PH热榜 | 2026-05-14

🚀 Hey Product Hunt! I’m Zino the founder of Spellar AI 3.0, and I'm beyond excited to be back with our biggest launch 🎉

Your support on our previous hunts (those #2 and #3 spots still give me chills 🏆) pushed us to ask a harder question: why do meeting tools only capture the moment — and never remember it?

So we built something different.

Spellar AI 3.0 isn't a note-taker. It's memory.

It joins your calls, captures every word, and builds context across your meetings — so you can actually use what was said, not just find it.

What's new in 3.0:

🧠 Cross-meeting memory — Ask what a client said three calls ago. Get an actual answer.

📁 Organized by client & project — Your context, structured the way you work.

📋 Templates — Set up the right AI context before the meeting even starts.

🤖 Your AI, your choice — OpenAI, Anthropic, Perplexity or Google. You decide who processes your data.

🔍 Open decisions & follow-ups — Never lose track of what still needs to happen.

We've always believed your meeting assistant should be native and bot-free, with support for 100+ languages. With 3.0, we're taking that further: meetings that don't just get recorded — they get remembered.

Would mean the world if you'd give it a try and share your honest thoughts. The Product Hunt community has shaped everything we've built, and today is no different. 🙏

Let's make meetings actually stick!

Zino

Today we launched Spellar 3.0 on Product Hunt, and I want to be honest about what this launch means to me.

The last PH launch was in June 2025. Spellar got great traction, kind reviews, and new users who took a chance on us. Then, for a while, we went quiet.

Not because something went wrong — but because we kept hearing the same thing in feedback: people would use our meeting assistant, get solid notes... and still lose context a few weeks later. They'd join a follow-up call and not remember what had been decided. The notes were there. The memory wasn't.

So we spent the past year rebuilding Spellar around that idea.

Spellar 3.0 isn’t just a meeting recorder with a better summary. It’s an AI companion that works quietly in the background — no bot joining your call, no one else noticing — and creates a persistent memory across all your conversations.

You can search for something from two months ago. Pick up a thread from a call you half-remember. Walk into a client meeting with the full story from every previous chat.

Most meeting tools give you notes.

Spellar gives you memory.

We’re live today. If this resonates with how you work, your support on Product Hunt means a lot to a small, bootstrapped team like ours

Thank you so much, your support means a lot!

hey team, my congrats! the memory feature is what got me. so if I have a related call today, can Spellar surface context from a meeting I had three weeks ago?

Hi product hunters, my name is Andrii, and I'm a developer on this product.We're live. Spellar 3.0 is on Product Hunt.

There's a specific kind of exhaustion that comes not from the hours but from holding a complex system in your head for weeks straight. Every edge case. Every architectural decision you made at 2am and then spent three days second-guessing. Every "it works on my machine" that turned into a two-hour debugging session the night before a deadline.

The memory layer was the hardest part. Making a system that reliably connects context across meetings — not just stores it, but actually retrieves the right thing at the right moment — is not a small problem. We rebuilt parts of it more than once.

But today it's out. And it works. And I'm pretty proud of that.

Spellar 3.0: AI meeting companion that remembers everything. No bot joining your calls. Just quiet, persistent memory running in the background.

If you want to see what we shipped — it's live on Product Hunt today

Congrats on the launch! I like the idea of a botless AI meeting companion. I'm curious whether it would be possible for Spellar to sort the recordings and meeting notes into related piles automatically. Would be a good time-saver!

Congrats on the launch!

Have a question about i18n support. What languages does Spellar support for transcription? Our team spans three countries, and not everyone runs meetings in English.

The summary quality is noticeably better than alternatives I’ve tried. Feels like it actually understood the conversation.

Looks great folks. Been struggling with AI notetakers' output being isolated - good to see someone is consolidating it with persistent memory:)

Hi all, my name is Yuliia, and as the QA engineer on this product, I've spent months doing one thing: trying to break it.
That's the job. Find every edge case. Every flow that almost works. Every moment where the product doesn't quite deliver what it promised.
What made this one different: the core idea held up.
The memory feature - the thing that connects context across meetings over time - is the kind of feature that's easy to get almost right and very hard to get actually right. The gap between "it usually works" and "you can trust it" is where I spend most of my time.
We're at the second one. I wouldn't have let it ship otherwise.
Spellar 3.0 is live on Product Hunt today. An AI meeting companion that remembers everything - no bot, no noise, just context when you need it.
Want to hear your feedback ❤️

I've been using Fireflies for about a year - what would be the main reason to switch to Spellar? Genuinely curious!

Congrats on the launch. A killer feature would be to make it open source

Congrats, team @zinovii_z @hotfixer Great update with memory angle! Can teammates each run Spellar independently and have their notes connected, or is it only for individuals? GL today!

My friend organizes their notes/knowledge from meetings in Obsidian, where he can interlink stuff with other knowledge.

Does your tool allow to go beyond meetings?

This is quite cool. The cross-meeting memory is what finally makes this category interesting. I've tried a few note-takers but the problem was always the same: it is great for the moment, useless two weeks later when you're trying to remember what clients actually said about pricing. The fact that it runs natively without a bot joiner is a big deal too! Does the memory search work across meetings with different participants or only the ones you attended yourself?

🚀 Already 11 hours into our Product Hunt launch day — and Spellar 3.0 is still holding the #1 spot 🥹

Is there a way to create templates for recurring meetings — like 1:1s or standups — so the summaries are always structured the same way?

Really interesting shift from meeting transcription to persistent memory. The cross-meeting context and ability to surface decisions from earlier conversations feels more practical than just searchable notes. Curious how Spellar handles conflicting information or changing decisions across long client relationships.

congrats, team! any plans for more integrations? I'd love to sync with Linear after calls.

The cross-meeting memory is the part that actually changes the workflow — most tools just dump a transcript per call and leave you connecting the dots yourself. Curious how it handles context when a client refers to the same project by different names across calls — does it pick that up automatically, or does it need some manual tagging to stay organized?

How does the 'bot-free' part actually work? Will anyone else on the call notice something's running in the background?

Most meeting tools just dump transcripts somewhere and hope you’ll revisit them later 😂

The cross meeting memory angle is way more useful in actual day to day work. Congrats on the launch!🚀

Noise! Nice update guys

Does the cross-meeting memory work retroactively? Meaning, if I'm starting fresh today, can I import old transcripts from another tool to seed it, or is the memory only built going forward?

Good luck with your launch!

The cross-device sync is underrated — record on iPhone during a coffee chat, summary is on the Mac when I get back. Question: any plans for Apple Watch quick-record (start recording from the wrist for impromptu hallway conversations)? Would unlock the "captured the idea before I forgot it" use case for me 🤔

Hi Product Hunt Community, 👋

I’m Anubhab, founder of Naptick. Really excited to finally share what we’ve been building. :)

Naptick is an AI sleep companion bringing an agentic era to sleep combining a bedside device, AI sleep coach, circadian light, adaptive soundscapes, distraction app-locking, and room condition intelligence to help people fall asleep faster, not just track sleep after the fact.

