Breadcromb
A browser that remembers everything and can act on anything
77 followers
A browser that remembers everything and can act on anything
77 followers
Your AI shouldn't start from scratch every time. Trace is the browser that remembers everything you've read, understands your work, and gives AI the context to help you research, write, and automate tasks without sacrificing privacy.
Breadcromb
'Act on anything' is super interesting! A browser already logged into my email and bank taking actions on its own is a much bigger trust jump than remembering what I read. QQ - where do you guys draw the boundary on what it can touch before a user hands it that much rope?
Breadcromb
@artstavenka1 Great question and I completely agree that there’s a big difference between an AI that remembers context and an AI that can take actions on your behalf.
For us, memory is the foundation, but action requires a much higher level of trust. The boundary we’re aiming for is that the AI should help users understand, prepare, and suggest actions before it ever executes anything sensitive.
For example:
Summarizing, organizing, extracting information, and helping you make decisions
Taking actions with clear user confirmation
Handling sensitive areas like email, banking, purchases, or private accounts only with explicit permission and strong safeguards
The principle is that the user should always remain in control the AI should be an assistant, not an autonomous actor with unlimited access.
This is one of the reasons we’re building Trace around transparency and user-controlled memory first. We believe trust has to be earned progressively before giving AI more agency.
Really appreciate the question it’s exactly the kind of conversation we want to have around the future of AI browsers. 🙌
Most AI tools feel like talking to someone with amnesia — you explain your project every single time. The "remembers everything you've read" angle could actually fix that. Does it work across different AI tools or is it tied to Trace's own interface?
Breadcromb
@ringo_td5 That's exactly the problem we're trying to solve. 😊
Today, Trace's persistent memory is built directly into the browser experience, so it works seamlessly across everything you browse. Our goal is to make this memory model-agnostic, so users are not locked into a single AI provider.
Long term, we want your personal knowledge to remain yours and be usable with different AI models, whether that's GPT, Claude, Gemini, or future models. The idea is that your memory should belong to you, not to a specific AI interface.
one thing i'd love is a way to manually tag or highlight snippets within pages so the AI can pull from those specifically instead of just the whole browsing history
Breadcromb
@glerobcu Thanks for sharing this. I really like this idea!
You’re right that there’s a difference between remembering everything you read and intentionally marking the pieces of information that matter most. A way to manually highlight/tag specific snippets and let the AI prioritize those when answering would make the memory much more precise and user-controlled.
This is actually very aligned with where we want to take Trace: giving users an active role in shaping their personal knowledge base, rather than just passively collecting browsing history.
I’ll definitely keep this in mind as we continue improving the product. Thanks again for the thoughtful feedback!
Curious to see how you draw the line between remembering useful context and keeping the user in control.
Breadcromb
@crystalmei Thank you so much for the upvote!
That question gets right to the heart of what we're building.
We believe the line is user control. Memory should feel like a personal notebook, not a black box. Users can always see what has been remembered, edit or delete it, and decide what deserves to become long-term knowledge.
Our goal is to remember only the context that genuinely helps you in future work not to keep a record of everything you do. By keeping memory transparent and user-controlled (with local-first storage by default), we hope people can benefit from persistent context without giving up their privacy.
We'd love to hear your thoughts when you end up trying it! 😊
Christophe, I live across an absurd number of tabs and forget half of what I read, so the idea of something that quietly holds onto all that for me is genuinely appealing.
Breadcromb
@matthieu_poitrimolt Thank you, that really means a lot to hear! 😊
That's exactly the kind of problem we built Trace to solve. So many of us spend our days jumping between dozens of tabs, articles, PDFs, and docs, only to lose track of where we found something.
If you get a chance to try it, I'd love to hear your honest feedback.
The app is completely free, and your thoughts would be incredibly valuable as we continue improving it.