Show HN: EchoStream – A Local AI Agent That Lives on Your iPhone

Thank you for your feedback, and I apologize for the repeated posts. You're absolutely right to call that out. I've been having an issue where I couldn't see my submissions appearing on HN, so I kept reposting thinking they weren't going through. I've now contacted the HN admins to understand what was happening, and I won't repeat this mistake.

The real purpose is that the iPhone native Agent , purely on your device. and we think that we could use more edge devices do more. that's we are focus and research for.

That's really cool! I just checked out HackYourNews.com and it's a great implementation of the summarization concept. I love that you're covering multiple sites beyond just HN, and including the comments summary is smart - often the real insights are in the discussion threads.

It's interesting that we both saw a similar need around staying informed without spending hours browsing. Your approach with Gemini Flash for the web-based solution makes a lot of sense, especially being able to offer it for free.

Our focus is a bit different - we're exploring what becomes possible when you move that intelligence directly onto the device itself. So while you're solving the summarization problem really well in the cloud, we're asking "what if this kind of AI agent lived natively on your iPhone?" The privacy, speed, and offline capabilities open up some interesting possibilities.

Thanks for sharing your project - it's always great to see different approaches to similar problems. The fact that multiple people are building in this space suggests there's real demand for better ways to consume information.

And, We will remove the paid feature in the later versions. Cause this agent is running on your device, there is no need to have any pay.

Very cool. Some ideas;

- would love to collapse all the news outlets at top and keep closed - would love a way to customize the wrap ups below that you have - would love the comments summary below

Yes, this is like a kind of flow - how to say - a kind of mind stream, replacing me thinking of any connections between any newest news and my memories.

As a person, we only need to listen to some insights and decide what to do. Others, let the ai agent do it. We think this may be the future in personal AI area.

Actually, imagine your device(not only the iPhone but also other edge devices) constantly building these connections in the background - when you read something new, it automatically surfaces related things you've seen before, conversations you've had, or notes you've taken. Instead of you having to remember "wait, didn't I read something about this topic last month?", the agent just presents those connections.

It's like having a research assistant that never forgets and can instantly connect dots across everything you've ever consumed or created on your device. The human stays focused on the insights and decisions, while the AI handles the memory and pattern recognition.

We're currently only available on the App Store in the US while we're in early development.

You could try the TestFlight version if you're interested in testing it out - that should work in more regions.

Our brains summarise all the time.

I can't remember exactly what texture/shade is on the ceiling of my bathroom, but I remember roughly what colour it is. If you asked me if there's a bit of pain flaking off and where it is, I couldn't tell you to save my life.

When I open HN (or most things) I will skim down the list of titles and it's keywords that will jump out at me.

When I read threads I will start skimming when the conversation evolves into endless debate, unless I'm interested in said debate.

Processing every possible bit of input in full would not make for a very efficient or worthy experience, but of course this is contextual. Someone a contract should read every word; though this has been weaponised in the form of long ToS - they know most people don't read those at all.

[dead]

In my knowledge domain, that's actually possible on iOS, but it will cause more privacy issues for sharing data between different apps, and also the clipboard issue. Apple has been pretty restrictive about inter-app communication for good reasons, but it does limit these kinds of automated workflows.

As you mentioned with that situation, I think Apple and Google may have better power to optimize the OS system if we only do it on the local device. They have the deep system-level access that would make this kind of seamless automation work without the privacy and security trade-offs that third-party apps face.

We're still developing this product for tech leaders and startup founders. We're also working on AI products for edge devices. Expanding to other regions would create too much pressure.

Just set up for opening to Australia, it may take some time to wait.

Yes, we’ve just made our service available in India.

1. llama.cpp can run on both Android and iOS devices. 2. For non-transformer models, Mamba and RWKV are also good options. 3. Additionally, you should have a better understanding of embeddings, tokens, and the structure of transformers.

And on Apple Silicon chips, MLX is preferred and is often the best choice.

Is it possible to download the app based on your App Store account's region, may be your account is registered outside the U.S.?

Also the app's age rating is 17+.