What is Naptick?

Naptick sits next to your bed and helps you build a better evening routine.

It can check your room conditions, start calming light and sound routines, help you talk through your day with an AI sleep coach, reduce phone distractions, and wake you up gently with light-based cues.

The goal is simple: make falling asleep feel effortless.

The Problem & Our Solution

Most sleep products today are passive.

They tell you your sleep score in the morning, but they don’t help much at the moment that matters most — before you fall asleep.

The real sleep problem is usually not lack of data. It’s late-night scrolling, inconsistent routines, poor room conditions, stress, noise, light, and a mind that refuses to switch off.

Naptick solves this by turning your bedroom into a guided sleep environment. Instead of saying “you slept badly,” Naptick helps you take action before the night begins.

What Makes Naptick Different

Most sleep apps still need your phone at bedtime. Naptick is built to be phone-free by design (the app is mainly for setup and preferences).

Users can start routines directly from the bedside device, control sound and light physically, talk to an AI sleep coach, and reduce doomscrolling before sleep.

It’s also not just another white-noise machine or sleep tracker. Naptick brings together the key sleep levers — light, sound, environment, coaching, routine, and phone discipline — into one connected bedtime experience.

Features & Benefits

1. Room condition intelligence: Naptick helps you understand whether your bedroom is actually sleep-ready.

Benefit: Better control over the environment affecting your sleep.

2. 1000+ adaptive soundscapes: Choose from calming sounds, nature, brown noise, piano, meditation, and more through our licensed content library.

Benefit: Easier wind-down without endlessly searching for the right track.

3. AI sleep companion: Talk to Naptick before bed to unload your mind, reflect, or get guided support.

Benefit: Helps reduce mental clutter before sleep.

4. App-locking for distraction-free sleep: Block distracting apps as part of your night routine.

Benefit: Less doomscrolling and stronger bedtime boundaries.

5. Circadian-inspired lighting: Evening light cues help you wind down, while morning light cues help you wake up gently.

Benefit: Your room starts working with your body’s natural rhythm.

6. Physical bedside controls: Tap, double tap, rotate, and long press to control sleep routines without picking up your phone.

Benefit: Fewer reasons to bring your phone back into bed.

Who is Naptick For?

Naptick is for anyone who wants better sleep without adding another complicated wellness habit.

Especially useful for:

• Founders and high-performance professionals

• People who scroll too much before bed

• Parents trying to create calmer nights

• Light sleepers

• Apple Watch / Oura / Whoop users who want interventions, not just scores

• Anyone trying to build a consistent wind-down routine

(Naptick integrates with wearables to create a stronger feedback loop.)

Our Ask

We’d love your feedback from the Product Hunt community.

Please check out Naptick, ask us hard questions, and tell us what would make this a daily bedtime habit for you.

And if better sleep matters to you, we’d love your support today. 🙌

Try Naptick, pre-book your device here 👉 https://launch.naptick.com/

Thank you PH ❤️

AG & Team Naptick

After building multiple 0 to 1 products and then scaling them upto 150M+ transactions per month, 220M+ active sessions per month, building Naptick had its own set of learnings - building scalable infra for 24x7 connected hardware devices, ensuring your devices work seamlessly all day and switching b/w places with multiple wifi networks is seamless unlike most other hardware devices that require rigorous setup after slight move, easiest controls by doing button taps on devices, seamless voice to voice conversations with the device that makes it your preference to use device by voice than by app. Go into a DND mode just by single voice command before drifting off to sleep..

I work at Isha Foundation, and I think even ashram like ours can adapt to these sleep tech devices for the tourists and visitors who come for the experience from the outside world. Congrats on launching!!

As a founder running more than 7 SaaS companies, I often get thoughts of tasks that are pending for me to be done while I am just heading to bed. :D

I guess with this device I can just share all of my pending tasks which come to my mind and then I am going to ask it to remind me tomorrow so that I get it out of my mind while sleeping.

I think the most interesting part of this device is that it operates phone free. The contradiction about other sleep tech is they are on your phone which means screen time before bed and right after waking up. Congrats Anubhab, Shilpi and team on shipping Naptick.

Our hardware vision is that sleep needs its own physical AI agent in the bedroom, a companion dedicated entirely to your rest.

Long term, we see it becoming the control layer for the bedroom: tracking your sleep, understanding your patterns, and taking action across connected IoT devices around you.

The current design makes it look like a bed side lamp. Do you plan to turn this into a Reddit Snoo-like plushie or perhaps a robot in the near future for kids, teens and young adults?

This phone free design seems intentional.
BTW, how easy is it to actually fight the urge to grab your phone in the middle of the night if say anxiety kicks in? Does the app-locking still work if you're stressed?

What stands out from a PM lens - the product decisions here actually match the problem:

Voice instead of screens (because nothing pulls you out of sleep mode faster than a screen).

Hardware at the bedside, not on your wrist (so the companion is there in the moment, not three hours later in an app).

An AI that learns your patterns (so the advice gets sharper every week, not staler).

Most sleep tech treats the night as something to measure. Naptick treats it as something to actually inhabit. That's a different product entirely.

I built Naptick because I was tired of being tired !!

The coolest part? It is not just a device. Its your personalized platform with a multi-modal four-model AI stack running under the hood. Sleep patterns + environmental sensors feeding our agent ensemble. The room itself becomes the input

Your Nap Agents are at work while you sleep. They sense the room, run your wind-down, watch the night, and learn what actually gets you to rest. By morning they have already shaped how your next day starts. Sleep that improves itself, night after night. Night 30, your Naptick is genuinely yours.

Drop your sleep struggles in the comments 👇 AMA on the agents, the hardware, the manufacturing chaos, anything!

While building Naptick, we were very particular that it be rooted in evidence.

We spent a lot of time studying the existing science around circadian rhythms, sleep health, performance, and health behaviour change, along with evidence-based guidelines on sleep hygiene, bedroom environment, and recovery behaviours, to shape the foundation of the product.

Naptick is designed to help you create an external sleep environment and internal habits that work with your biology, so you can slip into rest at the end of the day with less friction and wake up better prepared for the next one.

Congratulations

 Congrats on the launch, folks!

With the journaling feature, how does it work? Like, whatever we journal, does it then later reflect into our phones, or how is it captured so that we can revisit it later?

Combining hardware with AI for sleep is an interesting approach. Is the companion mostly focused on environmental factors like sound/light, or does it offer interactive guidance?

