PH热榜 | 2026-04-21

Boyuan here, founder of RankAI.

My team and I have been heads down building in search for years. What's happening now is that SEO and AI search are wildly overpriced by people claiming they have some proprietary playbook or that they have “cracked” search, and half the time they deliver meaningless results.

What people do not tell you is that search is not deterministic. There is no formula where you do X, Y, and Z and magically rank on Google & ChatGPT. That is just not how it works. After working with 200+ businesses and going through YC, what we learned is that the real way to win is iteration: find the right opportunities, publish, monitor reuslt, double-down on what works, and keep retrying for what didn't work.

So we built RankAI to do exactly that. And it's fully autonomous!

Check us out: https://rankai.ai/

You put in your website, and RankAI will:
- find the queries your customers already search on Google and AI search,
- conduct deep research into each sub-topic related to the query
- create optimized pages to target the query,
- track how each page performs,
- and rewrite them until they start working.

So instead of paying for a slow manual agency, or AI generate SEO slop, you get an engine that does everything iteratively to get you result.

A few outcomes:
An ecommerce skincare brand reached 1.2M search visibility in 3 months.
A consumer tech company grew search visibility 13x in just 2 months.
A B2B equipment company doubled organic traffic in 3 months after years of flatlined growth.

The future of organic growth is transparent and autonomous. RankAI is starting this wave.

Would love blunt feedback — what would make this more useful for you?

What's your stance on AI-generated content and Google's spam policies? How do you avoid getting penalized???

Really impressive autonomous approach, quick question for the team: how does RankAI handle highly competitive niches where even iterative content struggles to break through?
Congrats on the launch! 🚀

Congrats on the launch! Looks like we're in the same niche... or I mean we're direct competitors 😁

Just dropped in to say hi and good luck with your launch 🚀

Feels like this could replace a big chunk of what agencies charge for if the quality holds up over time.

Very interesting. Excited to try it out!

geo has been chaotic ever since, and fianlly some poduct that can actually work. grats on the launch and would love to see updated support for claude and openclaw

Does this mean that RankAI continuously monitors the performance of generated content and automatically optimizes content that is underperforming or dropping in rankings?

interesting.. we use SEOBot currently. Why should I switch?

Signed up with my own startup project which was just launched a few days ago, and I read your reply to @sandra_jirongo about new websites, which makes total sense. I loved the fact that prior to signing up the site audit gives actual important information which I didn't know that I needed, so wel ldone on this
The Technical SEO analysis sold me to be honest :-)

Question - how does the app solves the problem of a website being multiple languages? For example, one of my older sites that launched 3 years ago is transalated (using a plugin) in 12 languages.

Cool product! How much time does it give content pages before deciding they are performing poorly?

Si la empresa en la que trabajo (productora de eventos) usara RankIA para pósicionarse y 10 empresas lo hicieran igualmente con la app, cómo resolverían la ia para diferenciar , o cual posicionaría "primero"?

How does this compare to running Ahrefs plus manually testing prompts in Perplexity and ChatGPT? Curious where the "autonomous" part actually saves time vs being a wrapper around those workflows.

what's the suggested budget to start trying it out?

Are you guys the reason why I see more SEO/GEO related content on Reddit right now? Who says SEO is dead, they just changed.

This is huge! Question on GEO - LLMs like chatgpt like to cite reddit threads in particular, does RankAI help with visibility there?

So @boyuan_deng1 @dyllan_liu2 this effectively reviews my website identifies onsite optimisation/content holes and then generates the optimised content for Google/LLMs?

How long does it typically take to see the first real results for a brand new site with no existing authority or traffic?

Very cool product! Is this optimized for newly launched websites with a clean slate to work with or more established sites already ranking pretty high in SEO search and just need refinement and a GEO boost?

Congratulations with the great product and bombastic release. I can imagine how excited you feel today.

Could you give us some of the websites you've optimized so we analyze it with Ahrefs?

Are you using something like SerpApi for the rankings?

congrats!!

Congrats on the launch ! Is it based on the number of pages published, traffic generated, or a flat monthly fee?

Congrats with the launch!

This makes sense from a growth POV. SEO has always been about iteration, but most teams don’t have the bandwidth to do it consistently. If this can actually automate that loop well, it could unlock a lot of overlooked growth.

Interesting. How are you measuring real buyer intent vs just driving traffic that doesn’t convert?

Does RankAI work for brand new websites with zero domain authority, or is it better suited for sites that already have some traction?

Hey everyone 👋

Thomas here, co-founder of Twenty, back with 2.0

v1 launched with one pitch: an open-source CRM you actually own. No vendor lock-in, no black boxes.

The feedback we kept getting: "I don't just want to tweak my CRM, I want to build on top of it."

So that's what 2.0 is.

The core addition is our new SDK (npx create-twenty-app my-app). Data models, custom objects, workflow actions, layouts, widgets, the command menu, all defined in code, living in your repo, shipping through your normal dev flow with whatever AI coding tools you're already using.

We also shipped:
- AI built in: chat with your data, plus custom agents and serverless functions
- Custom layouts & navigation: record pages, widgets, command menu, navbar
- The full toolkit: custom objects, fields, permissions, views, dashboards, workflows.

Twenty is open-source, self-hostable, and now fully yours to extend.

