I mean docs are largely written for an LLM-in-a-harness. That’s how it goes! If the LLM bootstraps with the right understanding of the universe and knows how to quickly build specific context flavors… life is good.
I love this idea. I see the AI era having 2 competing views when building something new:
1. Build X with pure <language of choice>. Why? LLMs will have less context needed, and onboarding engineers would be easier since there’ll be less overhead and opinionated frameworks knowledge required
2. Build X using well establish frameworks. Painful in the beginning since you’ll not only need language knowledge, but framework knowledge. The upshot, is scaling and maintainability
I love that this ecosystem will heavily pressure teams to consider (2) more and more — solving the very real “AI slop” problem
i like it, but I think i would rather have a proxy, or atleast an auth redirect to those different tools.
I used to have flower at myapp.com/flower using an auth redirect in nginx to a simple view in django that made sure it was an admin user. I think if you can make that setup easier to leverage existing tools that would be nicer than rebuilding everything.
Totally understand - I am a long time flower user for example, and I am familiar with having to harden that installation a bit.
What I'm aiming for here is slightly different - keeping everything inside Django so there are no extra services to run or configure or proxy. As long as you surface the admin somewhere, then that is the place to find your tooling (including celery monitoring)
There will always be room for both approaches. A lightweight proxy/redirect could be something to explore in the future.
I think even if AI handles more of the CRUD side, you still need to understand what’s happening in the system once it’s running - this is where this project fits in.
To your point about framework use because of AI: As more applications are being built because of lowering barriers, I think it makes sense for full stack monolithic frameworks to be used more frequently.
I mean for one thing your garden variety LLM had been substantially trained to handle Django. That is less context for it to bootstrap every time you summon it.
Just like rolling your shitty homebrew framework is a bad idea because only you understand it, the same is probably true with LLMs. Sure they’ll scan the bejesus out of your codebase every time they need to make a change and probably figure it out eventually… but that is just a poor use of limited context. With something mainstream, the LLM already has a lot about the universe in its training. Not to mention an ecosystem of plugins, skills, mcp servers, wizbango-hashers, and claberdashers. All there for the LLM to use instead of wasting tons of time, tokens and money perpetually relearning your oddball, one-off, rat infested homebrew framework.
I think that explains some of the value for this project a bit better
README and site were definitely optimized for speed over perfection. The panels themselves got a bit more attention.
Curious what you’d want to see improved on the docs/site side.
I like the idea it can help for initial inspection and smell detection
1. Build X with pure <language of choice>. Why? LLMs will have less context needed, and onboarding engineers would be easier since there’ll be less overhead and opinionated frameworks knowledge required
2. Build X using well establish frameworks. Painful in the beginning since you’ll not only need language knowledge, but framework knowledge. The upshot, is scaling and maintainability
I love that this ecosystem will heavily pressure teams to consider (2) more and more — solving the very real “AI slop” problem
I used to have flower at myapp.com/flower using an auth redirect in nginx to a simple view in django that made sure it was an admin user. I think if you can make that setup easier to leverage existing tools that would be nicer than rebuilding everything.
What I'm aiming for here is slightly different - keeping everything inside Django so there are no extra services to run or configure or proxy. As long as you surface the admin somewhere, then that is the place to find your tooling (including celery monitoring)
There will always be room for both approaches. A lightweight proxy/redirect could be something to explore in the future.
I think even if AI handles more of the CRUD side, you still need to understand what’s happening in the system once it’s running - this is where this project fits in.
To your point about framework use because of AI: As more applications are being built because of lowering barriers, I think it makes sense for full stack monolithic frameworks to be used more frequently.
Just like rolling your shitty homebrew framework is a bad idea because only you understand it, the same is probably true with LLMs. Sure they’ll scan the bejesus out of your codebase every time they need to make a change and probably figure it out eventually… but that is just a poor use of limited context. With something mainstream, the LLM already has a lot about the universe in its training. Not to mention an ecosystem of plugins, skills, mcp servers, wizbango-hashers, and claberdashers. All there for the LLM to use instead of wasting tons of time, tokens and money perpetually relearning your oddball, one-off, rat infested homebrew framework.
Nothing has changed really…