As a founder who often finds myself 'doomscrolling' or thinking about backend architecture at 2 AM, this is exactly what I need. The phone-free design is a brilliant move. How did you decide on which 'circadian light therapy' patterns to include?

Congrats @krutiparekh16 Do you guys have plans to integrate with Apple Watch / Oura data beyond just the morning sync?

Oh yeah! How many times in the evening I want to turn on relaxing music on my phone to fall asleep, but I find myself scrolling through my social media feed.

Your project protects me from such things.

What's the microphone policy — does Naptick listen continuously or only when physically activated?

Will the distraction app locking work if someone manually reinstalls a blocked app during bedtime hours?

Can the circadian light automatically adjust to seasonal changes in sunset/sunrise times?

Can I use my own audio tracks like a meditation app  inside the adaptive soundscapes library?

This actually looks really usefulllllllll 😭
My sleep schedule is so bad these days, so an AI companion that helps you sleep without scrolling or wihtout listening to music for hours on your phone sounds like a good idea.

The phone-free design is actually what got me — most sleep apps still need you on your phone which defeats the whole point. The fact that it actively adjusts light and sound based on room conditions instead of just playing white noise is pretty cool. Would love to know how the AI sleep coach personalises over time!

When I read the tagline "AI sleep companion," I thought that it was some kind of toy that you can cuddle and sleep with. hahaha But yeah, I think I like the design of the device. It can just comfortably sit next to my bed.

What happens if I lose internet  does the bedside device still work locally?

Can the circadian light therapy help with jet lag or shift work sleep disorders specifically?

Great work, @anubhab_goel1 Is the AI sleep coach more like a journaling prompt or a two way conversation?

Congrats, Does Naptick require a monthly subscription or are core features one time purchase?

Hey Product Hunt, I'm Natalia from the Tendem team at Toloka 👋

The Problem

Builders today have more AI tools than ever, but turning AI output into work you can actually ship is still harder than it should be.

Most options follow one of two flawed approaches:

❌ Pure AI agents. Impressive in demos, then miss things, hallucinate, and leave you doing the last 20% yourself.

❌ Freelance marketplaces. They work, but quality is a coin flip. You end up project-managing strangers, and the ramp kills the speed advantage.

After watching teams burn hours fixing AI output or chasing freelancers, we built Tendem to actually solve the problem.

How Tendem is Different 🚀

Tendem is a task delegation platform where AI and vetted human experts work together to deliver verified output, for work pure automation can't finish.

✅ Backed by Toloka's expert network. 10,000+ vetted specialists across design, dev, copywriting, research and more — the same humans who help train the AI models you're using elsewhere.

✅ AI + human expert in every task. No "AI-only" tier. Every output gets reviewed, corrected, or finished by a vetted human before it ships back to you.

✅ Per-task pricing. Pay for the tasks you send. No seats, no contracts, no committing to volume you don't have.

Who is this for?

If you're a founder, builder, or small team that's tired of copy-pasting AI output and thinking "well, now I have to fix this" — then Tendem is for you. Especially if your work mixes things AI is almost good at with things that absolutely have to land right the first time.

🔗 Get started today: try Tendem at tendem.ai. 20$ for our first task is on us.

For anyone curious about the thinking behind Tendem: we wrote a white paper a while ago covering how Tendem works, the metrics we care about, and how it compares with other AI tools and workflows.

You can check it out here: https://arxiv.org/abs/2602.01119

Hey Product Hunt – I'm Egor, Head of Engineering at Tendem by Toloka. 👋

The post you're reading right now? I asked AI to write it for you.

And you know what? It was fast and technically fine.

Clean structure, correct grammar, all the usual launch words: “AI-powered,” “seamless,” “revolutionary.”

It was the kind of text that says everything and makes you feel nothing. The kind of post you scroll past before the second sentence.

And that's exactly the problem.

AI can produce something that looks finished very quickly. But “looks finished” is not the same as “ready to ship.” Real work needs expert judgment, context, human taste, and verification.

That's why we built Tendem.

Tendem combines AI with vetted human experts. AI handles the fast parts. Humans review, correct, finish, and QA the output before it gets back to you. So if you've ever used AI and thought, “Cool… but now I still need to fix this”, Tendem is for you.

We're launching our open beta today, and we’d love your support, feedback, and toughest tasks.

Your first $20 task is on us.

Try Tendem → tendem.ai

Share feedback in the comments – we’ll read everything.

I used to be just a digital marketer. That used to be enough.
With AI in the game, expectations on marketers grew tenfold.

My clients of the past months wanted someone who could vibecode a whole website from scratch, build a brand design system, create a verified leadlist, set up paid media campaigns, produce video and image ads, and write all kinda organic content.

I became that one person for them, a marketing builder, 'cause they needed speed, no over-explaining and full executional trust.

On my end, it meant hours of advanced AI prompting, setting up Claude Chat, Cowork, Code and Design to work in parallel on Max tier, all while verifying output by Perplexity and humanazing it with Gemini. ChatGPT was there occasionally to support me mentally.

Then I came across Tendem.

Tested it as a founder, and got stunned. I could hand off half of my tasks, and get the final results, ready for my clients' work. Hours saved on my end.

I first got kinda pissed off at them for making it so easy for users to delegate tasks. Who would need my help then?

Then I decided - I could just very well be part of the team scaling this platform, 'cause it's straightforward awesome.

Kudos to the team, and so thrilled to be part of this journey!

One thing I’ve personally been using Tendem for a lot is competitive and market analysis work.

A surprising amount of useful information still lives across fragmented sources, pricing pages, ads, PDFs, gated content, or places agents struggle to reliably interpret. Having a workflow that combines AI research with human review has been genuinely useful for getting a more complete picture instead of just surface-level summaries.

Also worth disclosing: I’m one of the people building it, so I’m obviously biased… but this is genuinely how I use it day-to-day.

Congrats on the launch Natalia and the Tendem team! 👋

This hits a real pain point. I’ve lost count of how many times I’ve had to spend an hour cleaning up “95% done” AI output that should’ve been shippable.

Love the hybrid approach - pure AI agents are great for speed but fall apart on the last mile, and freelance marketplaces are slow to ramp. The “AI + vetted human on every task” model feels like the middle ground that actually gets you to done.

A couple quick questions:

1. How do you match tasks to the right experts in the 10k+ network? Is it automated based on task type, or is there a human layer there too?

2. What’s the typical turnaround time for something like a landing page copy + design pass?

The $20 credit for the first task is a smart way to let people try it without friction.