- 📦 Start building today → https://www.npmjs.com/package/twenty-sdk

Felix, Charles, and I are here all day. Questions and feedback welcome. 🙏

Was deciding between making my own light CRM or building on Twenty 6 months ago, for a new thing I'm working on. Honestly, Twenty didn't have what I needed.

Oh child, how have I changed my mind since then! The speed you've built at the past 6 months has blown my pants off, and I can't wait to see what's more to come - really happy I didn't waste time slopping something together, when you've achieved so much!!

Believed in this vision since day 0, so good to see it ship esp knowing the ton of work behind it!

Most "CRM + AI" offerings are just a chat panel bolted onto closed software. Twenty is the opposite: a primitive you can actually build agents against - in your repo, your stack, your premises, or all of the above.

Go @Twenty team + @thomas_df @_felx @cbo_twenty 🖤🔥

Been working with Twenty for 1.5 years, before v1 came out. This is love from the first sight.
I have not seen another team in the CRM domain who ships this much real value while staying this small. Neither I seen a CRM that is as easy to configure, or build on top of it, as Twenty.

If you ever wanted something from your CRM - Twenty is capable. If not right now, then it's on the roadmap soon.

The twenty team has been shipping like crazy for the last 3 years at least. Love the team and the vision 🇫🇷🥖.

It's so nice to finally have an Opensource CRM! Is there a marketplace of integrations ?

In 2026, I'm a bit ashamed we're still running Pipedrive and not something custom-built with AI. It's time to move out of the saas era.

In your experience, what's the level of effort (in days, story points, whatever) to move a typical setup from Pipedrive to Twenty versus building something from scratch?

By a typical setup, I mean incoming leads from Typeform, enrichment, automated and manual responses, deals, a tracker, and targeted mass mailings, as well as Gmail and Intercom integrations.

One of the best launch today! Do you think the future CRM is something users configure or something they program?

an open-source CRM with an AI-friendly SDK is the kind of thing I didn't know I wanted until reading this. is the SDK typed end-to-end or do you generate types from the data model on the fly?

I wonder how sales manager with custom built CRMs. Many sales teams in certain industries have trouble adapting to newer technologies. How would this work for onboarding less technical teams?

The homepage looks great but it crashes when scrolling down:

Great idea!

Been watching Twenty on GitHub for a while. The self-hosted option is what keeps pulling me back. Quick question: how's the HubSpot import story these days? Last time I tried, custom field mapping was painful.

Hey guys!

You’ve been a total lifesaver. I just started a small software house, and your help with organization has been a game-changer. Everything is running so much smoother now.

Much appreciated! <3

I’ve been on K2.6-code-preview for a while, and now it’s officially K2.6. It has been kind of wild!

The model really shines on long-horizon coding: thousands of tool calls across hours of continuous execution, strong generalization across languages and tasks, plus the ability to generate rich, animated frontends with real motion and 3D elements. The agent swarm upgrades (300 parallel sub-agents) and proactive 24/7 agent support also feel like a meaningful step up.

As always, Kimi keeps delivering frontier-level models as open source. Respect🫡🫡

Hey PH 👋

Kimi K2.6 is our latest open-source model, built for long-horizon coding and agents - 4,000+ tool calls, over 12 hours of continuous execution, with generalization across languages (Rust, Go, Python) and tasks (frontend, devops, perf optimization).

Open-source SOTA on HLE w/ tools (54.0), SWE-Bench Pro (58.6), SWE-bench Multilingual (76.7), BrowseComp (83.2), Toolathlon (50.0), Charxiv w/ python(86.7), Math Vision w/ python (93.2)

Live at kimi.com, the app, API, and Kimi Code. Would love your feedback :)

K2.6 offers SOTA-level performance at a fraction of the cost.

It's open-weights, it's fast, and optimized for long-context tasks across the codebase, as well as the day-to-day work needed to support an always-on agent like @OpenClaw and @KiloClaw.

Impressive.

300-agent swarm orchestration is wild — curious how reliable the long-horizon execution actually is in practice. Anyone tried it on multi-hour coding sessions yet?

How strict is Kimi with sensitive topics? How would you rate it against the big three US models on filter sensitivity toward information security, copyright, interpersonal boundaries, etc.?

I'm not talking about explicitly dangerous activity, but about legitimate tasks that that trigger the filters occasionally. An example is Claude Code refusing to configure the Microsoft Entra dashboard because it looks like a hacker attack to it.

Hi Product Hunt! 👋 I’m Tim, and we built Dageno AI for startup and growth teams who need to understand one thing clearly:
When customers ask AI “who should I use?” — does your brand show up, and why?

We spent years building the data layer behind AI visibility workflows, so we know how messy this problem is firsthand. Most teams are still relying on classic SEO tools, manual checks, or dashboards that stop at reporting. That is not enough when AI answers are shaping buying decisions.

So we built Dageno AI — a GEO data and agent platform that helps startup and growth teams win visibility across 7+ major LLMs:

1. Enter your brand
Add your site and let us map the landscape around it.

2. Find competitors and prompts
We generate relevant prompts and benchmark you against the brands your customers are already seeing.

3. Analyze real signals
We combine signals from your site, traffic, prompts, social, and mentions against our proprietary data layer so that you can uncover visibility gaps, understand what changed, and know what to fix next.

4. Turn insights into action
We do not stop at reporting. Our agent workflows help teams execute content, source, and workflow changes so that you can turn visibility into growth.

5. Track performance over time
See how your brand is represented across major LLMs, compare against competitors, and measure progress as you improve.

