent teams that are fundamentally impossible with today’s single-agent or prompt-based tools?

I love how I can clone and adapt existing agents: great for learning, customizing, and experimenting. Congratulations on the launch!

70k GitHub stars is insane, congrats Arvin! Team of agents sounds cool until they start having their own Slack channel to complain about me:D Love the multi-model approach. Does LobeHub have smth like a boss mode where I can see if my agents are actually working or just hallucinating together?)

Congrats on the launch.

Agent teammates that grow with you is a strong concept. What’s the biggest problem LobeHub solves compared to single-agent setups?

Congrats on the launch 🚀
Curious to see how teams evolve over time once memory and co-creation really kick in, feels like this could redefine what “working with AI” actually means.

This feels like something you integrate into how you work, not something you try once. That’s rare.

I'm pretty excited about this launch. Excited to have a place to bring everything in one hub to build and have the memory and ability to shift from model to model. I just signed up and I'm ready to start learning how to get Lobe working for me! Congratulations on the launch here at Product Hunt!

love the "teammate" framing vs just "bot". since they are teammates, can multiple agents collaborate on a single complex task (like one researching, one coding), or do I have to manage the handoffs myself?

congrats on the launch @arvinx love the "agents that grow with you" framing. The persistent memory piece is exactly what's missing from most agent tools today/ curious how you're handling state persistence across conversations? Building something adjacent at mio.xyz. Congrats on the launch!

I noticed the LobeChat project shortly after it was open-sourced. I'm not a programmer, but this was indeed the first open-source application I deployed on a server.

Initially, it was to take advantage of the free quotas offered by various platforms, as many large language model (LLM) providers offer generous free credits upon registration. I used all of these through LobeChat.

Later, I found that LobeChat's design philosophy greatly helped me in understanding "how to interact with AI" and "how to use AI" in the early stages. I even shared my deployed LobeChat with colleagues and friends, which was truly a wonderful memory.

Although I rarely use it now, I'm delighted to see them introduce a new generation of AI interaction methods!

Really impressed by LobeHub’s vision of agents as evolving collaborators rather than disposable tools. The idea of agent teams with persistent, editable memory and multi-model support could be a game changer for workflows that require continuity and context. Looking forward to exploring how this redefines AI-driven productivity!

The co-evolving human–agent model is interesting. Curious how you think about trust and control as agents accumulate long-term memory.

Hey Product Hunt 👋 I’m Tommy, co-founder of Timeless.

Meetings weren’t the problem. Nothing happening after them was.

For a while, we tried notetakers. They helped us remember what was said. But remembering wasn’t the hard part. The hard part was everything that came next.

Your face to face? It’s a prompt.

Your Zoom? Prompt.

Your phone call? Prompt.

Your “quick sync” by the coffee machine? Also a prompt.

Timeless captures every kind of conversation, even the passive aggressive ones, and turns them into follow ups, proposals, tasks, stakeholder docs, and more.

The things you said you’d do tend to already be there when you go looking for them.

This is our first time saying all of it at once, and we’d love your support.

just say the word, and all this is there:

🧠 Agents that build themselves

Agents form from what you already do. Your calendar. Your calls. Your conversations.

Follow through Agent. Proposal Agent. Eat the Frog Agent.

📞 Works across Zoom, phone, and in person

No links. No setup rituals. We even built a phone number for real life conversations. Put it on speaker.

🗂️ Rooms that organize themselves

Every doc, decision, and next step grouped by project or client.

Shared with one link. No recap needed.

📄 Outputs, not summaries

Briefs. Slides. Next steps. The things you usually scramble to send.

Already written. Already filed.

We believe the most important moments don’t happen inside tools. They happen in the moment.

So we built something that listens closely and makes sure something actually happens next.

Join the conversation/Let’s talk —> www.timeless.day

“Action-taker, not a note-taker” is a strong framing but the hard part is precision. Curious how Timeless avoids over-acting or misinterpreting intent when conversations get messy, ambiguous, or half-joking.

When all is said... it's done! That's the beauty of timeless. We don't need to keep having meetings to talk about the meetings to follow-up on the action items to async the syncs. It can all happen at the same time, in the moment. Loved all the conversations that went into making this thing real!