Definitely going to test this for some content + research tasks I’ve been putting off. Upvoted 🚀

I’ve played with Tendem for a few tasks! Love it, and genuinely happy to see tasks take a bit longer for the human in loop part. Helped me with sales leads, just more precise. Also felt nice to have a handshake with a person

congrats on the launch team Toloka 👏

Can we keep a specific message for the experts, or is the prompt meant for both the human and the AI when we need something specific to verify, for example?

this is honestly the part nobody talks about enough.. ai gets you 80% there and then you spend more time fixing the last 20% than it wouldve taken to just do it yourself. the human expert layer makes sense but how do you handle turnaround time, like is it minutes or hours? because speed is usually the whole reason people use ai in the first place

How do you solve QA process, as both humans and AI can maky mistakes (as we all know)?
If human - it should be expensive, if AI - it subject to hallucinations.

I’m building out my GTM motion, can Tendem substitute tools like Apollo or Clay for target lists?

Just submitted a research task that's been on my desk for ages. Excited to see the results, as it's exactly the kind of thing that requires human judgment to get right. Congrats on the launch!!

Interesting how it blends AI with human handoff — most agent platforms go fully autonomous but knowing when to escalate to a human is actually the harder problem. Curious how it decides the threshold for handoff

"Love the honesty here—most agents are impressive in a demo but fall apart on complex, multi-step work. Since you’re using Toloka's specialists, is the goal for the AI to eventually 'learn' from the human corrections on the platform to become more autonomous over time, or is the human-in-the-loop always going to be the core philosophy?"

World class team behind it! 🫰

Nice one, Natalia. The "no AI-only tier" thing is the bit that stuck with me ... everyone else treats human review as the upsell, you're making it the floor. Bold pricing move.

Building in a similar space (AI drafts, human approves before anything ships) so I feel the "last 20%" pain hard. Two things I'm curious about:

1. What's a typical turnaround look like? Per-task pricing is great until you're refreshing your inbox wondering where the human is 😅

2. Does the expert see the AI's first pass or work blind? I keep going back and forth on this for our own stuff ... seeing the draft saves time but sometimes anchors them into the AI's mistakes.

Good luck today! 🚀

How are you making money on this given that the work involves the human learning the context of the job, which isn't as fast as an LLM, and therefore how are you able to pay the experts fairly, assuming you take a cut?

A job that doesn't require much getting-up-to-speed would be better done by an LLM. If it requires strategic thinking, it should be packaged and compensated accordingly.

Second question, how do do your experts deal with context-switching? If I had to do 20 different freelance jobs a day, I'd get tired, which would make me reduce them to 10, at which point I won't be making much money, unless they're full projects, in which case I'd rather go to Upwork or TopTal.

And what's special about how you curate quality versus a freelance marketplace? At 10,000 experts, that's a lot of of competition.

Hey Product Hunt 👋

Ana here, co-founder of Theneo.

A quick note on why this launch matters to us personally.

When we started Theneo, we made one bet: AI was going to change how APIs get documented. We were right, but we underestimated the second half of that bet. AI didn't just change how docs get written. It changed who calls APIs in the first place.

Over the past 18 months, our customers have shown us things we couldn't have predicted. Fintechs watching agents hit their endpoints in ways no human ever would. Engineering teams spending weeks hand-rolling MCP servers that crash in production. PMs realizing their API catalog was last updated two years ago and nobody knows what half the endpoints do anymore.

Elva is what came out of those conversations.

A few things I'd love your honest take on:

1. Does the agent-readiness audit actually surface things you didn't know about your own APIs? (I'm hoping yes. Push back if no.)

2. The production-grade MCP claim is bold. Try it on a real repo and tell me where it breaks.

3. What's missing that would make you replace your current API client tomorrow?

I'll be in the comments all day. The harder the question, the better.

the api catalog audit is a smart angle.. most companies have no idea half their endpoints are outdated until something breaks in production. curious how theneo handles versioning tho, like when an api changes does it auto update the docs or does someone still need to trigger that manually

A lot of API tools feel overwhelming unless you’re technical, so I like the focus on making API management easier for both builders and users

I love the API catalog audit feature. When you are an early stage startup this is one of the hardest thing!

Do you offer that as a standalone product in itself. :)

Hey Ana, congrats on shipping. The framing ... "it's not just how docs are written, it's who's reading them now" ... is the most interesting thing in this post. Most API doc tools are still optimizing for humans skimming on a Tuesday afternoon.

Two questions on the agent-readiness audit:

1. What does "agent-readable" actually mean in your audit? Is it stuff like consistent error shapes and idempotency hints, or are you scoring something fuzzier?

2. When an agent hits an endpoint "in ways no human ever would" ... what does that look like in practice? Curious whether it's pathological retry loops, weird param combinations, or something else entirely. Feels like a goldmine of edge cases nobody documents.

Good luck with the launch! 🚀

As someone who deals with complex API integrations, the idea of an 'agent-readiness audit' is huge. It’s one thing to have docs for humans, but another entirely to have them structured for an autonomous agent. Does Elva suggest specific schema improvements, or does it try to wrap the existing mess into a more 'consumable' MCP layer?

Claiming production-grade MCP (Model Context Protocol) servers is definitely a bold swing! Most of the ones I’ve seen are basically toy projects. How do you handle authentication and rate-limiting when spinning these up directly from a repo? @mariam_lekveishvili1

Designing APIs for AI consumption is a very specific challenge. How does Theneo help with auto-generating documentation that agents can actually parse effectively?

Cool new launch guys. I like the free investors database - I'm not quite ready to raise. How often do you refresh the vc info?

Hey PH - Ivan here.

Fundraising today is still painfully manual:
finding the right funds, figuring out the right partner, writing outreach, tracking replies... repeat for months.

We built Causo to automate the entire workflow.

Upload your deck, and Causo’s agents:

• find relevant investors

• identify the right people at each fund

• research them using live data

• craft personalized outreach

• monitor campaigns and replies

No stale VC databases. No hallucinated matches. Just targeted fundraising outreach that actually saves founders time.

Would love your feedback 🙏

Love the honesty ... "$80M raised and it was NOT fun" is the most relatable founder line I've read this week 😅

The bet on a live-enriched DB vs prompting an LLM is the right one. LLMs confidently pitching you to partners who left the fund 18 months ago is a real failure mode.

Two questions:

1. How fresh is "live"? If a partner moved firms last Tuesday, when does Causo know?

2. Does it learn from outcomes ... e.g. if a thesis-match looked perfect on paper but the partner ghosted three founders in a row, does that signal feed back into ranking?

Rooting for the raccoon 🦝

this hits home. fundraising outreach is basically the same problem as b2b outreach -- matching the right person, right message, right timing. curious how you handle personalization at scale without sounding generic

Taking a deck and turning it into an end-to-end outreach campaign is a perfect use case for multi-step agents. Since Causo is monitoring the campaigns, I'm wondering how it handles the data logging. Does it integrate directly with CRMs like Salesforce to log these touchpoints, or does it act as a standalone dashboard for the raise?

This looks really interesting. I've spent hours on VC research in the past. Curious if you're adding linkedin outreach soon? That's where I've had a lot of success in the past, it seems you're email only for now?

