Oasis Browser for Mac - A privacy-first AI browser you can train anonymously

@akansha888 thank you so much for sharing this genuine feedback about Oasis, Akansha. We hate Ad Nauseam, and we are thrilled you feel like Oasis helps you focus and stay aligned. If you have any suggestions for us to double down on this or add new features to help you focus and do deep work in new ways, please let us know! You have no idea how much we appreciation your trying it out, and that you feel this way feels like a miracle to the engineering and product teams :)

@aj0671 You know more than anyone how much work has gone into this, and it's an honor to go to war with you today, Atharva. I just wanted to take a moment here and thank you for all that you've done for Oasis, from hooking up wires, to refreshing the most obscure ui and branding assets, to building the next wave of privacy-first ai features in browsing. You're a great engineer, and I'm so proud to work alongside you and have this moment where we can showcase all that we've poured in.

@ferdi_sigona Big thanks for raising this important question!

Short answer: Imported site passwords stay on your device, in Firefox’s Login Manager vault (encrypted at rest). They are not part of the local history/bookmark/semantic-index layer, and the Oasis Assistant has no tool or API path to read the vault. Using the assistant still sends prompts and replies (and sometimes page text or tab URL/title) to cloud models—that’s a separate boundary from the password store.

Long answer on how everything works under the hood:

https://www.producthunt.com/p/kahana/oasis-browser-import-bookmarks-passwords-history-and-autofill-data-from-other-browser

@rohan_mehere Woo! We are over-the-moon to hear feedback like this. When we set out to build "Oasis" we chose the name for a reason. We wanted to have a refuge we could genuinely fall in love with, a space where we'd be safe, where we could do deep work while feeling undaunted. If you have any ideas for new ai workflows we can add for you to amplify your capabilities more, please shout them out and we'll get right to shipping away.

@pournami_pottekat that's crazy! The same thing happened to me recently. I had my hands tied while eating a snack and watching TV, and I remembered I could just speak and say "Open Github in a new tab", and then the voice-to-text worked smoothly and Github opened. It was pretty insane lol

@fberrez1  Short answer. Yes, it's not too heavy from what we've seen so far!

Testing with 125 internal beta testers so far, we've received 1 mentions so far of Oasis slowing down performance on a device, but we haven't yet pinpointed whether that's an issue with the application itself or the device/user. We’re still early and tuning performance. If you try it and it feels heavy on your machine, we’d love to know your OS and whether the assistant was open. We’re optimizing for “powerful when you need it, quiet when you don’t.”

This version of Oasis is built on Firefox, so day-to-day browsing experience is in the same ballpark as Firefox.

A bit more detail:

• Browsing & privacy: History, bookmarks, and semantic search indexes stay on device.

• Assistant: Replies are powered by cloud models when you ask Oasis something, so you get capability without keeping a huge local LLM loaded all the time. Though, we have received questions already about supporting local models too

• In practice: If the assistant is idle, resource use should feel close to Firefox. When you’re chatting, using voice, or running browser actions through the assistant, you’ll see the usual spikes (CPU/network) you’d expect from AI + automation.

@gkanev Hey Gabriel, thanks for raising these points.

1. Peace of mind comes from the anonymous experience of using Oasis AI. None of your personal data is collected, only raw prompts and outputs which help us create our own "amplifier" model that seeks to improve the speed, accuracy, and quality of commands for you and the broader userbase. I explain this in detail in this video covering our privacy-first approach to training anonymously. You can also see examples of the exact type of JSON payloads that we receive when you send a prompt and learn more about this in our doc on interaction data.

2. We do not support self-hosted models yet, however, that is something we've been exploring and would be open to focusing on in the next sprints on the roadmap. We would also explore allowing you to use your own API keys for other models. Are there any local models you prefer? I've been using Qwen series off Ollama mostly.

3. I'm having trouble understanding your observation here. Which "desktop navigation" or URL are you referring to? Our main website is [https://kahana.co/] if you could point me to an exact page that is not loading, that would be helpful. Or am I missing something?