Hundreds of conversations went into building this. Which feels fitting... because that’s the whole point.

I love the shift going from passive scribe to active agent that actually helps you do the work. Congrats to the team 🔥

Love it! How is it different from a normal note-taker?

One thing that truly is timeless, our conversations. love this!

Love the idea of agents forming from existing behavior. Curious how you handle context across informal, in-person conversations.

We didn’t want better notes. We wanted things to actually happen.

Congrats on the launch! I like the framing of conversations as prompts and the focus on outputs rather than summaries. From a product perspective, how does Timeless decide which outputs to generate from a single conversation, especially when discussions are messy or exploratory, and how do you avoid over-producing tasks or docs that don’t actually need to exist?

I love the demo video! "action-taker" is the right framing, but context is tricky. if i say "send them the deck," how does it know which file in my drive i'm actually referring to?

looks great @tommy_bar_av1 . we're using fathom but this looks like a nice upgrade to go beyond 'meeting notes'.

Really like this action taking approach! I am curious about how is context maintained across multiple conversations that happen across platforms? Do the agents link conversations on the basis of shared participants / topics of discussion?

Congrats on the launch @tommy_bar_av1 ! Action-taker > note-taker is spot on, actually extracting what matters is pretty daunting task (i'm doing memory for agents)
curious how do you decide what's signal vs noise?
onwards!

Hi Product Hunt! 👋

I'm Tullie, the founder and CEO of Shaped. Previously I was a researcher at Meta AI, leading several ML teams including one focused on Instagram Reels and Ads video ranking. I also created PyTorchVideo and was a core contributor to Pytorch Lightning.

We built ShapedQL because we realized that while retrieval has become easier (thanks to Vector DBs), ranking and relevance are still incredibly hard.

Most engineering teams we talk to are stuck maintaining a "Frankenstein" stack. To build a "For You" feed or give an AI Agent personalized memory, they have to glue together a vector database, a feature store (like Redis), a reranking service, and thousands of lines of Python spaghetti code.

We built ShapedQL to turn that "house of cards" into a single interface.

ShapedQL is a domain-specific SQL dialect that compiles down to a high-performance, multi-stage ranking pipeline. With a single query, you can define the four stages of modern relevance:
1. Retrieve: Fetch candidates from multiple sources (Hybrid Search, Collaborative Filtering, Trending).
2. Filter: Apply hard constraints (e.g., "in stock" or "within 50 miles").
3. Score: Rank results using real-time ML models (optimizing for clicks, purchases, or watch time).
4. Reorder: Enforce diversity so your users (or Agents) don't see the same 5 items repeatedly.

We're seeing teams reduce 2,000+ lines of maintenance code down to ~30 lines of ShapedQL, while shipping features like "Cart Upsell" or "Agent Memory" in days instead of months.

If you're not a fan of SQL you can also choose from Python or Typescript SDK's.

I'd love to hear your feedback and answer any questions about the syntax or how it works under the hood! 🚀

This is sick - more teams able to build better feeds. Are you able to share any feeds Shaped is powering that you're particularly proud of?

Real-time retrieval and ranking tends to break at scale on feature freshness and training serving skew when event volume spikes and backfills happen.

Best practice is strict offline to online feature parity with a streaming fed online feature store plus impression logging for eval and safe shadow or canary rollouts.

How does ShapedQL handle feature definitions and model versioning plus A B testing while keeping low p99 latency across retrieve filter score reorder stages?

Very cool. Congrats on the launch, fellas.

Congrats on the launch! Looks so good

Congrats on the launch @tullie_murrell ! How do teams actually iterate on relevance once they’re on ShapedQL? Is the biggest win faster experimentation, or just not having to touch infra every time they tweak ranking logic?

I had this idea when I was building @Tandem (a virtual office for remote teams) and saw how awkward it can be when someone can't mute themselves. We built the ability to mute a teammate, but I thought it would be even better if mute/unmute could become muscle memory...so I remapped caps lock to mute via @Karabiner .