This is the kind of AI automation founders actually need. Researching VCs manually is exhausting 😅 Congrats on the launch!

@dawid_baranowski @ivan_sem Congrats on the launch 🚀 How do you handle the relevance signal? Does Causo match on stage + sector only, or does it factor in things like check size, recent deal activity, or thesis alignment? Building an investor list right now for pre-seed and “match quality” is the hardest part.​​​​​​​​​​​​​​​​

Congrats on launch! Now this one is interesting and one of few projects I would personally you. I understand the startup side of the project but what about investor side. Is it cold outreach from their point of view or they also have Causo account and are reviewing matches directly on the platform?

Congrats, guys, you're shipping faster than Anthropic 😂
Btw how do you avoid these emails getting into spam folder?

Onboarding right now - stoked for this!

Congrats @dawid_baranowski , a must have tool for founders, specially the 1st-time-to-raise ones.

QQ, the platform sources and does the outreach to the VCs and angels, right? or just sources the best matches?

Pitch the right VCs, skip the grind reads as one of the cleaner positioning lines I've seen this week. Ran a quick Lastest baseline on the homepage while looking — turned up a small UX finding (Start Finding Investors anchors to a #start id that doesnot exist on the page, so the click goes nowhere):

https://app.lastest.cloud/r/ESO6NqR2Nzn2jQJMNJGrow

Hi everyone!

Notion used to be a place to work in. Now it’s a place to build on.

This release makes Notion more than a workspace and closer to agent infrastructure. By introducing hosted Workers and a dedicated CLI, developers can now deploy data syncs (like @Zendesk or @Linear) and custom agent tools entirely on Notion's infrastructure.

With External Agents API, you can now bring your own agents—whether it's Claude, Codex, or an in-house build—directly into your workspace. They operate as first-class collaborators that your team can mention, assign tasks to, and review.

The Developer Platform unlocks the full potential of agents for your entire team. What will you build?

Why?
Why not focus on fixing the editor and the slugginess first?!

As a developer .... I feel like Notion is an incredible workspace tool for note-taking, but it is fundamentally not built to be a reliable production database or a headless CMS.

Has anyone actually managed to build a stable, scalable production app on top of Notion without losing their mind?

Think about my biggest pain points:

• Expiring S3 Media URLs == game over

• The block payload is worse than walking on magma

• API Ratelimits is 3/s

What does your caching infrastructure look like?

This looks pretty neat. Can you integrate custom code snippets directly, or is it more about using pre-built templates?

The dedicated CLI for deploying custom agent tools is exactly what was missing for serious builds. I love the idea of agents as first-class collaborators you can actually 'assign' tasks to. Does the API allow the agent to report back its 'reasoning' steps within the Notion comment thread, or is it strictly focused on the final output/task completion?

It’s really pulling me in!

Hey PH! This is Alexis, Co-Founder of Raindrop.

Your agent fails at 1am, traces are in some SaaS dashboard, harness is on your machine, eval suite is in a third place, and Claude sees none of it.

We've been stuck in this loop. So we built our way out of it. Workshop is the first sane way to debug your agent locally.

It has two parts: a local UI and an MCP.

Every span streams live to a local browser UI and you can replay any agent run with edited prompts, models, and tools.

The MCP lets you create self-healing eval loops. Claude Code reads your traces, writes evals, and fixes what's broken.

It's free, open source, and works with all the agent SDKs you already use.

One command to install: curl -fsSL https://raindrop.sh/install | bash

Excited to hear what you think!

Welcome to the self-improving agentic world. Very solid team, great launch. Congrats Alexis!

This was so easy to integrate and view my agent traces. No more console logging!!

We've been building an agent and this is exactly what we've been looking for! so excited to give this a spin

Really excited about this! If I'm a fan of Vercel's AI SDK local debugger what should I expect from this one? Raindrop integration I assume?

This is incredible! Using it to improve your prompting and execution is already a massive improvement, but your agent being able to use it itself just tightens the feedback loop. Huge!

This is really interesting, excited to play with it

What’s the easiest way to hook Raindrop into an existing agent project?

Agents have been such a black box, you're all doing an amazing job and helping me make sense of building in the new world. Congratulations and thank you for this!

Finally, a proper debugger for the agent era. Bridging traces over to other agents via MCP is such a smart architectural move. I love the focus on local-first—it makes iterating on sensitive agent logic much more viable. Does Raindrop support custom visualizers for specific tool-call outputs yet?

@Resend announced Automations during their launch week a month ago, and @zenorocha just posted an update on X: 518,265 automation runs since then.

Impressive!

This feature is exactly what I've been waiting for — it makes arrangement so much more intuitive. Huge thanks to the @Resend team!

Resend's DX is already top-tier. For the new automations, can we trigger flows based on custom metadata passed through the API, or is it strictly pre-defined events for now?

Meta just launched a standalone app for sharing raw, unedited photos with your inner circle.

What it is: Instants is a photo-sharing app built on Instagram's infrastructure that lets you send real-time, unedited photos to your Close Friends or mutual followers, where they disappear after viewing or after 24 hours.

What makes it different: The no-edit, no-upload constraint is baked into the product at a mechanic level, not just a UI nudge. You can only share what the camera captures in the moment. Combined with disappearing delivery and screenshot protection, the authenticity signal is structural, not optional.

Key features:

• Real-time capture only — no uploads from your camera roll

• Photos disappear after viewing, and can't be viewed after 24 hours

• No edits, no filters before sending

• Screenshots and screen recordings blocked

• Undo button to retract before a recipient opens

• Private archive for the sender only, retained for up to one year

• Recap to Stories: compile past instants into an Instagram Story

• Snooze control to pause incoming instants without blocking

• Teen Account integration with shared time limits, Sleep Mode, and parent notifications

The interesting thing about Instants is that it frames authenticity as a product constraint rather than a user choice. Whether that model holds at scale depends on whether the Close Friends list is actually small and trusted for most users but as a design stance, it's a cleaner take on ephemeral sharing than most.

Note: The standalone app is currently rolling out in select countries only. The feature itself is available globally through Instagram.

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

so basically bereal but with instagrams social graph and disappearing messages. the no-edit no-upload thing is interesting but lets be real people will find workarounds.. they always do. the screenshot blocking is a bold move tho, curious how well that actually holds up technically

this makes me nostalgic for the spontaneity of social media. I love it, I hope it sticks around!

A standalone app just for this one feature?

Making authenticity a hard constraint rather than just a filter option is a bold move. I'm definitely curious about the anti-screenshot tech—handling that reliably across different OS versions sounds like a headache. Does it actually block the capture entirely, or just flag it?

Heheh like used to be snapchat?

👋🏻 Hi Product Hunt! I'm David, co-founder of Asteroid.

