Show HN: I built a tool to create AI agents that live in iMessage | Hacker News Reader

Show HN: I built a tool to create AI agents that live in iMessage

tryflux.ai

54 pointsdanielsdk32 comments

Hey everyone, I made this thing: https://tryflux.ai/. Here's a demo video: https://screen.studio/share/1y2EnC26

Context: I've tried probably 15 different AI apps over the past year. ChatGPT, note-taking apps, productivity apps, all of it. But most of them are just clutter on my iphone.

They live in some app I have to deliberately open. And I just... don't. But you know what I open 50 times a day without thinking? iMessage. So out of mild frustration with the "AI app graveyard" on my phone, I built Flux.

What it does: - You describe a personality and what you want the agent to do - In about 2 minutes, you have a live AI agent in iMessage - Blue bars. Native. No app download for whoever texts it.

The thesis that got us here: AI is already smart enough. The bottleneck is interaction. Dashboards get forgotten. Texts get answered.

This was also my first time hitting #1 on Product Hunt, which was surreal.

It's still rough and probably broke something. If you try it, feedback is super welcome, weird edge cases, "this doesn't work," or "why would anyone use this" comments all help.

That's all. Happy to answer questions.

Open articleView on HN

This seems to be a wrapper around LangAgents that can be linked to iMessage. I would not say the agents "live in iMessage," rather that they interface through iMesssage.

You can find a bit more on what's (likely) behind this in the github repo: https://github.com/photon-hq/flux

(NB: the deployed version may not be exactly the same as the code in the repo)

I couldn't figure out where to go from the homepage. Eventually I spotted the grey-on-black GitHub link but that took me to an organization page, not a repository.

Here's the repo for anyone else who didn't find it: https://github.com/photon-hq/flux

Key detail from the README to help understand what's going on:

  Now text +16286298650 from your phone to interact with your agent!

So photon are operating a currently free relay service at fluxy.photon.codes (that's the address that flux talks to). You register your own phone number, then any time you send a message from your number to their +16286298650 they pass it back to your agent.

... but that means your agent needs to stay connected and running on a server somewhere. That's what this command does:

  flux run --prod

It stays running and maintains a connection to their relay and triggers your agent code any time you send a message to their number.

Based on https://github.com/photon-hq/imessage-kit my best guess is that Photon achieved this by running a Mac server somewhere that scripts iMessage via AppleScript against their own Apple ID account that owns that +16286298650 number.

Question for danielsdk: does your paid enterprise plan involve you running a Mac that's signed into a separate iCloud account that's assigned the phone number that your paying customer wants to use?

Thank you, that summary answered all questions I had, in the order I had them.

I feel like Apple makes this very hard and against their ToS right or in a gray area, aka shut it down at any moment. There are businesses that solely try to manage provisioning of verified apple accounts and having worked with a bunch of them reliability is pretty hard.

I tried this and the flow is sketchy UX even if it’s probably not malicious. It opens Messages, has you send a code to a number they control, and that “verifies” you. Probably to multiplex many chats on one phone number. That’s not a takeover by itself (you’re not forwarding a bank/Apple 2FA), but it’s sketchy. It also doesn’t work. It said hello and stopped replying.

I want to know what this is and what it does before I give it permissions.

Hi! I'm sure this is very cool, but want to know what this does before I try it. I typed in the chat: what does this do? Will you build me an agent behind a phone number I can chat with? And it tried to open up iMessage which I denied because IDK in the first place what this is.

I asked it to build me a weather machine and it's responding to me like a teenager.

AI: Authentication successful! Go back to the webpage to continue.

Me: Hey!

AI: yooo whats good!! u got any weather stuff u wanna chat about or just saying hi?

Me: how’s the weather in New York City

AI: ohh nah i don't have real-time data so i can't tell u the actual conditions rn! but if u check like weather.com or just search "nyc weather" u'll get the live stuff instantly tho if u tell me whats goin on there rn i can get HYPED abt the atmospheric conditions!! like is it humid?? what's the pressure doin??

So, about as useful as a teenager. Source: I was a teenager once.

Less useful than a teenager. A teenager today would just look at their weather app and tell you the answer.

probably uses some free model, like shapes.inc bots we had to discord couple months back

The most awful thing is that you can’t punch it in its face for that reply, due to its not having a face.

I asked it to build an app to track water consumption and medication. It can’t track anything. After logging “water consumption” and “taking medication” I asked it how much water I’d had and when I’d taken medication. It replied “mhm thats whats up” (like that - lowercase and no apostrophes)

Is that expected behaviour?

The tiresome bro-speak is very irritating but I’m maybe not the target audience.

united has free messaging on their flights and i found myself using meta's ai chatbot via whatsapp a few times. it's system prompt forces it to ask me inane questions at the end of every action so i would be happy to switch to something else.

just tried making one using this and it initially was way too casual for me and i didn't like how it split responses into many messages but it's very responsive to my prompts to change its style.

I made an interface to a Perplexica instance over Telegram (which I use constantly through the day) a few months ago, but actually haven't used it much as honestly the quality just isn't as good as Perplexity. It does seem the logical next step is to switch to something with more raw agent-like access.

Having worked with mobile messaging and AI and trying to wrestle with iMessage and WhatsApp because I'm pretty sure this goes against ToS as they monitor how many new contacts and messages you're sending and get flagged to be shut down... If this is not the case, it would be very interesting....

This is the path forward, we need more protocols and systems to help integrate the future rather than more apps! Excited to try it out.

> protocols and systems to help integrate the future rather than more apps

iMessage is kind of the opposite from what this is implying though, right? It is by definition an app (and extremely closed one at that).

I would think that regular SMS or even just email would be more inline with "protocols and systems".

(I have dabbled with a similar thing for myself using my twilio phone number over regular SMS)

I agree with you, iMessage ubiquity obfuscates the fact that it is also a client app and we would be much better off with an open protocol behind it. Would that then be RCS instead?

Surprised this got a second chance after hitting the front page almost a week ago; good luck!

Thank you!

Does this use an official Apple API or...

It uses Photon!

Really nice landing page!

Tysm!!

Thanks! I designed and built the landing page. Had to defend it pretty aggressively internally, lmao, glad it was worth it.

SOB now i am going to be texted by some AI trying to scam me . Fantastic!

Flux lets you build an agent that only you can text. I'd be interested to learn how you plan to scam yourself.

There are many things named flux. It’s not a big deal for one more, but maybe a more unique, descriptive name would differentiate.

Yeah, I agree. Thanks for the suggestion!

[flagged]

Can you please not post like this, especially in Show HN threads? It's good to let people know when your attempt to use their software runs into problems. But please do it in a helpful way, not a dismissive way.

From https://news.ycombinator.com/newsguidelines.html: "Please don't post shallow dismissals, especially of other people's work. A good critical comment teaches us something."

It's also good to remember that there are special rules for Show HNs in https://news.ycombinator.com/showhn.html:

Be respectful. Anyone sharing work is making a contribution, however modest.

Ask questions out of curiosity. Don't cross-examine.

Instead of "you're doing it wrong", suggest alternatives. When someone is learning, help them learn more.

When something isn't good, you needn't pretend that it is, but don't be gratuitously negative.

Reply on news.ycombinator.com