It quickly became one of my favorite hacks - I think it's because Caps Lock is a large key, so it's much quicker to press than a combination of keys. Yay Fitts' Law!

When I got a new laptop and my Karabiner setup broke, I missed it so much I built a dedicated app - now others can use it!

I added a meeting app selector so you can change the keyboard shortcut to work with other meeting software.

I find myself using it all the time!
- About to sneeze? -> Hit caps lock to mute.
- Want to dictate a note via @Aqua Voice while on a call? -> Hit caps lock to mute
- Child throwing a tantrum? -> Hit caps lock to mute
- Loud car horns near your house? -> Hit caps lock to mute

Remapping mute to caps lock is more satisfying than:
- backing into a parking space
- changing your keyboard layout to Dvorak
- customizing launcher shortcuts in @Raycast

It's absolutely crazy that laptops are still shipped with a Caps Lock key and not a mute toggle, given how much time most of us spend on video calls. If anybody knows a way to remap the LED to your mute state, that would be even better, but as far as I can tell, it's not possible on MacOS.

Hey, my coworkers are going to be jealous because I can finally mute without looking down or fumbling with a keyboard shortcut.

The real struggle happens when you forget that you muted yourself and talk for one minute muted :D 😅

@rajiv_ayyangar Remapping Caps Lock to app-level mute is such a clean Fitts’ Law win; at scale the hard part is reliability across Zoom/Teams/Meet updates + making sure the mute state never desyncs.

Best-practice: read mic/app mute state (where APIs allow), add a tiny “you’re muted” HUD + optional haptic/earcon, and ship shortcut profiles w/ auto-detection per app.

Curious: are you planning a Meet shortcut profile + a failsafe that re-syncs if the meeting app loses focus or the shortcut changes?

My curiosity is high because this makes me wonder what other underused keys could become productivity heroes and I want to experiment immediately.

My meetings have survived sneezes, crying babies, and car horns thanks to this concept alone and I can’t stop thinking how much calmer my Zoom calls could be.

It would be amazing if the LED could show mute status because my habit is muscle memory but a small visual cue would make me feel extra confident on calls.

My meetings are constantly interrupted by kids yelling or dogs barking, and this would have saved me from muting with clumsy key combos more times than I can count.

I wonder if this works seamlessly across all apps because my calls jump between Zoom, Teams, and sometimes weird beta software and I don’t want to hit Caps Lock and stay unmuted by accident.

Honestly shocked laptops still don’t ship like this by default. We spend more time on calls than typing in ALL CAPS anyway. This just makes sense. @rajiv_ayyangar

Finally a real job for Caps Lock. I always miss the Zoom mute combo when the dog barks. Big key = easy. Going to try this. If you ever hack the LED to match mute state, that’d be perfect.

LOVE THIS! Congrats on the launch :)

I actually built something similar a while back but never launched it (I was trying to get too fancy with it). I love your approach of making it software specific vs trying to tackle it all at once.

@rajiv_ayyangar can you share a bit more on what you mean about remapping the LED to your mute state? Do you mean the caps lock LED light?

👋 Hey Product Hunt!

I'm Rushi, CEO & co-founder of Alfi. We're launching the first chat app with social skills, built to bring friends together.

💡 The Problem

Gen Z is leaving social media. But they're not going offline. They're retreating to group chats, smaller spaces where they're honestly themselves. People are still using 2010 utilities like WhatsApp and Messenger to manage their 2026 social life.

The problem? Group chats are where plans go to die. Someone has to be "the planner" or nothing happens.

Alfi fixes that.

💙 Meet Alfi

Alfi is your new social friend. The one who remembers everything, plans effortlessly, and keeps the crew together. Alfi also has a proprietary framework that knows when to talk (and when not to). We taught Alfi… social skills! No commands required.

It doesn’t simply process but understands context, group dynamics, and whether you’re actually asking for help or just chatting with friends.

Here's what that looks like:

• Book restaurants together, with Yelp built in(exclusive partnership), directly in the chat. No more "where should we go?" spirals. Reserve instantly.

• Create images with your friends: drop your crew into funny hats, wild scenes, anything you can dream up. Multiplayer image gen, first of its kind.