We spent the last year perfecting the experience of building computer use agents.

RPA, Playwright, and browser automation are as old as bad software itself. The problem got bigger with AI agents: they're useless if they can't connect to systems that lack APIs. But every computer use agent framework until now has overpromised and underdelivered.

Thankfully, 2026 LLMs like Claude Opus 4.6+ make this hard problem actually solvable.

We went deep on complex customer use cases and took a clear stance: it's better to automate 50 complex browser portals extremely reliably than 1,000 at surface level. We focus on repeatable workflows, not one-off tasks.

A year of iterations, major updates, and small shipped features later, we have a full-stack platform that combines everything you need to build, run, and optimize computer use agents.

Just last month, Asteroid agents completed 150,000+ executions across 1000s of EHRs, benefits portals, insurance carriers, Citrix, desktop apps, VPN-protected environments, including:

We're the only platform supporting:

• 🌐 Any environment: browser, Linux, and Windows desktop

• 🖱️ Any interaction mode: visual computer use, structured DOM, or scripted execution

• 🧠 Any frontier LLM: OpenAI, Anthropic, Google, and more

• 🤖 Meta-agent: builds and debugs your agents for you

• 💾 Persistent filesystem: each agent stores scripts, files, and memories for reuse across runs

• 🔐 Session persistence: agents stay signed in and pick up where they left off

• 📬 Email inboxes: every agent gets its own inbox for 2FA and notifications

• 🕵️ Proxy support: stealth browsers, residential proxies, custom VPNs, and Tailscale

• 👀 Live view & recordings: watch runs in real time, review sessions, and add human-in-the-loop checkpoints

• 🛡️ Secure & compliant: HIPAA and SOC II Type II certified

• 🧩 API-first: webhooks, MCPs, and SDKs out of the box

You need all of these together to deploy computer use agents across critical industries at scale. Our use cases inlcude:

• 🏥 EHR / Patient Record Extraction: hospice and post-acute patient ingestion across agency EHRs

• 🩺 Voice-to-Form Clinical Intake: GP receptionist call transcripts turned into submitted triage forms

• 💳 Health Insurance Enrollment: ACA enrollment, plan handling, CRM sync, and binder payments

• 📅 Patient Creation & Scheduling: ECW patient creation, referral intake, approval checkpoints, and scheduling

• 📄 Insurance Policy Management: creating, filling, updating, and endorsing policy forms across carrier and agency portals

• 💬 Insurance Quoting: submitting quote requests, configuring coverage, and extracting results from carrier portals

• 🔁 Insurance Renewals & Endorsements: initiating renewals, applying mid-term changes, and tracking endorsement status

• 🧾 Insurance Claims Submission: end-to-end claim entry via Citrix VDI and status tracking across portals

• 🗓️ Intelligent Scraping: schedule extraction across 10,000+ venue calendars

• 📝 Form Filling: government portals, banking dashboards, and internal finance tools

If you've used Cursor, Lovable, or Claude Code and felt that magical moment, you'll feel it again when you see Asteroid automating portals you'd already given up on.

Try it now here or book a scoping call to talk through your use case.

Use promo code ASTRO10 for $10 in free credits to get started.

Would love your feedback! 🚀

To give you some stats from our small engineering team, over the last 12 months

• we've shipped 3,093 pull requests

• we've closed 1,351 Linear issues

• we've scaled our infrastructure to support over 100x the volume

It's been a long journey, but I'm incredibly proud of what we've built and I'd love you to check it out!

I've used several Agents, specifically browser agents, over the past year and found Asteroid to be the best one. It was actually able to visit sites that require cookies and managed some of the harder to do things I wanted. Excited to see that the team is now going for computer use agents!

Great to see how the product has grown since inception, congrats on the launch!

(and the new landing is slick ✨)

I'm so proud of the work we've achieved as such a small team. I joined right after the last Product Hunt launch, and looking back I can't believe just how much the platform has changed. The way it looks, and what it offers. The things our agentic workflows are able to do for customers are truly game changing. So many people now rely on automations running on Asteroid, and they won't even know it! Unless they stop and think: "Wait, I got that GP referral so quickly this time around". Let's go! ❤️‍🔥

Healthcare interoperability is exactly where computer-use agents shine - EHRs don't expose decent APIs. How does Astro handle sessions that need MFA or CAPTCHA mid-flow? That's typically where automation gets flaky.

I've been looking forward so much for this launch, and to be able to share what we've been working and show what our computer use agents are really capable of!

The launch of Astro v2 is a game changer, which has enabled the intelligence to be injected into the repeatable workflows much earlier - resulting in faster, cheaper and more accurate executions. This is the natural evolution of the human layer interaction with agents at scale.

Can't wait to continue the work with such a great team!

The 'Meta-agent' for debugging is the feature I’ve been waiting for. Debugging browser agents is usually a massive time sink, so having an agent that can troubleshoot its own execution is brilliant. I'm really interested in the HIPAA compliance—how much of the 'Live View' and recording data is redacted or processed locally to maintain that standard?

Love it!

We've been using Asteroid for last-mile automation at Lodo and it's the only thing that's reliably handled the messy switching portals we couldn't crack ourselves. Astro genuinely feels like the magical moment you described. Chapeu!

Awesome to see all the progress

Excited for this! Curious, how would Asteroid work for businesses using legacy systems that are gated (ie Iconote for Rehab Clinics)? Is there any vertical or situation where it can’t be used?

Really stoked to try this out. I’ve been selling agents to mid-market/enterprise businesses and the biggest bottleneck has been gated legacy apps.

This is epic congrats on the launch! What's been some of the most surprising use cases that have popped up? I imagine things can get a little wild in the world of agents. Would love to hear what your guys' journey has been!

Hey everyone, Tammuz here, CTO and co-founder.

When we first built this, our power users were PMs — people who think at the functional level: "users should be able to filter by date." The agent ran with that and shipped it. Huge unlock.

But as more designers started using us, we noticed something different. The functional part wasn't the hard part for them — the iterative tweaking phase was. They wanted their screens pixel-perfect, and they didn't want to wait on a developer who might or might not land their exact vision.

So we built Design Mode. Same experience as sitting next to a developer, pointing at the screen and saying "no, move that up a bit — the padding is wrong" — except now you're directing AI agents. Click the element, tweak the property, see it live. If you're working in Figma or Claude Design, your design intent flows straight in — no manual transfer, no context lost.

Ask me anything about how it works.

AI agents that directly edit live UI pixels is a completely different approach from prompt-to-code tools. This feels more like pair programming at the pixel level. Wondering how it handles complex component states.

The 'build in production' workflow is bold. Does Fei offer a way to rollback or version-control the changes made in Design Mode, or is it intended for real-time hotfixes?

This is awesome!! Congrats on the launch!

That's an awesome addition!