If you want your brand to be more visible in AI answers, try it out and share your feedback — we’re building this in public!

This is a big issue with similar products we've used.

How do you keep prompt generation clean and avoid weird brand mentions?

One of the best launches today! Quick question - how are you guys handling user support right now? Asking because I build AI support agents for dev tools and always curious what founders are dealing with at this stage!

Is this built around adding blog and other text context layers?

GEO is becoming more and more important, I'm very excited to see new tools working on that, congrats on the launch!

we’ve seen our organic traffic shift so much lately, but our 'mentions' in perplexity are basically a black box. classic seo tools just aren't built for this. having a proprietary data layer to track why an llm picks a competitor over us is the only way to stay relevant in 2026. @Dageno AI @tim_dageno_ai

hi congrats on your launch. May I know more about the base of how the content opportunities being decided? from public forum like reddit that people mention feature need in this product domain where the target product is currently missing? or it's from other AI generated insight using competitor research.

Interesting product direction. Turning AI visibility into something measurable and actionable is hard — most tools stop at dashboards.

If you can really close that loop, this is a strong product.

This is super relevant. Feels like we’re still using SEO tools for a world that’s already shifting to AI answers. Curious how accurate the visibility tracking is — would love to try this.

My website, pokecut.com, has been undergoing geo-optimization. This tool looks great and should be effective for my site. I am currently using it.

I’m still trying to understand GEO better. Is it basically the same as SEO, just for AI search, or is there something more to it?

Worth a look if you're seeking a set it and forget it GEO solution for your app, product, or startup!

Not only does Dageno create the content for you, but it also observes trending topics as they emerge so you're always relevant and on time!

Hey Product Hunt 👋 I'm Symo, founder of Devaito.

I ran a digital agency for years. Every week, someone walked in with a brilliant idea. They needed a website, a brand, a store, an app, SEO, marketing. I'd quote them weeks of work and thousands of dollars. I'd watch their face drop.

I didn't lose clients. I lost believers.

That's the day Devaito started. Not as a company, as a promise: no one should ever need permission, money, or technical skills to bring their idea to life.

Today, you describe your idea. AI plans everything, site, store, app, branding, SEO, content, marketing, sales, support. You review. You approve. AI executes. You stay in control, AI does the work. 24/7.

We didn't build a tool. We gave every idea the power to become a business.

Dream it. Approve it. Devaito builds and scales it.

If AI could take over ONE part of your business tomorrow, what would you hand over first?

Congrats on the launch, Symo.

The agency backstory makes sense as the origin, and you saw the gap firsthand. My hesitation with Devaito is the scope. Website, store, app, SEO, blog, social, support, and sales automation is a massive surface area to maintain quality across. Most tools that promise everything end up being the best at one or two things and forgettable at the rest. What does a real customer actually look like 3 months in? Are they still using all of it, or gravitating toward the parts that actually work?

Really nice! But you say “no one should need permission or money to start”, so how do you balance that with the reality that distribution (not building which is realtively easy based on idea) is the hardest part today??

hi Symo, congrats for the launch. I find this concept very nice especially the readiness of ios and android app comes along with the web stores. Could you explain more about the payment part, is it also fully integrated with built-in stripe or other alternatives?

the all-in-one approach makes sense for solo founders who don't want to juggle 10 different tools. we've seen clients struggle with connecting their site builder to their email tool to their analytics... having everything work together from day one sounds like it could save months of integration headaches.

this is interesting - the "AI agents keep working" part caught my attention. most no-code builders dump you after setup, but ongoing automation for content and customer support could be huge. what kind of tasks are the agents actually handling day-to-day? curious how hands-off it really gets after launch.

The real value isn’t launching—it’s continuous execution. If the agents actually perform, this is powerful.

Website, store, app, SEO, blog, social, support, and sales automation from one description — ambitious scope. The question is whether "AI agents keep working behind the scenes" holds up in practice or just sounds good on a landing page. 🤔

NO ONE is downloading another app. It's time to bring your AI agents to the channel that people actually use. Spectrum makes this super easy, allowing you to bring in your agent in less than 30 seconds. Fully open source so that you can run everything on your local machine. We also provide a free cloud-hosted plan for people to get started without any friction

The Photon team is cooking.

Earlier this year, @Flux already hinted at something important: people want agents to show up inside the interfaces they already use. Spectrum takes that idea and turns it into infrastructure.

What makes it interesting is not just the channel coverage, but the native rendering layer across iMessage, Telegram, WhatsApp, Slack, Discord, Instagram, and more. One API, but the interaction still looks and behaves like it belongs on each platform. That is a much harder and more important problem than it sounds.

Agent backends are getting crowded fast. The more interesting question now is: who becomes the interaction layer users actually like? Who builds the bridge between frontier agent capability and the communication surfaces people already live on?

@Spectrum has a real shot at becoming that infrastructure.

I often find myself wondering what the world of the future will look like. AI and agents have existed for barely three years, yet their impact on every facet of our society has been extraordinary. They have reshaped how we search, how we code, and even how we understand ourselves as humans. So what will the next five years bring? Will they truly become, as science fiction once imagined, an integral part of our society? Will we work alongside them as real colleagues, relate to them as friends, or even form deeper bonds? And when future generations look at them, will it feel no different from how we look at one another today?

With this belief, we created Spectrum.