• Social calendar: upcoming hangs, all in one place. Everyone is on the same page. No more managing separate cals.

• Group Wrapped: who's the chattiest? Top inside jokes? Now you'll know. Something we all have wanted for the longest time!

• Reminders & Memory: Alfi remembers your spots, your preferences, and nudges you to actually stay connected.

• Customize your Alfi: Customize how Alfi responds, creates, and helps to fit in with your group’s style. Custom AI for each chat.

Of course, you have your solo Alfi too! The group chat is simply sacred.

🌍 Why This Matters
We believe the group chat is the real social network. Where the funniest moments and real conversations happen. With Alfi, we hope to bring you together more expressively, and a lot more often.

🔄 The Journey Here

We had first built text.ai, our AI that lives inside SMS, WhatsApp, and Telegram. With near-zero marketing:

• 10M+ messages processed

• 100K users across 150+ countries

• Growth driven almost entirely by word of mouth

But we learned something unexpected. People didn't just want answers. They wanted a social friend who understands their unique group and actually helps them hang out more.

So we built Alfi from scratch.

💬 What beta users are saying

"It reminds my bf when he forgets our Farmers Market dates. Finally, someone else keeping him in check 😏"

"We put ourselves in cute dresses together. So cute. I wanna buy now"

🌍 Why This Matters

We believe the group chat is the real social network. Where the funniest moments and real conversations happen. With Alfi, we hope to bring you together more expressively, and a lot more often.

🙏 We'd love to hear from you:

• What group would you add to Alfi first?

• What feature would make this indispensable for your crew?

• How many plans have died in your group chat?

Alfi is live now on iOS and Android. Available in 10+ languages.

👉 Download today: text.ai/app

With gratitude,
Rushi, Prahar, & Paras

The new group chat era begins now. Building Alfi was about figuring out how to make human connection effortless. It’s not just a chat app; it's the glue that will helps you hold your social life together. Can’t wait to see how your crews break it!

Learn, listen, build. After 10 months in the trenches of consumer AI we heard what our users wanted... unlocked access to AI as a social enabler, but in a premium chat experience. Hence Alfi. Dig in with your friends and family & tell us what you think!

This is cool - does it remember things about me and friends?

Been using Alfi in our group chats and it’s honestly great. Arguing with it in savage mode is way too entertaining 😂 but it’s also clutch for handling all the logistics so one person isn’t stuck herding the group every single time.

Alfi looks fun, good work team! What are some of the next targets on your roadmap?

Congrats on the launch! Treating the group chat as the real social network feels very on point, especially with planning friction being the core failure mode. How does Alfi decide when to step into a group chat versus staying silent, particularly across different group dynamics, without becoming intrusive or dominating the conversation?

Reading this description made me reflect on my own habits. I save so many links and never return to them. This approach feels slower but in a good way.

I really relate to this idea because my inbox is full and I barely finish anything. Reading on screens always distracts me even when I want to ficus. This feels like a calmer way to actually enjoy content.

This looks useful overall. I feel giving users more visible control over tone adjustments could help people like me trust the replies more.

The idea feels simple but thoughtful. My experience tells me that sometimes small changes in how we consume content make the biggest difference.

Hi, I was just scrolling and stopped here because this felt different. My attention struggles a lot online and this idea made me pause and read properly.

Oh, how fun! I love this so much. Looking forward to print and deliver. 🖤

Such a non-trivial product to launch in our eco-friendly era! You know, potentially, it can also help combat the problem of having a million of links saved for later and never revisited again. If you, say, print a couple of articles and keep them at sight - you'll defo read them.

Congrats on the launch! Making Canva pages conversational instead of static feels like a natural next step, especially for lead capture and self-serve flows. How does the chatbot decide when to answer directly versus pushing someone toward a form, booking, or other next step, and how much control do creators have over that decision logic?

Beyond basic questionnaires, it’s impressive to see it support payment processing, report generation and workflow automation, making it an all-around tool. For forms that require complex logical branching (e.g., showing different questions based on different answers), is the conditional logic setup intuitive, or does it come with a certain learning curve?