The workflow of pointing at an element and having an agent tweak the code live is a huge unlock for the handoff process. I'm really curious about the code quality that comes out of Design Mode, though. If a designer makes twenty micro-adjustments to a single button, does the agent refactor the CSS cleanly, or do you end up with a bunch of layered visual hacks?

the designer-to-code gap is one of those problems everyone complains about but nobody fixes well. if this actually pushes to the real codebase from a visual editor, that's a huge unlock for small teams shipping fast

Hey PH! I'm Adir, CEO of AutonomyAI.

We built Fei Studio because of a pattern we kept seeing across every product team we talked to: iteration speed is what determines how fast a company moves. Every change needed a full cycle. PM describes it, designer specs it, engineer builds it. Repeat.

Design Mode collapses that cycle. PMs and designers ship directly from Figma, from Claude Design, or straight inside the live preview. Design intent moves into production regardless of where it originates.

If your team is still waiting on engineering for every small change, this is for you.

Ask me anything.

Hey Product Hunt! 👋
Thrilled to be sharing OpenClaw-OS with you today.

⏳The story
We've been deep in the generative UI world for a while now, building OpenUI as an open standard. Over the past few months, watching how people actually use OpenClaw, the AI that actually does things, a pattern kept showing up. Folks were running OpenClaw inside Telegram, Discord, Slack DMs. Your agent can do the work. The question is whether you can keep up with it.

You'd ask OpenClaw to analyze something, schedule something, run a script. It would. But the output would scroll past, get buried under the next message, and three days later you'd be searching threads trying to find what your own agent did for you. The agent was doing real work. The chat window was the bottleneck.

🔑Apps are the unlock
The thing we're most excited about: OpenClaw-OS lets your agent build apps for your specific use cases. Not one-shot answers. Actual apps that stick around and keep working.

💹A sales dashboard that pulls from your CRM.

💌An inbox triage app that sorts and drafts replies on a schedule.
🗞️A standup digest that assembles itself before 10am.

🧾A pipeline tracker your whole team can open and act on.
🔬A research workspace that watches a set of sources and tells you what changed.

You build the app once, you stop re-prompting forever. That's the shift. You're not chatting with an agent anymore, you're running a small system the agent built for you.

💬Why this beats chat
Once your agent is doing real work, chat falls apart. You can't see what's running. You can't manage multiple agents. Files get lost in scrollback.

OpenClaw-OS gives every part of agent work a real home.
🤖Agents (run multiple, switch like Slack workspaces).
🧵Sessions (named, saved, resumable).
📱Apps (the persistent ones above).
▶️Artifacts (documents and datasets the agent produced).
💿Context (the files and sources it can see, all visible to you).
⏰Cron jobs (scheduled work, in one place instead of buried in a config file).

And one more thing. We brought generative UI into the chat as well. Your agent responds with live UI: a chart when you ask for a sales breakdown, a form when it needs to confirm a meeting, a table when you want a list. The output adapts to the task.

Check it out at

openui.com/openclaw-os.

Drop feedback in the comments, especially if you're already running OpenClaw seriously.

What's missing? Which parts do you actually reach for? We're listening.

The jump from a chat interface to a 'usable app' is where most agents fail. Does Openclaw generate a standalone frontend, or is the 'app' hosted within your own ecosystem?

Super fun built something with Generative UI that is for everyone. I built a dashboard to monitor Github stars with OpenClaw OS.

The jump from one-off chats to persistent apps is exactly where AI tools should be heading. Most chat interactions die after the conversation - turning them into something reusable is a real unlock. How are you handling state persistence across sessions?

This is exactly what I've been looking for. I use browser automation heavily for B2B workflows and the local-first approach with real Chrome profiles is a game changer - no more fighting with headless detection. The MCP server integration is a nice touch too. How does tab claiming work when multiple agents need to coordinate?

Running against your actual Chrome profile is smart - sites behave very differently with real history and cookies. How do you handle CDP detection? Thats where most local automation setups get flaky.

Browser agents become really powerful once they can reliably navigate authentication flows and multi-tab sessions. I like the local/open-source direction here since debugging agent behavior is still a major challenge in production environments

Open-source browser use is the missing link for local agents. Does this support standard MCP (Model Context Protocol), or is it a custom automation framework?

Hey,

I'm Ilias, a content creator and software developer. I've been creating content on YouTube for more than 3 years (80 videos, with an average duration of 1 hour). It's become harder for me to find video files and moments

After I got a bill from Google Video API for a couple of video analyses, about 450$ (couple of videos).

So I built a local-first tool that:
* Transcribes your footage on-device
* Analyzes frames: faces, objects, scenes, and text on screen
* Indexes everything so you can search in plain language ("find me when I'm at my desk looking excited")
Now, you can search using natural language to find the exact moment using local AI models. I don't wanna upload my videos to the cloud to get them indexed.

Start indexing your videos: https://edit-mind.com

Congrats on the launch @iliashaddad3 and @rohanrecommends !

Can the user select multiple found moments and send them as a batch to the editing timeline?

The whole picture looks like an intro from the series "Lie to Me" with Tim Roth. Detecting micro facial expressions etc. :)

Love the concept as I have about 20TB of videos to search and sort through. Congrats on your launch. Besides adding to the integration with the various editing programs, what is on your upcoming roadmap for the foreseeable future?

Hi Ilias, good luck. Which local models do you think work best with it?

Searching video like a conversation is a game-changer for editors. Does this index the visual elements of the frame, or is it primarily relying on the audio transcript?

I've been using Pressmaster for a few months and it's got a really smart way of generating content using your own voice. Particularly like the 'interview' mode for crafting articles.

The interview-to-content pipeline is clever. As a founder I have a ton of half-formed thoughts and opinions but turning them into LinkedIn posts or blog articles always falls to the bottom of the priority list. If this actually captures my voice from a quick conversation and publishes it - that's a real time saver.

The interview/source-material angle feels like the right direction for founder content. The hard part with “authentic voice” usually is not just matching vocabulary; it is preserving the founder’s actual judgment: what they reject, what tradeoffs they care about, and where they are willing to be specific.

I’d be curious how much the workflow captures negative examples or “don’t say it this way” notes. That is often where the real voice shows up.

Thought leadership often suffers when it sounds too 'robotic.' How does Pressmaster.ai capture a founder's specific 'voice' or contrarian takes without smoothing them out?

Exciting to see a new, full featured Mastodon + Bluesky client!

The business model is also interesting; maybe similar to @Ivory for Mac?

Indigo is free to download and try out, but interacting with posts and creating posts requires payment. Our premium tier is called Ultraviolet and has monthly, annual and one-time purchase options.

Prices are regional, based on a purchasing power index and are based on the US prices, which are:

$4.99 per month

$34.99 per year

$119.99 one-time purchase

Please note that the one time purchase may not include all future features in perpetuity as ongoing and that the services, Bluesky and Mastodon, may change what features are available without notice.