Our mission is to make AI a living part of society - to bring its extraordinary capabilities within reach of every individual. Spectrum enables developers to connect agents to the platforms people already know and use every day: iMessage, WhatsApp, phone calls, Telegram, even hardware devices. We believe this is the first real step toward integrating agents into human society - they must appear where human life already happens.

And there is something undeniably powerful about this moment. For the first time, we feel agents not as distant tools, but as participants - joining group conversations, exchanging messages like friends. This shift opens the door to widespread adoption. My mother, for example, used an agent through iMessage for the very first time - and she’s someone who has never used any other app on her phone.

Agents aren’t built for ordinary people - yet.
But we believe that together, you and we have the chance to change that.

We’ve been using Photon for 3 months. After a brief conversation, Daniel and Ryan got our API key set up in under five minutes. The service is extremely stable.

It’s a young team full of energy, passion, and talent, I'm excited to see see more interesting progress you guys build next!

Supported and this will be a tool that developers love to get.

love this framing — meeting users where they already chat instead of forcing another app install. how do you handle context/memory across the different platforms? does each interface get its own session or is it unified?

What a launch video!

This is a interesting offering, and your thesis aligns with one of mine, which is that users don't want more apps or logins and would prefer to use the tools they already have. Are you able to do SMS messages to and from a phone number?

The interface war is already over—agents need to live inside existing user habits, not new apps.

Cool Julie! Does it use official APIs in every messaging app? Cause if so it's super useful for many founders here. Wish you all the best on this impressive launch!

The "omni-channel" approach for AI agents! 🔥 Bringing intelligence to where people already hang out is the most logical path to mass adoption. Can't wait to see more frameworks like Spectrum making agents truly "ambient".

the wearables integration piece caught my attention - are you pulling from Apple Health, Fitbit, or going broader? we've seen so many health apps that only look at one data source when the real insights come from connecting everything together.

Es excelente para poder analizar de manera rapida nuestros resultados medicos

My father who has a pretty complicated health history has been using this and swears by it. He has tried others and finds Perplexity to be the most accurate. I still caution him to have a bit of AI-skepticism but he feels Perplexity is highly effective.

Me parece genial tener una alternativa de IA de Salud, especialmente a quienes hemos tenido un historial clínico de cirugías nos ayuda mucho.

Interactive demos are one of those things I keep meaning to set up properly but never get around to. Static screenshots lose context fast. Does this work for flows behind auth, or is it mostly for public marketing pages?

Definitely worth trying if you’re building software products. Just recorded a few demos of my own projects and shared them on socials, curious to see where this goes next.

Absolutely nailed one of the issues founders and specially tech founders have and it is to create storytelling demos I have tried in a couple of my projects and it makes a difference compared to how I was doing it before.

interactive demos as a category feels overdue for an open-source option. how does it compare to something like Arcade or Storylane in terms of capture flow — is it browser extension based?

Interactive demos are a criminally underused format for technical content creators. I record a finance/financial modeling podcast (ModeLoop Podcast: https://open.spotify.com/show/0m...) and every time I interview a practitioner, the hardest part isn't the conversation — it's letting listeners "see" the model they're describing after the fact. An interactive demo layered alongside an episode could fix that. Does LiveDemo support embedding into longform content like a podcast episode page, or is it mostly SaaS landing-page oriented?

screen context as memory is such an underrated direction. curious what the privacy boundaries look like — does it stay local or sync somewhere? feels like the kind of feature that lives or dies on that answer.

Been thinking a lot about this problem, most tools don’t fail because of capability, they fail because they lose context too quickly.

This feels like a big step in the right direction.

Hey Product Hunt! 👋

I’m Yang, co-founder of Cosine (cosine.sh), along with @alistair_pullen. We’re excited to introduce Cosine Swarm!

For the past year, we’ve seen AI coding move from simple autocomplete to bounded task agents. At Cosine, we believe that the system surrounding the model intelligence has become increasingly important.

💡What is Cosine?

Cosine is an AI software engineering agent that works across every surface developers use. Instead of stitching together separate AI tools for the terminal, editor, desktop, and remote execution, Cosine gives you one shared runtime across all of them.

You can start in the CLI, coordinate through the desktop app, and scale tasks in the cloud without changing products or workflows.

👾 Unleash the Swarm

Cosine Swarm lets a single prompt launch parallel subagents for research, exploration, implementation, QA, and coordination. Each subagent works in its own isolated context, with its own tools and trajectory, so complex work can branch safely, stay reviewable, and move faster without turning into chaos.

Instead of a single-agent thread straining under the weight of multiple tasks, Swarm mode creates a coordinated hierarchy. An Orchestrator breaks down goals, Task Owners manage workstreams, and Workers execute tactical changes in parallel.

Assign complex tasks in the evening and find them completed, tested, and ready for review by the morning.

🖥️ Available now across our entire ecosystem

I’m incredibly proud of our team for this release. We built Cosine to feel less like a collection of disconnected agent experiences and more like a real system for AI-native software engineering.

Swarm mode is available now across CLI, Desktop, and the Cloud.

We’re looking for feedback from developers and teams working across large codebases, multiple tools, and complex workflows – especially anyone who wants more control, better collaboration, and a coding agent that scales beyond a single interface.

🎁 Grab 1 month of FREE access to our Hobby tier using the code above. 20M tokens to swarm with.

→ Get started with Cosine: cosine.sh

→ Read the docs: cosine.sh/docs

I'll be in the thread all day answering questions!

