Phoenix.new – Remote AI Runtime for Phoenix (opens in new tab)

I am a long time php dev who has been interested in learning elixir/phoenix for a while but never quite motivated enough.

I saw this and thought, if this doesn't get me to give it a go, nothing will.

Less than 45 minutes after signing up for fly.io, I have a multi-room tic tac toe game deployed.

https://tic-tac-toe-cyber.fly.dev/

I had it build the game, opting for a single room at first to see if that worked. Then I had it add multiple rooms on a different git branch in case that didn't work. It worked great.

I learned very little about elixir, phoenix, or deploying to fly.io up to this point, and I already have a nice looking app deployed and running.

I know a lot of devs will hate that this is possible, and it is up to me now to look at the steps it took to create this and really understand what is happening, which are broken down extremely simply for me...

I will do this because I want to learn. I bet a lot of people won't bother to do that. But those people never would have had apps in the first place and now they can. If they are creating fun experiences and not banking apps, I think that is still great.

You guys have been releasing amazing things for years only to be poorly replicated in other languages years later.. but you really outdid yourselves here.

I'm blown away.

edit: is there a way to see how much of my credits were used by building this?

Phoenix creator here. I'm happy to answer any questions about this! Also worth noting that phoenix.new is a global Elixir cluster that spans the planet. If you sign up in Australia, you get an IDE and agent placed in Sydney.

This is very cool. I think the primary innovation here is twofold:

1. Remote agent - it's a containerized environment where the agent can run loose and do whatever - it doesn't need approval for user tasks because it's in an isolated environment (though it could still accidentally do destructive actions like edit git history). I think this alone is a separate service that needs to be productionized. When I run claude code in my terminal, automatically spin up the agent in an isolated environment (locally or remotely) and have it go wild. Easy to run things in parallel

2. Deep integration with fly. Everyone will be trying to embed AI deep into their product. Instead of having to talk to chatgpt and copy paste output, I should be able to directly interact with whatever product I'm using and interact with my data in the product using tools. In this case, it's deploying my web app

I’m torn on “is this the future”.

I worked all day on a Phoenix app we’re developing for ag irrigation analysis. Of late, my “let’s see what $20/mo gets you” is Zed with its genetic offerings.

It actually writes very little Elixir code for me. Sometimes, I let it have a go, but mostly I end up rewriting that stuff. Elixir is fun, and using the programming model as intended is enlightening.

What I do direct it to write a lot is a huge amount of the HEEX stuff for me. With an eventual pass over and clean it up for me. I have not memorized all of the nuances of CSS and html. I do not want to. And writing it has got to be the worst syntactic experience in the history of programming. It’s like someone said people lisp was cool; rather than just gobs of nested parentheses, let’s double, nay triple, no quadruple down on that. We’ll bracket all our statements/elements with a PAIR of hard to type characters, and for funsies, we’ll make them out different words in there. And then when it came to years of how to express lists of things, it’s like someone said “gimme a little bit of ini, case insensitivity, etc”. And every year, we’ll publish new spec of new stuff that preserves the old while adding the new. I digress…

I view agentic coding as an indictment on how bad programming has gotten. I’m not saying there wouldn’t be value, but a huge amount of the appeal, is that web tech is like legalese filled with what are probably hidden bugs that are swallowed by browsers in a variety of u predictable ways. What a surprise that we’ve given up and decided the key tools do the probabilistic right thing. It’s not like we had a better chance of being any more correctly precise on our own anyway.

Ah man I'm really happy to see this and excited to try it out.

As an Elixir enthusiast I've been worried that Elixir would fall behind because the LLMs don't write it as well as they write bigger languages like Python/JS. So I'm really glad to see such active effort to rectify this problem.

We're in safe hands.