This is really clever! I actually try to have a beat in my head that I run to but never thought of a "pulse" or "metronome" to do this.

Personally LOVE the Apple Watch Haptics feedback, I don't run with headphones so this is perfect. The design is really nice as well. Congrats @pipe_abello - overall well done :)

finally a metronome that doesn't scream over my podcast. does the audio lower the volume when the beat hits, or does it just play a subtle click over the top?

hey @pipe_abello love the idea! will start using it today on my run... just a quick suggestion: add "product hunt" to the options on the "how did you find us" onboarding section 🏃

congrats on the launch! Runo seems extremely useful especially for those who are in the beginning of running journey (like me haha)
I like to run and I try to make it as professional as possible, but I can notice myself struggling with rhythm, especially when listening to different types of music. was curious is it possible to change my rhythm in Runo during the run, or shoud I stop, finish one session and start another with different rhythm?
thank you, and I wish you continuing success and great runs!

The move from just keystroke visualization to full mouse tracking is a game-changer! @rahul_mula I'm curious about the technical implementation - how did you handle the mouse drag visualization without showing the "Null" issue that one reviewer mentioned? Also, with Tauri as the framework, what was the performance impact of adding real-time mouse tracking on top of keyboard detection? This seems perfect for my coding tutorials where I need to show both shortcuts AND where I'm clicking in the IDE.

Hey everyone, Eldar here 👋

Thanks for checking out Outtalent.

At its core, Outtalent is about people. I built it after realizing that the biggest difference between engineers who make it into top tech and those who don’t is often the environment around them. The right peers, mentors, support, and accountability matter more than any single course or tool.

Our software exists to support this community and give our fellows an huge advantage, but the real value is the people and the culture we’ve built. Happy to answer questions or hear your thoughts 🙌

Join us outtalent.com

congrats with your launch!! wish u all the best!!!

Outtalent is like a career power-up for engineers. It gives you a system, feedback, and tools so you get more interviews, prep smarter, and land better tech job offers even if your skills were never the problem, just the strategy.

Congrats Outtalent!

Congrats Tilek!

Wish you the best!

Well done on the launch guys! I've been following you since Day 1 and I can surely tell you accomplished a lot so far. Well done also to your CEO Tilek which is surely driving the team with passion and energy, every single day! Congrats! 🎉

Hi! It looks really impressive and I love the idea. I get that there is a lot that you get out of it, but many people looking for a job might be fresh out of uni and lowww on money.
With that in mind, it does feel like the price point is very steep. I understand that the 1:1 coaching is a lot of work, but maybe it is possible to make it more modular, so also include a very basic package that would be more suitable for starters and doesn't include all the 1:1 support?
Great job though!

clean stack. curious how you're handling multi-chain address detection?

Very cool! How is this better than an SEO best practice cursor rule?

Congrats on the launch! Pulling real-time GSC and SERP data directly into the editor context makes a lot of sense, especially for reducing context switching. How do you keep the data fresh without running into GSC or SERP rate limits when agents are querying often, and how do you make sure the suggested SEO changes don’t accidentally do more harm than good?

So your target audience is founders? If yes, there might be a struggle even to come up with a right question, as SEO is very specific and requires some skills. There is a chance to stand out if you provide some guidance. Cheers!

Congrats on the launch! The laconic design looks nice, and the Bites are well done. For Topics, I like that they’re recommended based on interests, and that everything is built in a chat format overall. Do you plan to add more visualization to Topics, or an audio/video format? It might be difficult to read longer text for users, what do you think?

I've been building software for over 20 years, often for clients and frequently for myself. Trailward started as a tool to manage my own habit of "constant learning", but it evolved into something wanted to share.

It began as a system for structured learning plans. However, I found that rigid structures often led to procrastination. That led to the concept of "Bites" - 90-second sessions designed to extract a single, clear insight. The goal isn't to replace deep study, but to make the entry point for curiosity so low that you actually start.

I also integrated an Ikigai discovery tool to help map curiosities to potential career paths, generating tailored learning plans based on the results.

It’s a solo project, built with React Native and FastAPI. Give it a try, it’s designed to be effortless, and I guarantee you’ll walk away knowing something you didn’t before.

