Flowly
Your personal AI assistant, native to your desktop
Start new thread
Hakan

Flowly - Your personal AI assistant, native to your desktop

by
Flowly
•
Flowly is a native AI assistant that takes action across your apps and browser tabs. Summon it as a full chat, from the menubar, or through a notch overlay — one global hotkey away on macOS, Windows, and Linux. Since our January launch we shipped a browser agent, voice coach, and end-to-end encryption — your chats stay yours, even from us. Free forever — Pro plan for unlimited usage.

Add a comment

Replies

Best
Hakan
Maker
📌
Hey Product Hunt — Hakan from Nocetic, the team behind Flowly. We built Flowly because every "AI assistant" we tried was either a chat tab in the browser or a Slack bot that could describe what to do but couldn't actually do it. We wanted something that lived natively on the desktop, knew what was on screen, and could click the buttons for us. Flowly is the result. It runs as: - a full native chat app, - a menubar dock, - a notch overlay summoned by a global hotkey, and - a browser extension that lets the assistant act on the page — fill forms, drive Google Sheets, navigate Maps, draft Gmail replies. A few things we're proud of: Native everywhere. Real binaries for macOS, Windows, and Linux. OS-encrypted token storage (Keychain on macOS, DPAPI on Windows). Global Fn-key hotkey via a small native module. Persistent sessions. Long-running agent task? Walk away, come back — the stream picks up where it left off. iOS-style continuity on the desktop. Browser agent, not browser link. The extension reads the DOM and Flowly plans → acts → verifies. Not "here's a URL, you do it." Honest state of things: this is v1.0.18. Voice coach and notch overlay shipped recently. The browser agent works on most sites we've thrown at it but we'd love your edge cases. Free during launch. Two questions we'd love answered in the comments: 1. What's the one task you'd want an AI to actually do for you on your computer? 2. Which site should the browser agent work on next? Thanks for taking a look — we'll be in the comments all day. — Hakan & the Nocetic team
Ryan Hudson-Peralta (RHP)
💡 Bright idea

@hakanorensy Congrats on the launch of Flowly. I’m a product designer, accessibility consultant, and content creator focused on how technology can create more independence for people with disabilities. Personally, I was born without hands and unable to walk.

Flowly caught my attention because the idea of an AI assistant that can actually act on your computer, not just tell you what to do, has a lot of potential from an accessibility and independence standpoint.

I’d love to connect about possibly writing an article, and creating some content around Flowly.

Can you email me at ryan [at] eqacc [dot] com when you get a chance?

Hakan
Maker

@hudsonperalta  Hi Ryan — just sent you an email. Looking forward to talking; this is exactly the kind of use case we want to get right.

Saul Fleischman

@hakanorensy This is a really thoughtful breakdown of what makes Flowly different—the persistent sessions and actual DOM interaction (not just URL passing) seem like the real wins here. The native everywhere approach with proper OS-level security is solid too. Curious how you're thinking about the agent verification step when it inevitably encounters edge cases or novel UI patterns.

Hakan
Maker

@osakasaul Honest answer: today the user is the verification layer — actions sit behind approval modes (always-ask, allowlist, full). For novel UI the agent surfaces the step and waits. Where it falls short: the agent doesn't yet flag "this DOM looks different than expected" before trying. That's the next gap. What edge case were you thinking?

Natalia Iankovych

It doesn’t work on phones?

Hakan
Maker

@natalia_iankovych The desktop app is native to macOS, Windows, Linux. But we have separate Flowly apps on iOS and Android too — same account, conversations sync. App Store and Play Store links are on the right side of the page in Company Info.

Philip Kubinski
🧐 Good find

Hey Hakan - nice idea & a clear differentiation from others on this market. How do I ensure agents knows me? You know - who I am, what I do, etc. Without it, its output might be off or mediocre.

Does Flowly keep learning about me along the way?

Hakan
Maker

@philip_kubinski Yes — it learns as you talk to it. When you mention something about yourself or your workflow, it picks it up and uses it next time. On top of that, it also evolves in the background — reviewing past conversations and quietly refining what it knows about you, building up a small knowledge graph along the way. Everything's stored locally and you can read or edit it inside the app.

Hakan
Maker

Hi PH

First, thank you. Flowly hit #4 today and the day isn't even over. The sign-ups, DMs, and feedback have been wild.

Quick context for anyone landing here: Flowly is a desktop AI assistant that doesn't just chat — it actually clicks buttons in your apps and browser tabs through OS-level controls and a Chrome extension.

Native to macOS, Windows, Linux. One global hotkey away from anywhere.

Free during launch and staying free for now also paid options available.

If you try it and something feels off, reply here or DM me — I'm reading every comment today.

— Hakan, founder

Agrim Chopra
Hey Hakan, Great product you have here, congrats on #4 launch. I was also working on something similar before Regent, but with 1 additional feature : given enough context and inputs, the model could also actually predict what you might need, and when, and ask you before doing it : almost like a real human secretary. I did make some architectures for that which worked to a good extent, so should be possible. Just something to think about :)
Hakan
Maker

@agrimchopra  Hey, thanks for digging in. That proactive layer is honestly where we want to take Flowly next — we already run a background agent every few turns that reviews context and updates memory, so the foundation's there. Turning passive observation into proactive "you might want X" suggestions is the natural next step. If you ever feel like sharing what worked for you, I'm all ears.

Samir Asadov

The native, always-a-hotkey-away framing is the right mental model — chat windows are the wrong primitive for tasks you do 50 times a day. I built DishRoll (https://dishroll.netlify.app/) on the same principle for weekly meal planning: it has to live in the background, not be another app you remember to open. End-to-end encryption + local action on top of that is the part I'd love more PWAs and consumer AI tools to adopt by default. Question on Flowly: how does the assistant handle multi-step actions that span apps (e.g., "pull these three numbers from the model in Excel and paste them into the IC memo in Word")? Does it ask for confirmation per app, or is there a single pre-flight review?

Hakan
Maker

@samir_asadov  Thanks — and DishRoll's framing is spot on, "in the background, not another app" is exactly the bar.


On multi-step: today it works per-action with configurable approval modes (always-ask / allowlist / full access), so your Excel→Word example would prompt before each app touch unless you've pre-approved the patterns. We don't have single pre-flight plan review yet — agree it's the better UX for trust,

especially in finance flows. It's on the list.

Curious how you'd want it to surface — full plan up-front with one approval, or per-step with the option to "approve all"?