Most AI tools make you explain the context before they can help. Goldfish already has it. It privately remembers what you’ve been working on across your Mac, then helps you write better from any app. Press Option in a text field to draft replies, summarize threads, rewrite sentences, or recall important details from your recent work without copying, pasting, or re-explaining the whole backstory.
This is great! How big can the context get?
I also saw you using graphs, I guess it is to increase accuracy. So how is the memory layer structured?
Goldfish
@vugar_javadov the context can get pretty large, but we keep it scoped rather than just dumping everything in.
The rough structure is:
1. Immediate screen/app context from the current field
2. Recent activity from the last few minutes/hours
3. Longer-term memory summaries and searchable snapshots
4. A small identity/voice layer for how you usually write
The graph is mostly there to connect people, projects, apps, threads, docs, etc. so the memory retrieval is more accurate than plain keyword search. When you press Option, Goldfish pulls the relevant slice for that exact surface and intent, then ignores the rest.
@vugar_javadov yeah exactly. graphs are one part of it, but the main thing is scoped context, current surface first, recent exact activity second, and longer-term memory after that.
curious what your best use cases have been so far? 🐠
PixFit
Hey Ben, Goldfish looks like a genuinely clever take on ambient AI. How does it handle context switches between work projects or clients that have very different tones?
Goldfish
@mbertone911 Thanks Marco! It uses the focused field and nearby context as the anchor first, then pulls in only the relevant recent memory for that surface. So a client email, a dev task, and a Slack reply each get treated as separate writing situations, with different tone and context.
Have you tried it? What do you think? 🐠
@mbertone911 yeah, exactly. we try to keep it anchored to the current app/thread first, then only pull in broader memory when it actually helps. curious what use cases have been working best for you so far? 🐠
The challenge with AI memory is often trust and relevance. How do you prevent important details from getting buried as more conversations accumulate?
Goldfish
@harini_mukesh This is basically the core design problem for Goldfish.
We don’t treat memory as one giant chat history. It’s split into layers: the immediate screen context, recent activity, longer-term summaries, and entity/project-level memory. When you press Option, Goldfish first anchors on what you’re doing right now, then pulls in only the memory that looks relevant to that surface.
Important details should also become easier to retrieve over time, not harder. So instead of relying on raw conversation logs forever, Goldfish distills repeated people, projects, preferences, and commitments into more structured memory.
And there are controls for excluding things you don’t want remembered, because trust only works if the user can shape the memory.
@kar_re This makes a lot of sense. The layered memory approach feels much more practical than treating everything as one long chat history. User controls for memory seem like a must-have for trust.
Goldfish
@harini_mukesh exactly, user controls are non-negotiable here. People need to be able to exclude apps, domains, and sensitive surfaces, otherwise the product starts feeling creepy instead of useful.
@harini_mukesh yes, exactly what Kaspian said. We keep memory useful by ranking what matters right now, not just what exists somewhere in the past.
What have been your strongest use cases so far? 🐠
Finally someone built contextual AI without shipping my data to the cloud. “Memory stays on your Mac” sold me immediately. Tired of pasting Slack threads into ChatGPT. The MCP integration for Claude Desktop is genius too. Curious how you handle performance with 1k+ founding members indexing everything locally?
Goldfish
@oreofe_oluwatipin1
Thanks Oreofe! Performance is one of the main things we obsess over. Goldfish only indexes the context it can read locally on your Mac, and we keep retrieval focused around the current surface, recent context, and the people/projects involved rather than trying to shove everything into every request.
With 1k+ founding members, the biggest lesson has been that relevance matters more than raw memory size. Recent visible context usually wins, older memory helps with tone and recurring work, and everything stays local so the privacy model does not change as usage grows! :-)
One big bet is that the context will become more relevant in the future as well, so even though we cannot use the context to the max today, we definitely will in the future.
@oreofe_oluwatipin1 yeah, exactly! Local first for the win, people don't want their sensitive data anywhere else. Local is the future!
what are your best use cases so far? 🐠
This looks really interesting. My English is not very good, so I couldn’t fully follow the demo video. Subtitles would help a lot, especially for users like me in China. Funny enough, Goldfish is helping me write this sentence right now haha. I’d still love to try Goldfish and share feedback.
Goldfish
@xiya_ai thank you for the heads up, that’s super useful feedback. Subtitles are definitely a good point. Really cool that Goldfish helped you write the comment too haha 🐠
Subtitles will definitely come in the upcoming updates!
@xiya_ai so cool that Goldfish helped write this haha. and yes, subtitles are a great point, we should make the demo easier to follow for non-native English speakers.
would love for you to try it and tell us what works best for you 🐠
@joel_edholm I’ve tried Typeless before, and I think their onboarding is very simple and easy to follow. Just a small suggestion for Goldfish: could there be an option to skip the intro video?
Goldfish
@xiya_ai There is a "skip onboarding"-menu setting when you start it! :- D
Just got set up. Onboarding was a bit clunky but overall really novel and refreshing. Love it so far. Interested to see how useful this is after a few weeks of use. The option key doesn't seem to work with Atlas browser.
Goldfish
@goldhaxx Thanks for flagging this, Zach! Atlas is likely blocking or exposing the focused field differently than Safari/Chrome, so we’ll take a look and add a fix if we can. Appreciate the comment - we're looking into increasing stability as we speak! 🙏🐠
@goldhaxx yeah Atlas support is useful to know, we’ll fix it asap.
what are your best use cases so far? 🐠
Can you speak a little more to the privacy element? Goldfish seems awesome, but some clarity on just how secure the private memory layer is would be reassuring - especially for a little dude on your screen who's reading everything that you're reading!
Goldfish
@oscardog Totally. Privacy is the part we care most about getting right.
Goldfish memory stays locally on your Mac in a local database. There is no cloud sync and no backend where we can browse your data. You can pause capture whenever you want, exclude specific apps or domains, and delete stored history.
For AI calls, we use zero data retention. The goal is that Goldfish can use the context needed for the task without turning your work into something we store or inspect. We do collect basic telemetry, but only for product usage, like how often features are clicked or messages are sent. Never the content itself
@oscardog totally fair. the short version is: we’re designing it so you stay in control of what it sees, remembers, and uses + the data collected can't be seen by anyone else but you + AI calls folow Enterprise level standards with zero data renention
curious what your best use cases have been so far? 🐠