Thank you again for checking out our campaign and sharing these questions and observations :)

@hasitha_sigatapu When we first brainstormed the concept of Oasis, before a single line of code was written, we described our design philosophy as "buffing the ice rink." We have tried to focus rigorously on the baseline user experience and preserving the user's focus. We want Oasis to feel smooth as ice, with zero friction. Hearing comments like yours is the ultimate positive reinforcement for us, and it is humbling to feel like we are moving towards are most ancient North Star. While there are other "ai browsers" that may feel like AI has been added for the sake of hype, we have tried to stay true to the idea that Oasis is meant to be an environment for your mind. AI is there to preserve and enhance that environment, but never intrude on your mental flowstate and be the star of the show. The star of the show is always your mental state and capacity to focus and do deep, meaningful work. If you have other constructive feedback about features you'd genuinely want us to add in next, we are chomping at the bit to create new sprints and get back to hacking away :)

@jan_bremec Great use case! Thanks for sharing. Competitor research and doc spelunking is exactly the kind of workflow we had in mind.

Short answer: Today it’s more “help me find and work with what I’ve already seen” than “the browser predicts the next site you’ll want.” It gets more useful as your history, bookmarks, tabs, and hubs fill up locally, but the main lever is what you visited and what those pages were about, not deep tracking of how you scroll or click.

Two layers (worth separating)

1. On your machine — surfacing past work (no cloud model training your profile)

When you ask the assistant things like:

“that Unity doc on rewarded ads I read last week” or “the competitor store page with the gacha breakdown”

It can search:

• Semantic history search — recent browsing (~500 pages) indexed on-device with embeddings built from title, URL, domain/path, and a short text snippet captured when the page was visited. So it’s biased toward meaning (“IAP,” “LTV,” “App Store”) not just “you opened this domain 40 times.”

• Memory search — full-text search across open tabs, tab groups, bookmarks, and history (titles/URLs)

That index grows as you browse (incremental updates, persists across restarts). It does not today behave like a recommender that proactively pushes “you usually open Sensor Tower at 9am.” You invoke it via chat (“find…”, “what did I read about…”) or related tools.

2. How you interact with a page — when you’re on it

That’s a different path: tools like summarize this page or questions grounded in the active tab read visible page content and send an excerpt to the cloud LLM to answer. So for docs/API references, it’s interaction in the moment (you’re on the page + you ask), not the browser silently learning your click patterns over time.

What “trainable” means in Oasis (so expectations match)

• Thumbs up/down + comments on assistant replies → product improvement (anonymous or account-linked; your choice). That’s not on-device fine-tuning of a personal model on your machine.

• Optional “Personalize Oasis Assistant with my account” → ties assistant interaction logs to your account for better signed-in experience over time, still separate from uploading your whole browsing graph.

For your workflow specifically

• Competitor store pages / GDC posts / SDK docs: semantic history + memory search are the wins — “surface that article again” without digging through 200 tabs.

• Long doc sessions: summarize / ask-about-this-page while you’re reading.

• Repeat visits to the same sites: you’ll see them more in search results because they’re in history, but we’re not (today) ranking purely on visit frequency like a dedicated “favorites brain.”

To be super transparent, we’re not yet a full “workflow OS” that learns how you research (e.g. always cross-reference App Annie → spreadsheet → Notion) and automates that sequence without you asking. We’re closer to local recall + assistant actions on tabs/bookmarks/history that compounds the more you use Oasis for that work. Though, we are very interested in expanding our system to handle more complex workflow sequences of any type.

If you decide to experiment, we'd love to know:

1. When you’d want a nudge (start of day, after closing many tabs, when landing on a store URL, never).

2. What the nudge should do (reopen group, search history, summarize, bookmark to a hub — not all of the above).

3. Whether it should be local-only (patterns never leave the machine) vs okay with account-linked suggestions when signed in.

4. One moment last week where proactive would have saved you time — and one moment where it would have annoyed you.
   
   Thanks again for checking us out and supporting the launch! :)