I’ll be here all day to answer questions about the product or the technical side of the build.

Congrats on the launch @stephanschulz ! Love the "recommended based on interests" piece - tailored learning paths only work if the app actually knows the user. Curious how you're tracking what someone's interested in vs just what they clicked on. YC founder here, researching user context for AI apps.

Hey PH! 👋

Built this after getting rejected by Apple one too many times.

You know the drill: you submit your app, wait a few days, then get a vague message like "Guideline 4.2 - Minimum Functionality." Cool, thanks Apple. Super helpful.

I tried pasting rejections into ChatGPT but it kept making up guideline sections that don't exist.

So I built AcceptMyApp:

🔍 Before you submit → scans your metadata, terms and privacy policy for rejection risks
📩 After rejection → explains what went wrong + gives you a ready-to-send reply

The AI runs on actual Apple Guidelines (not hallucinated ones), and spits out structured results: risk level, specific issues, fixes checklist, appeal drafts.

Pricing: First 50 apps are free. After that, one-time €3.99-€9.99 per app. No subscription bs.

Would love to hear your feedback and your worst App Store rejection story 😅

Very cool! Just went through frustrating, non-transparent process of hitting issues that weren't clear upfront

This is something that is unique and would love try our app before putting it up on App store

Clever, I see myself using this tool

without risking bans is the boldest claim here Reddit's spam filters are ruthless. does the "pre-flight checker" actually simulate the post against current subreddit rules, or is it just checking for banned keywords?

Hey everyone! 👋

I'm Kirill, the solo maker behind Zush. As a designer and developer, I was drowning in thousands of files named IMG_1234.jpg and DSC00567.png. Finding anything was a nightmare.

I tried existing tools, but nothing worked fast enough for batch processing or handled RAW files well. So I built my own app.

The toughest decisions were:
Local AI vs cloud APIs: chose cloud for speed (files processed in seconds, not minutes)
Tauri vs native Swift: went native to keep it at 14MB with minimal resources

One thing I'm really proud of is the speed optimization. Zush handles heavy files and batches of hundreds of images supafast (© to @zanderwhitehurst). It's still early access, so there might be some bugs — but I'm fully committed to jumping in and fixing things fast.

Whether you're a designer with tons of screenshots, a photographer with RAW libraries, or a developer documenting bugs — Zush should save you hours.

Would love to hear what features you'd find most useful!
Drop your feedback here or on our roadmap: https://zush.canny.io/

🎟️ GET 50% OFF your first month with this promocode: PRODUCTHUNT

Cheers! 🥃
Kirill

Hi Product Hunt! We built The Password App after getting a compromised password notification. While fixing the account, I found out that I had 8 other compromised passwords that I needed to fix. On top of that, I had the same password across dozens of sites from my early years without using a unique password per account. Changing them all manually felt like an impossible task.

That's the problem: everyone knows they should rotate compromised passwords, but the friction is too high so nobody does. Attackers know this too, which is why credential stuffing works so well.

We built AI agents that handle the tedious part: they open each site, navigate to password settings, and change your credentials automatically. The design principles we wouldn't compromise on were security and privacy:
- your passwords stay on your Mac, they're never sent to the cloud
- AI agents never see any of your sensitive login data, such as username and password
- built-in Guardian agent which acts as an additional defense against prompt injection attacks

But this is just the beginning. Your data is already everywhere. It's exposed in breaches, scattered across platforms, processed by systems you'll never audit. And it's accelerating: more accounts, more services, more exposure, faster than any person can track. You can't secure your digital life manually anymore. It's time to put AI on your side.      

We're building local-first AI agents that go far beyond passwords. These are agents that delete accounts you no longer need, send data removal requests on your behalf, and map your entire digital footprint to show you exactly where your information is exposed. This is what personal security looks like now: continuous, autonomous, and finally yours.                  

Currently macOS only, with a Chrome extension for browser integration. Would love to hear what features you'd want next!    

Great team solving a widespread pain point! I've worked with this team across multiple companies. I trust their rigor and genuine interest in solving real world security problems.