With the federated web growing so fast, this is a much-needed bridge. Does Indigo allow for cross-posting, or is it primarily meant for consuming feeds in one place?

Excited!

aggregate GEO scoring gets weird the moment a brand operates in more than one language. ChatGPT and Perplexity often disagree on the same brand depending on whether the prompt is in English or Polish, so a single number hides which market actually needs the fix first.

Having partnered with brands, sports and the creator economy in the last shift to social and mobile, thrilled to do it again in the era of ai search driven discovery!

This looks super helpful. Searched for my brand on Perplexity last week and the answer was completely wrong. Didn't know where to even start fixing it. This is timely. Congrats! 👏

Looking great! I was actually trying to tackle the same problem manually for https://warpply.com, so I’m really curious to see how well this works in practice. Haven’t tested it yet, but it definitely looks like a huge time saver 👏

Thanks for your post and comment

Love the idea of standardizing computer use via MCP. It opens up so many possibilities for custom agents. Do you think it could be used to automate data entry from legacy apps directly into something like a structured database or even a spreadsheet?

Hey Product Hunt! 👋

I'm Rashmi, cofounder at Troopr Labs. We make Enjo, used by 600+ companies including Netflix, Snowflake, and Grammarly for support automation. We're launching Enjo Help Center today.

Here's a conversation we've had with dozens of support teams:

"We know we need a help center. We've been meaning to set one up. But nobody has time to sit down and write 20 articles from scratch."

So the help center never gets built. Or it gets built once and nobody maintains it. Customers can't find answers, they file tickets, and the team answers the same questions every day.

We asked ourselves: what if you didn't have to write anything? What if you just pointed the AI at your website, answered a couple of questions, and got a working help center?

That's what Enjo does. Pick a template, paste your URL, and the AI generates your articles, organizes them into collections, and gives your customers a portal that actually answers questions.

But here's the part I'm most excited about: it gets smarter on its own.

When a customer asks something your help center can't answer, Enjo escalates it to your team. When they resolve it, the AI drafts a new article from that resolution. Next customer with the same question? Answered. No ticket needed.

Every conversation makes the help center more complete. Escalations go down over time, not because someone wrote more docs, but because the system learned from your team's real answers.

Free to start. All features, 200 AI replies/month, no credit card. Your help center stays live even after AI replies are used up. Starter plan is $95/mo when you need more.

What's been the biggest thing stopping your team from building a help center, or keeping the existing one from going stale? Genuinely curious what the blockers look like for people here.

Rajesh here, CEO at Troopr Labs 👋

Building a help center is the easy part. Keeping it alive is what kills you.

Every team we talked to had the same story. Built one, it was great for a month, then it slowly rotted. Nobody had time to maintain 40 articles.

The core idea behind Enjo is that a help center should be a compounding knowledge base. Sources go in, AI compiles articles, customers query, and every interaction feeds back to make it more complete. When a question gets resolved by your team, that resolution becomes a new article. The loop closes. Coverage grows from use, not from someone manually writing docs.

To keep that loop running long term, we built the Help Center Agent. Tell it "rewrite all billing articles for our new pricing" or "reorganize everything by user journey." One prompt, bulk changes. You review and approve.

The website URL is the fastest path in, but not the only one:

→ Zendesk/Intercom/Freshdesk tickets? Enjo extracts articles from your team's repeated answers.

→ Docs in Notion, Google Drive, Confluence? Connect them directly.

→ Most teams end up combining 2-3 sources.

We've been building support infra since 2019. Same AI layer that Netflix, Snowflake, and Grammarly run on daily.

Ask me anything. Genuinely.

Great to see this live! The closed-loop system where team resolutions automatically become draft articles is definitely the right way to think about knowledge management. It shifts the burden from writing to just reviewing. Congrats to the Troopr team!

Wow, this is how help desk should be redesigned AI-first, ground up. Love that it can build articles from customer quieries. Congrats @raj42 @rashjbp and the team!

I tried Enjo Help Center today and was genuinely impressed by how quickly it understood our product. Right after signup, it automatically generated a clean, well-structured set of help articles using our website and documentation. The content was surprisingly relevant and saved us a significant amount of manual setup time.

Really excited to see companies start using this!

We've been building it for months, so there's a lot to explore but the best part is we made sure resolved conversations don't just disappear, they become the raw material for the next article.

Would love to hear how it lands for your team! 🙌🏻

Really excited to finally see how teams use Enjo 🚀.

We actually built this to solve our own support headaches first, and used it internally before deciding to launch because we wanted to be sure it genuinely worked in everyday support workflows.

Would genuinely love to hear what resonates, what doesn’t, and how other teams end up using it. Let’s go 🙌

Super excited to be a part of this! Creating a help center has always been a daunting task, and now with Enjo, you can generate contextual articles to answer complex customer queries.

Hi Product Hunters!

I'm Dan, co-founder of .dodocs — and yes, this is our second launch here, but a very different one.

Our first launch was a document extraction API. Useful, but narrow. What we actually kept seeing — across hundreds of accounting firms and startup founders — was a much bigger problem hiding underneath.

Accountants spend over half their working hours chasing documents, matching invoices, and reconciling statements.

Work that adds zero value to their clients. Meanwhile, founders wait 3–14 days to get a simple answer from their fractional accountant — "what's my cash balance?" or "how much tax do I owe this quarter?"

So we stopped building a tool and started building an OS.

.dodocs is now the Accounting OS — it sits on top of QuickBooks, Xero, and Zoho and runs the work that accounting teams currently do by hand.

What's live today: 🔁 Chase Loops — AI-powered document chasing via email and WhatsApp. Clients get followed up automatically. Escalations happen only when a human is truly needed. Saves 24h/month per firm. Reconciliation — Automatic invoice ↔ statement matching with fraud detection. Flags duplicates and vendor anomalies before they hit your books.

Shipping next: 📅 Period Close · 🏥 Client Health · 📊 Team Performance · 💸 Payroll Agent

We're at 1,000+ users, growing 3× YoY — and we're just getting started.

As a thank you to Product Hunters, we're giving 2 months free — the code is in the last screenshot of the carousel.

Backed by Techstars Atlanta. Built by a team that has lived this pain firsthand.

Would love your honest feedback — especially from accountants and founders in the comments. What module would make you switch tomorrow? 👇

— Dan & the .dodocs team

Which part is live today, the document chasing or the reconciliation piece?

Hi. Reconciliation is the part I’d want to try first. Is matching deterministic, AI-assisted or a mix? Also curious how you show what was auto-matched vs human-approved.

Good luck

The 'no busywork' promise is huge for founders. How does DoDocs handle the categorization of complex edge-case expenses that usually trip up automated systems?