Burnt through some credits pretty quickly. The first few minutes of using it felt like a glimpse of what it was supposed to be like, but otherwise it spent a lot of time not getting basic things right with the UI and before I knew it I was done. Roughly 90 minutes for $20 it's not a good value when you can ostensibly have the same experience on your computer and have control over every aspect. I still can't clone the latest revised version of the codebase it created to my local computer. Between that and Fly's non-existent documentation, no usage meter of some kind, and the lack of unpaid support (even though I am paying to use the service?) makes me want to avoid Fly, which is unfortunate because I think it does a lot of things right, especially the tunneling and dev experience outside of phoenix.new.

I've wasted a lot of time and energy on stuff that doesn't matter, so I can hardly judge anyone else on what they focus on, but man does it feel bad to have community leaders actively focus on building out tooling that is anti-worker. I think the only way I'd feel more conflicted is if Fly.io started building weapons systems for the military. I guess that wouldn't be shocking considering some of their lead's beliefs.

This is very neat, and right up my alley as both someone really into Elixir and who thinks agentic AI is the future.

I have a question about how you manage context, and what model you use. Gemini seems the best at working with give context windows right now, but even that has its limitations. Thinking about working with Claude Code, a fair bit of my strategizing is in breaking down work and managing project state to keep context size manageable.

I'm watching the linked video and it's amazing seeing it in action, but I'm imagining continuing to work on a project and wondering if it will start losing its way so to speak. Can you have it summarize stuff, and can you start a session clean with those summaries, and have it "forget" files it won't need to use for this next feature, etc?

"Sign in with fly.io" takes me to a page asking me to pay $20 but the plan details are vague - what exactly is included in "$20/mo of Built-In AI Assistance Builds, refactors, and debugs right in your IDE"?

Phoenix.new looks powerful, and I'll definitely play with it.

I've been daydreaming of an agentic framework that maximally exploits BEAM. This isn't that, but maybe jido[0] is what I'm looking for.

0. https://github.com/agentjido/jido

I'm a little surprised by the sentiment here that LLMs don't do well with Elixir. I've had a pretty good experience using AI tools on Phoenix/Elixir side projects.

Hmm, just signed up to check it out but no trial just "$20/mo of Built-In AI Assistance" without any mention of usage limits?

it's wild to me all this progress with AI, but at the same time, on my brand new mac, going through the readme to try out phoenix, it can't even get past the first step without an error (says it can't find postgres), yet the docs say I don't need postgres and it will default to sqllite if postgres and mysql can't be found.

hard to put confidence in AI vibe hacks when the basic stuff just doesn't work.

* (Mix) The database for Myapp.Repo couldn't be created: killed

I am probably doing something wrong but I hate agents for coding. I like the autocomplete and the prompt to generate snippets but when it starts modifying code in many files so fast all at once, tries a bunch of stuff and never know when to stop it pisses me off more than anything else. Because most the time if it had stopped and let me do the 10% last part it would have been actual legit code.

Mindblowing! This is 100x VB6. The generated UIs are beautiful and professional. I am still trying it out and building an app for tracking expenses but it is working very well. The conversational dialogue it has with the developer is just fantastic. I am amazed at how clear the user experience was in Chris' mind. I am not sure what LLM is being used but this is better experience than any LLM. Given this is first version, I look forward to what comes next!

Few issues:

1. The 150 message limit is understable but it suddenly pops up and you lose significant work. I was working on UI mockup and just as I had finished and was ready to go on implementation, this limit appeared and significant part of my work was lost. 2. After the first credit, the credit seems to exhaust pretty fast which makes it expensive, especially when you are trying it out. 3. Also I don't understand when you ask it to prototype different screens, why does it overwrites the same file. 4. It is not able to stop to seek user feedback but keeps trying different approach which kind of exhausts the credit. It would be nice if it describes its approach, so the human developer can provide their feedback. 5. It seems it is using OpenAI because it is often self-congratulatory to the point of being annoying sometimes.

I love the idea of Phoenix and server side rendering (I happen to work on SkyMass, a related project).

This is a tangential comment and should not detract from what Chris and team have created. I think closing the loop between agent and the running output is a great/critical step forward.

However, I find using AI to build transitional Apps with a UI is a bit like improving the way automobile steering wheels are made. In a world that soon won't need steering wheels at all.