I cant tell you how nuts this feature is.

Here was my workflow previously:

• get on the train

• start 5 tasks

• get to work and debug the code for each and iterate back and forth, taking up most of my time for the day

• get the 5 tasks in by EOD

Here's my workflow now:

• get on the train

• start 5 swarms

• get to work, start 5 more

• come back to each, find that everything already works (because all the code has already been thoroughly reviewed, QA'ed and iterated on)

• keep merging them and starting more throughout the day

• get praised for how much work im getting done

• profit

• EOD ive merged 10-15 features

Its genuinely changing the game, this is what AI coding has been promising to be for so long and it finally is

Swarm has been pretty fantastic so far. It's been quite a nice workflow to just hand over a large task, let it work through it, come back later and review what it did.

I can finally feel my backlog of tasks and experiments start to clear... Well, it's got a higher throughput at least. Theres always another idea to experiment with

Such an exciting release. I’ve been loving using Swarm mode. As someone working in ops, it’s been a game changer for my workflow - helping with research, project planning, spreadsheet analysis, reporting, and turning messy information into clear outputs I can actually use. It’s also been great for hobby projects too. I even used it to build an interactive intro-to-coding app! Really excited for more people to try it 😊

Swarm has expanded my workflow so much. I can start up multiple swarms to tackle seriously big new features, additions or tackle tough bugs. I just have to check out the branch afterwards, give it a quick once over, and it's done.

If ever a mono-agent is failing you just kick it into swarm mode, trust me

Swarm is such a game changer. We were able to generate a fully-functional, design system-aligned mobile version of our website with a single swarm and prompt. It even included tests!

This is huge. Biggest problem I have with AI coding is context rot. Problem is too big to digest and AI just starts making sh*t up.

Cosine has give the AI the ability to go from coder to Tech Team HR, spinning up a whole org chart to make big problems easy to digest.

Congrats!

Congrats on the relaunch, Yang.

The orchestrator/task owner/worker hierarchy makes sense on paper.

It mirrors how real engineering teams divide work. The hard question is whether the coordination overhead between agents stays manageable on genuinely messy, real-world codebases, or whether complex tasks still quietly derail in ways that only surface at review time.

"Assign tasks in the evening, find PRs ready by morning" is a bold promise.

What's the failure mode when it goes wrong? Does Cosine surface a clean explanation of where it got stuck, or do you end up debugging the agent's decisions as much as the code itself?

congrats on the launch!

The orchestrator/task owner/worker hierarchy is a smart decomposition for parallel agent work. Most multi-agent coding setups I have tried fall apart when two subagents touch the same files. Curious how Cosine handles merge conflicts between parallel workers. Does the orchestrator prevent overlapping file edits upfront, or is there a resolution layer that reconciles after the fact?

Honestly the parallel angle is what scares me. Every time I've tried coordinating two agents on the same repo they end up stepping on each other's files and I spend the saved time resolving merges. Does Swarm coordinate which files each agent can touch, or is it more of a post-hoc merge thing?

The Orchestrator and Task Owner split is a cleaner hierarchy than I usually see in swarm setups. I am curious how the Orchestrator decides when to fan out versus keep something in a single thread. Is that driven by the task graph it builds up front, or is there a heuristic about codebase scope that triggers parallel branching?

Hi Product Hunt! 👋 I’m Yunfei, the founder of PageOn.

It’s a bit emotional to be back here with our 3.0. We built PageOn on a simple belief: Ideas are everywhere; impact is rare. Our mission is to bridge that final, painful mile of information delivery. We’ve seen brilliant professionals, growth-drivers, and students struggle—not with their ideas, but with the 'pixel-pushing' required to make them shine.

In the earlier versions, let’s be real, our AI had its growing pains. So, we spent the last year listening to your frustrations and re-engineered the whole brain.

We’ve evolved from a "Slide Maker" into a Multi-format Visual Agent.
🎯 Smarter Context Awareness: The Agent now truly "gets" you. Whether you’re pitching to a picky CEO or launching a campaign to a Gen-Z audience, the tone, layout, and narrative will shift accordingly.
🎨 One Prompt, Any Canvas: It’s no longer just about slides. Our Agent now generates a cohesive set of visuals—from social posters to infographics—instantly, ensuring your message looks stunning across all platforms.
✨ Point-and-Chat Editing: We’ve minimized manual labor. Just select a specific element, chat with the Agent to modify it, and let the AI handle the "heavy lifting" of redesigning.

Whether you’re a Sales lead needing a last-minute deck, a Small Biz owner creating a week’s worth of social assets, or a Professor visualizing complex data—PageOn 3.0 was built to take the visual burden off your shoulders.

We are here to see if our Agent’s brain is finally “smart enough” for your real-world workflows. We desperately want your brutal feedback.

What’s the most frustrating visual task on your plate right now? Tell us, and let’s see if our Agent can solve it.

⚡ Special Launch Perk: We want you to really push PageOn 3.0 to its limits. Use the code [M2L-W6D-N5F] to unlock a free month of Pro access. Heads up: This is limited to the first 1,000 redemptions and must be activated within 1 month. Jump in and see what our AI can do for you!

Let’s bridge that last mile together. 🚀

Will there be an API available soon? I'd love to integrate this agent into our internal tools to auto-generate reports.

How well does it understand complex layout requests like "make the visual hierarchy more aggressive"?

It's refreshing to hear a maker admit their AI had growing pains. Re-engineering the whole brain sounds like a massive undertaking. Can you elaborate on what specifically changed in the backend logic?

"Ideas are everywhere; impact is rare." — Love this mission statement. The tool definitely bridges that gap. Congrats on the 3.0 launch!

The slide-first AI output is interesting to me from an M&A perspective — the bottleneck on most deals isn't the raw analysis, it's how fast you can turn it into an IC deck or one-pager without losing structural integrity. I've felt the same pain on the modeling side: reusable templates with the right shape beat starting from scratch every time, which is why I put my project finance and valuation templates on Eloquens (https://www.eloquens.com/channel...). Does PageOn handle dense financial tables and footnote structures well, or is it more optimized for narrative/marketing decks?

Magic Layers is Canva's answer to a problem AI tools created for themselves.

It takes any flat image and converts it into a fully editable, layered design inside Canva, separating text, objects, and layout into components you can actually change.

The core problem: every AI image generator produces locked files. A JPEG has no layers, no live text, no movable parts. One typo or a slightly off background means starting over.

What's different here is the shift from tracing to interpreting. Conventional tools outline pixel regions. Magic Layers identifies that a block of pixels is a headline, not a decorative shape and restores it as a live text box you can type into.

Key features:

• Flat PNG or JPG in, fully layered Canva file out

• Text restored as editable text boxes — fix typos, swap fonts, translate copy

• Objects separated as independent movable elements

• Compatible with any AI image generator output

• Premium feature — usage counts against your monthly AI allowance

Built for content teams producing visuals at volume, creators iterating across formats, and marketers who can't afford a rebuild every time a detail changes.

How much time does your team lose today going back to reprompt or redesign when an AI visual lands at 80%?

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

This is pretty much unusable for anything other than some obscure prototyping. I gave it a simple task: First scroll of slack homepage. While it does really well interpreting where the fonts are and doing OCR, any flat objects it buchers hard and creates a low-res version bitmap of it (instead of sharp vector - (look at GET STARTED button background)). Meaning in the end, you can't use it at all as the final quality is just so bad that you'd have to reposition/draw everything by hand.

ORIGINAL:

MAGIC LAYERS:

ORIGINAL:

MAGIC LAYERS (note how it buchers the REQUEST A DEMO button cutting out it's bottom border):

This is amazing! Something I've really been missing while using Canva for my job as well as my personal life! Will this add on eventually be part of the free or the paid version? Congrats on the launch!

I have been using this today within Chat GPT/Canva and it is a game changer for converting great AI images into fully editable templates/projects!

Pretty good detection and masking. Inpainting of textured backgrounds is blurry, you'll need to touch those up.

Private STT in 2026 needs a clear answer to one question: does Harker run the model fully on-device, or is it local recording with cloud transcription? Those are very different privacy stories. Where do you land on this, especially for anyone handling client calls or medical notes. and also, congrats and good luck on the launch :)

Hey Product Hunt 👋

I'm Manuele, the maker of Harker. I shipped v1 last June because I was tired of speech-to-text apps that streamed my voice to someone else's server.
Harker transcribes 100% locally on your Mac. Nothing leaves your machine unless you explicitly ask for it. Your voice never does.

With v2, I'm doing something that scared me for months: I'm making the entire local transcription engine free. Forever. No trial, no nag, no feature gates on the private core.

Here's the logic. Local transcription is the thing people need to trust. Trust is built by giving it away. What I charge for is Premium ($5.75/month): cloud-based AI that takes your transcript and rewrites it, summarizes it, translates it, turns it into an email, a bug report, a Linear issue. That part genuinely costs money to run, and it's opt-in every single time. Your default stays private. Your voice, your biometric data, never leaves your machine.

A few things I'd love feedback on:

1. Which text transformations would you want in Premium next? I have a shortlist but I'd rather hear yours.
2. If you're a v1 user: the migration path. Tell me where it hurts.

Launch-day offer for the PH community: 40% off Premium annual with code PHLAUNCH40, valid 72 hour.

I'll be here answering every comment today. Ask me anything 🎙️

Harker has been a great app to use especially with long form typing! I find it to be very accurate and a better deal than others out there.

I've been using Harker for nearly one year now and have had a great experience so far! I love how easy it is to just talk when prompting AI, instead of writing everything down. Also very happy with how responsive Manuele has been, I've shared feedback multiple times and got back great responses!

You had me at “works in airplane mode”. I’d love to see something like this on iOS.

@mnlcpc Congratulations. And happy product launch.

ran into this building with voice input — the moment you route audio through a cloud stt service, you're adding a data processor to your privacy policy whether you want to or not.                                                           

local changes the equation: no dpa, no "what are they doing with this audio," no explaining to users why their voice ends up on someone else's server.                                                                               

the only missing piece for teams is windows — the compliance problem is platform-agnostic.

What I love is Harker is constantly improving and the accuracy gets better with every iteration.
Also I am thinking faster and able to complete work at a much higher speed than before.
I'm so grateful this tool exists and if you don't have it GET IT NOW!

I want the mobile app version please

Congratulations on the launch Manuele! This looks great - I hope you are also looking to build a companion mobile app shortly.