If the AI is so good to write the code for an App, how much longer before you won't need those Apps in the first place? And then the question is, what will fill the role that Apps play today.

This looks really cool, but I gotta say I'm a bit uneasy with the apparent(?) closed-source + hosted + branding. "mix phx.new" is the way to generate a new Phoenix project, but "Phoenix.new" is closed source Fly.io product for building Phoenix projects?

Feels like we're getting into a weird situation if LLM providers are publishing open source agentic coding tools and OSS web app frameworks are publishing closed source/non-BYOK agentic coding tool. I realize this may not be an official "Phoenix" project but it seems analogous to DHH releasing a closed-source/hosted "Rails.new" service.

This is very cool! I will say though, my spidey senses kick in when looking at stuff like this, and make me wonder about how much I'm going to get vendored in here. Could I use it develop a site off fly.io? If the answer is no, then I'd say this is cool and I think useful for people who need something quick and simple and out of the box, but not something I would ever use on a serious production project.

The mental models of Elixir/OTP and AI Agents are very compatible. I’ve felt for a long time that it would be one of the best platforms for building AI agents.

So is Phoenix.new a Fly.io product, or just under the fly umbrella? Also, is pricing clearly laid out anywhere (including what the additional costs are for permanently deployed/hosted services that arise from Phoenix) ? Didn't dig too hard admittedly, but wasn't obvious where to find or look for pricing information on the front page on mobile

Any chance of open sourcing the model instructions for this? Do you feed it all the Phoenix/LiveView/Elixir docs, or have you written more specialised instructions?

I find Claude to have quite a bit of problems trying to navigate changesets + forms + streams in my codebase, just wondered if you had any tips of making it understand better :)

A bit sad that the language and framework so enjoyable to write an read will be mostly hidden in a coding box.

And thinking about it made me realize that soon there will be a completely different programming language used solely by coding agents. ChatGPT gives an interesting take on this, "The fundamental shift is that such a language wouldn’t be written or read, but reasoned about and generated. It would be more like an interlingua between symbolic goals and executable semantics, verbose, unambiguous, self-modifying, auto-verifiable, evolving alongside the agents that use it").

This makes me very uneasy. Not in what it is per se, but in what it shows about the direction of Phoenix.

I've been working with Phoenix a lot the last few months, and I like it a lot. But I do get the sense that the project suffers from wanting to perpetually chase the next new thing, even when that comes at the expense of the functional elegance and conceptual cohesiveness that I think is Phoenix' main strength.

LiveView is a great example. It's a very neat bit of tech, but it's shoe-horned surprisingly awkwardly into Phoenix. There's now a live view and non-live view way to do almost everything in Phoenix, and each has their own different foibles and edge cases. A lot of code needs to work with both (e.g. auth needs to happen at both levels, basically), meaning a surprising amount of code needs to have two, nearly identical variants: one with traditional Plug idioms, and then another using LiveView equivalents. Quick little view helpers end up with either convoluted 'what mode am I in?' branching, or (more likely) in view-mode-dependent wrappers around view-mode-independent abstractions. This touches even the simplest helpers (what is the current path?) and becomes more cumbersome from there. (And given the lack of static analysis for views, it can be non-trivial to even find out what is and isn't actually working where.)

Not every website should be a live view (e.g. hiking directions, for example), but that is clearly the direction of travel in Phoenix. Non-live views get the disparaging moniker 'dead views', and the old Phoenix.HTML helpers have been depreciated in favour of <.form />-style live components. The generators depend on those, plus Tailwind, Hero Icons and (soon) DaisyUI, all fetched live from various places on the Internet on build. This tight coupling to trendy dependencies will age poorly, and it makes for bumpy on-boarding (opinionated and tightly coupled isn't necessarily a smoother experience, just a more inflexible one).

So with all of that in mind, while I'm not shocked to see Phoenix jump on the vibe coding hype train, I guess I am disappointed.