Measuring how a brand is perceived by LLMs is basically a new form of sentiment intelligence — and I'd argue the financial world needs an equivalent for prediction markets. Polymarket trader behavior is leaking macro signal all the time, and most finance teams still aren't watching it. That's actually the problem we're tackling with PolyMind (https://polyminds.netlify.app/) — AI-powered alerts on large Polymarket trades as a real-time signal layer. Curious if you're seeing Gauge Sentiment used by finance or risk teams, or if it's still mostly marketing today?

1. If AI becomes the interface, then brand narrative inside models is the real battleground.

GET 75% OFF with Promo Code: PRODUCTHUNT26

1. Only available for first 100 signups
2. Offer ends in 4 days! (April 24th)

----

Hey everyone 👋

So prospecting has gotten pretty tricky in 2026, right?

→ We've got tons of tools
→ All this automation
→ AI is literally everywhere

But somehow we still find ourselves stuck trying to find ideal customers, figuring out what to say to them, how to get them to reply, and how to actually fill up our calendar with sales calls.

And to make matters worse:

→ Old outreach playbooks are burned out and no longer work
→ Founders and teams have zero time to do it manually
→ Crap reply rates mean no sales calls, which means zero growth

We've all been there.

So we built Flow AI - and since our first Product Hunt launch, we've been heads-down making it into the one and only platform you NEED to book meetings on LinkedIn.

Here's how it works:

1. Find your ideal buyers
Search 300M+ contacts to build targeted lead lists - no Sales Navigator or expensive third-party databases needed.

2. Reach them at scale
Connect multiple LinkedIn accounts and let Agent Maya manage your outreach on auto-pilot—using a proven campaign. She warms prospects up naturally and builds your network while you sleep.

3. Generate, schedule and send DMs—instantly!
Maya jumps into co-pilot mode - reads your profile, understands your offer, checks out the prospect, and suggests the perfect reply to move it toward a call. You stay in control and can tweak outputs before sending.

4. Never miss a lead or opportunity
Every conversation, follow-up, and lead status lives in one inbox and smart CRM. We'll nudge you on follow-ups, highlight the leads worth focusing on, and make sure nothing slips through the cracks.

All of it follows a proven outreach playbook we've refined across hundreds of campaigns.

What's NEW since our first launch:

→ Lead Finder — 300M+ contact database built in

→ Auto-pilot — lets Agent Maya run proven campaigns (without any guesswork)

→ Co-pilot — generate, schedule and send LinkedIn DMs—instantly!
→ Multi-sender — run outreach across multiple LinkedIn accounts with safe rotation
→ Unified Inbox — teams can manage all conversations from one place
→ Integrations — connect your existing GTM stack (HubSpot, Zapier, and more)

And we've given a few clients the keys:

"I've had over 100 sales calls." — Simon Leich, CEO, CS Partners
"We landed our first 50k enterprise deal within 30 days" - Ashutosh Saitwal, Founder & CEO, KlearStack

We'd love for you to check it out.

Takes 60 seconds to set up.

Welcome to Flow AI.

We're here to help you build more pipeline.

Can't wait to hear what you think :)

Watched the video, looks very cool. However, every time someone on Linkedin reaches out to me and it feels AI, I challenge the sender and they subsequent message always falters. Also, I'm not happy when people send me AI generated messages. How is flow handling this?

Warning:
Checksum of brew install doesn't match.

Cask reports different checksum:
5f7a22bb3abf3986e49d84efc2143af937c34efe835a906e29d7e4ffcda069db

SHA-256 checksum of downloaded file: 9fe67225facf50d83c283db495851f1abba076dbfe01c6c53672cbbcee20c092

I want to install but also hit the same error reported by @digisome

First I tried this brew install as shown here: https://xisland.app/docs/

brew install --cask xisland
✔︎ JSON API cask.jws.json                                                                                                                                                   Downloaded   15.5MB/ 15.5MB
✔︎ JSON API formula.jws.json                                                                                                                                                Downloaded   32.0MB/ 32.0MB
Warning: Cask 'xisland' is unavailable: No Cask with this name exists.
==> Searching for similarly named casks...
Error: No casks found for xisland.

Then tried the brew install path shown on the landing page:

brew install --cask bluedusk/xisland/x-island
==> Auto-updating Homebrew...
Adjust how often this is run with `$HOMEBREW_AUTO_UPDATE_SECS` or disable with
`$HOMEBREW_NO_AUTO_UPDATE=1`. Hide these hints with `$HOMEBREW_NO_ENV_HINTS=1` (see `man brew`).
==> Tapping bluedusk/xisland
Cloning into '/opt/homebrew/Library/Taps/bluedusk/homebrew-xisland'...
remote: Enumerating objects: 40, done.
remote: Counting objects: 100% (40/40), done.
remote: Compressing objects: 100% (20/20), done.
remote: Total 40 (delta 9), reused 0 (delta 0), pack-reused 0 (from 0)
Receiving objects: 100% (40/40), done.
Resolving deltas: 100% (9/9), done.
Tapped 1 cask (13 files, 10KB).
==> Fetching downloads for: bluedusk/xisland/x-island
✘ Cask x-island (0.12.0)                                                                                                                                                   Verifying     7.7MB/  7.7MB
Error: Cask reports different checksum:     5f7a22bb3abf3986e49d84efc2143af937c34efe835a906e29d7e4ffcda069db
       SHA-256 checksum of downloaded file: 9fe67225facf50d83c283db495851f1abba076dbfe01c6c53672cbbcee20c092

Hello Product Hunt!

Andrey here, co-founder of illumi. Ling and I have been building toward today for a while, and we're excited to finally share it.

The idea came from watching the same patterns repeat across consultants, marketers and founders. After any ideation sessions, someone still had to turn scattered notes and half-finished thinking into a more polished result, and that transition consistently takes the most brainpower. When AI tools arrived, we started asking everyone we talked to the same question: do you actually feel less exhausted, or just differently exhausted? (like brain fry?)

Before you can generate anything with good quality by AI, you have to structure the chaos: half-formed ideas, sticky notes, workshop recordings, competing frameworks. Most AI tools hand you an answer once you've assembled everything yourself; we want to give your thinking somewhere to live while it's still messy, then carries it through to a finished output.

illumi is a visual workspace that connects multiple AI models on a shareable canvas, built for those accountable for turning messy group input into clear outcomes. You start with the raw input, explore and structure it on the canvas, keep your context intact as the thinking evolves, and end with a publish-ready deliverable in a single workspace.

We built it for teams that carry heavy, evolving knowledge across client projects, such as marketers, consultants, PMs. The longer goal is to make that thinking layer portable: a shared memory layer where business and technical teammates draw from the same knowledge base, feeding both their own thinking and their AI tools, and easily share across.

If you've ever lost your train of thought between messy ideation and the final deliverable, try illumi.

Andrey

Pretty landing page (illustrations) :)

Suggestion:
– when adding a sticky note, I would welcome the "form" it (resize)
– also I didn't notice where to change its colours (maybe some inspiration from the Figma Jam toolbar could make it more visible)

Have been using it since beta and a big fan of it 🙌 It’s great for teams that want to collaborate in a visual way with AI help!

love it! been using it for over a year for my brainstorming process in writing

This looks promising! I can see this being useful for product strategy sessions. Our workflow today is usually Miro -> docs -> ChatGPT / Claude → slides or our PM tool, so there’s a fair bit of back-and-forth.

Do you have plans to support more AI models or open up APIs/webhooks for integrations?

Hey Product Hunt! 👋

Sanket here — some of you might remember when we launched RapidNative last year.

200,000+ screens generated. Hundreds of real apps shipped. But I knew we could do better. So we spent the past year rebuilding everything from scratch.

Why we rebuilt it

I've been building dev tools for a decade — GeekyAnts, NativeBase, gluestack, BuilderX. Same frustration every time: why is it still so hard to go from idea to working app?

Every app builder gave you pretty screens. But no real backend. No database. No auth. You still needed a developer to make it actually work.

RapidNative v2 fixes that

You get a full-stack app — database, auth, file storage, real-time updates — all generated together.

My favourite part? Built entirely in the browser. No downloads. No Xcode. No setup. I've been shipping apps from a caravan while travelling. 😄

💬 Would love your feedback

Full mock data in design mode — as useful as we think it is?

PWA / Add to Homescreen — does skipping the App Store matter to you?

What integrations are next? RevenueCat, Google Maps, and AI SDK are on our list.

Try it free — no credit card, no demo calls.

I read every single comment.

The fullstack angle is the interesting bit. Most AI app builders stop at the UI and leave you wiring auth, db, and realtime yourself, which is usually where things actually break. How do you handle schema changes after the first generation, regenerate or incremental migrations?

Hey everyone, I’m Sachit.
I built YourMemory because I hit a wall with my own coding workflow. My agents were brilliant, but their memory was a mess. They either forgot my architectural 'gotchas' by lunch, or they got so bogged down in stale bug fixes from last week that they started hallucinating.

I realized we don't need a digital filing cabinet for our agents, we need a filter.
YourMemory treats context as a living thing. It uses 'biological decay' to let transient noise fade away while reinforcing the patterns and facts you actually use.

For the skeptics:
I’ve provided the full benchmarking scripts and the LoCoMo dataset on GitHub. We’re hitting 52% Recall@5, which nearly doubles the industry average. Why? Because our v1.3.0 Graph Engine doesn't just do keyword matching, it pulls in related architectural 'neighbors' that standard vector search completely misses.

It’s 100% local first (DuckDB), zero infra, and it’s finally stopped me from repeating myself to my terminal.

I’d love to hear how you’re all handling context amnesia right now. Tell me what your agent keeps forgetting that drives you the most crazy!"

Self-pruning memory is the right instinct — stale context is what makes most "personalized" AI apps lose their edge over time. I ran into this when building DishRoll (https://dishroll.netlify.app/), a weekly AI meal planner — old preferences (the chicken recipe you loved three months ago that's now boring) kept contaminating suggestions until we added explicit decay and recency weighting. MCP-level memory hygiene is a much cleaner place to solve this than at the app layer. What signals do you use to decide what gets pruned vs. kept?

Using the Ebbinghaus curve for context decay is a genuinely clever framing. The biggest failure mode I have seen with agent memory is not forgetting too much but forgetting the wrong things. Architectural decisions that were made months ago and rarely referenced can still be load-bearing. Does the graph engine help protect those kinds of low-frequency but high-importance memories from decay?

Ebbinghaus decay plus a graph engine is a clever combination, but the edge case I keep running into is old

architectural decisions that are still load bearing. A design choice made six months ago that constrains current code

is rarely in active use, so the forgetting curve would drop it, but the graph edges to it are exactly what new code

needs. How does the engine decide when decay wins versus when the graph pulls something back into scope?