The revelation that AI is now writing PRs for Phoenix itself is not confidence inspiring. I rely on frameworks like Phoenix because I don't want to have to think about the core abstractions around getting a website to my users; I want to focus on my business logic. But implicit in that choice is the assumption that someone is thinking about those things. If it's AI pushing out Phoenix updates now, my trust level and willingness to rely on this code drops dramatically. I also do not expect Phoenix' fraying conceptual cohesiveness to get any better if that's the way we're headed.

Phoenix is still an amazing piece of tech, but I wish I felt more at ease about its future trajectory.

I love everything I read about Phoenix. It's definitely the framework I'm going to use the next time I need to scratch an itch.

Having a full stack that is easy to use as a learning sandbox is incredibly helpful in that regard, so this looks amazing.

This is incredible. It does seem quite expensive compared to Zed or Claude Code now it's on Pro. But neat enough I've burned through the $20 subscription credit despite being a bit of an AI sceptic. This seems to have a much better handle on UI design (unless I'm missing something with the other agents), but as a solo dev I'm becoming quite convinced. It's also got me to try out fly again.

I couldn't get Tidewave working but I must try again to see if Tidewave with Claude Code would offer this level of awesome.

ps. @fly - please let me buy more credit, I just get an error!

Hey @chrismccord, very confused but this is a collab between you and the FLY.IO project right? Like I can't eject the app from this and run it myself? This isn't an open source Phoenix project?

Chris is a hacker’s hacker.

Technically this is really. Practically I don’t understand the point. Who is the target demographic? People that don’t have a local dev environment?

Beautiful demo! How do I build agents like this?

Does anyone know any great resources to learn how to design agents? Tool agnostic resources would be awesome.

This is great. I had to back out of a phoenix project and rewrite it in Django because I couldn't get good AI assistance. I'm pretty inexperienced with Elixir and Phoenix but understand the benefits enough to want to make projects in it. So this is really cool.

This is really cool! And, that you did it in a few weeks is insane.

How did you get VS Code embedded in your app? I'm aware of projects like Monaco, and that vscode.dev exists - so it's clearly possible - but I didn't realize it was something others could build upon?

Again, kudos!

What LLM does Phoenix.new use?

What model is it using for all the agentic work?

What usage limits do we get with the $20 monthly price?

Thank you!

I really wish I could move to those nice new editors, but as a vim user I just feel paralyzed when I cannot use vim bindings. And all "emulations" I tried are just incomplete.

Phoenix needs an ActiveRecord-like database abstraction layer. Many Rails developers try Phoenix at some point because they may need better performance. They’re so accustomed to the Rails structure that they assume Rails has done everything right. However, Ecto and ActiveRecord are two very different beasts. When Rails developers try out Ecto, they often feel there’s too much boilerplate and believe the Rails design is much more intuitive. This, I think, is one reason Phoenix struggles to attract Rails developers. If it can’t please Rails users, it will rarely appeal to others.

My primary concern with Phoenix is how bad it performs in LLMs and I basically went back to Node/React/Rails for LLMs.

This is very exciting and I’ll check it out!

Why wouldn't LLMs work as well with Elixir as python or JS? LLMs don't parse an AST.

Very cool. Does it use Phoenix 7 or 8 in the video?

Integrating AI assistance directly into an Elixir IDE could boost productivity, especially for newcomers. Excited to see how remote SSH and local workflows develop!

For us SaaS Founders, GrowthHackers, Product folks, us non-devs.. What are some solid use-cases for it today?

Is the site down?

this is the excuse i needed to revisit phoenix-elixir after a decade

ty fly team

Well of course I have to test it.

Disabled on news.ycombinator.com

Heyy! Now we know why fly.io was shilling for AI a couple weeks ago.

"They are not making money off AI".was the most common response to my pointing out they were shilling AI. Feels good to be right.

I'm surprised they are investing into this. I checked Phoenix recently because I was interested in LiveView and there isn't even an official AWS SDK for Elixir.

Honestly doubt the AI stuff is going to move the needle much if you can't even have a dependable S3 client.