andric.dev
Indie software dev.
Striving to build software that’s pragmatic, ergonomic, and collaborative.
Bootstrapping https://promptbible.ai with @ideosyncretic.com to help you better manage your AI prompts
121 posts
902 followers
483 following
Regular Contributor
Active Commenter
comment in response to
post
Sounds like a Facebook Page or Group from the early 2010s
comment in response to
post
AI will make it so we can organize around purposes more easily, because you don’t need to be an expert to operate these tools when an AI model can assist you.
Maybe this is too idealistic, but I think humans specializing around the ends and not the means leads to more meaningful work.
comment in response to
post
Yeah I agree with you there: dialogue is important. What I want is dialogue that’s contextualized with the actions I’m taking.
A lot of what we do tacitly imply what we mean. If these tacit decisions and ideas aren’t made explicit, it’s impossible for an agent to really understand you well enough.
comment in response to
post
Current “AI agents” aren’t like this at all. They do things on your behalf in ways that their developers decided they should do.
An AI agent should instead do things exactly how I would do them.
Do as I do, not as I say.
comment in response to
post
Like my personal apprentice.
Just like training an apprentice, a true AI agent should amplify your abilities to the limits of how much compute you can buy, so that you can do a lot more with the same amount of time (which is true productivity).
comment in response to
post
*understand
comment in response to
post
Statistical prediction of the sort that deep learning models do (where past data predict future data well, i.e. ergodic) isn’t quite the same as predicting socioeconomic developments, though.
The latter is non-ergodic.
comment in response to
post
The other thing is that the nature of said upside is more “downside protection”. You don’t really gain anything. Instead it prevents losses.
And the nature of this prevention almost always probabilistic.
In many ways it’s similar to insurance.
Which is hard for people to grasp.
comment in response to
post
The issue is that the upside is collective at the population level (which is hard to intuit) and the downside is at the personal level (which is more easy to imagine)
comment in response to
post
It’s a smaller quantized model (q2_k). The bigger ones don’t work on my computer.
Mostly I just want to see if I can use it for local coding. I have Perplexity Pro for chat, already
Managed to get it to work with @continue.dev (using @lmstudio-ai.bsky.social). Will use it more and see how it goes
comment in response to
post
Not familiar with it!
What does it help you do and how have you found it so far?
comment in response to
post
Oh this is a neat one!
I do a similar thing with my hand drawn notes (on an e-ink tablet) and turn it into text.
I find scribbling notes on a freeform page much more conducive to creative thinking than typing
but text is more easily searchable and readable, so this closes the gap
comment in response to
post
What other boring but useful ways to use AI have you found?
comment in response to
post
Just tried this in Perplexity Pro (with GPT-4o) and was worried it might hallucinate text that wasn’t there.
But I double checked with the PDF and it got it!
I know, it’s nothing groundbreaking. But I was happy to be able to read a PDF on my phone now!
comment in response to
post
So I prompted it:
I haven’t read this PDF. I want you to help me summarize it and walk me through chapter by chapter.
First, list all the chapters with a short summary.
Later, when I tell you to continue, start responding me with each chapter’s text verbatim. Do not miss any details.
comment in response to
post
It takes a bit of work for me to get good sleep. If is slip up on sleep hygiene the night before then I’ll sleep poorly. It sucks.
comment in response to
post
Recommender algos don’t have a concept of “quality”:
If it uses popularity as a proxy for quality, then it’s at best uncorrelated, and at worse inversely correlated
Whoever chose music for GTA radio stations should finetune the Spotify algo. Since it’s their only good algorithmic playlist!
comment in response to
post
Oh man I don’t really want to go back to WordPress (it’s definitely dated).
I’ve settled on Next.js and Typescript so it’s got to work with that. No PHP.
But if I were strapped for time and had to do a client project I’d probably pick that too.
comment in response to
post
The choice seems to boil down to:
Do I want to store my data in a database and then have to deploy & manage that infra myself?
Or do I want to keep it in Markdown files in Git but potentially tie myself to a custom engine for querying data?
comment in response to
post
There’s isn’t a consensus choice that made me go, yeah you can start small with this and scale up if needed
On the “complicated” end you have Sanity, Directus, Strapi
On the “incomplete” end you have Outstatic & Contentlayer
Then there’s PayloadCMS & TinaCMS that seem half-baked and complicated
comment in response to
post
Might be that recommending incumbent products isn’t really considered advertising.
Whereas if they play favorites with a lesser-known product, that would be suspicious.
comment in response to
post
My understanding is that Remix is still server-first.
See: x.com/ryanflorence...
TanStack is at least architected to be client-first.
I’m still mostly using Next.js with TanStack Query since it’s the most stable/mature.
But definitely keeping an eye on Router and Start.
comment in response to
post
Thanks Tanner! Glad you liked it.
I can’t help being excited about client-first making a comeback!
I continued a bit here: bsky.app/profile/andr...
comment in response to
post
Don’t get me wrong, a slow loading webpage is frustrating.
But a snappy local-first interaction and realtime multiplayer will always feel satisfying in ways that a fast initial webpage load cannot.
I want to use (and be able to build) more such satisfying apps!
comment in response to
post
Arguably apps that work this way feel faster in more profound ways than server-rendered websites that load fast on fiber Internet.
You’re manipulating data! Text! Images! Diagrams!
And these manipulations are instant, both in terms of what you see on screen and what collaborators see in real time.
comment in response to
post
With CRDTs and local-first sync engines, data is first mutated on the client, and then broadcast to online peers and reconciled with server state (which might or might not be authoritative and “roll back” data).
comment in response to
post
I’m excited about CRDTs and local-first sync engines because the way they work is different from the traditional client/server model where you fetch some data, cache it, optimistically mutate the cache, fire off a server mutation, then revalidate the cache
comment in response to
post
People talk about granular data invalidation with server components as a solution, as if the UI is just a cache of the data on the server.
When data could be an *active replica* of data on the server, that conception of how data is treated feels wholly unimaginative.
comment in response to
post
Even with these imperfect home-grown data syncing stacks, these apps are miles ahead of other app they compete with.
Server components bake data into components and make non-interactive UIs faster, but they don’t speed up interactive UIs (which includes navigation & data shared between components).
comment in response to
post
And Notion. I was surprised about them, because I’ve always thought they supported real-time collaborative rich text editing… but they don’t.
It’s all an illusion. They cheated by using last-writes-win on a block level. If you concurrently edit a block with someone else, one person’s edits will win
comment in response to
post
In Linear, the rich text portion of the app (documents) and the structured data portion (issues) use different syncing methods, and they don’t share the same undo/redo history.
The structured data uses Linear’s home-grown sync engine, and documents use Yjs.
comment in response to
post
That combination isn’t really something that exists right now.
Even if you look at best-in/class apps like Figma, Linear, or Notion, you see some cracks where the UX could be a lot better.
e.g. In FigJam if 2 people edit the same text box, one editor’s update will clobber the other’s.
comment in response to
post
I didn’t know anything of the extended universe, but if they diverged from it I can see why people might hate it. The only part I hated from BoBF was the mods.
I can see the pun they’re going for, but they felt out of place. Felt more Doctor Who than Star Wars.
comment in response to
post
Seems to be that being able to render node views as React components and a first class Yjs extension is what TipTap brings
comment in response to
post
Aye, TipTap’s comments are more for the collaboration use case.
Although you can build what you want with TipTap’s extensions API. TipTap is just a layer over Prosemirror.
Although for out of the box, this might be what you’re looking for: www.remirror.io/docs/extensi...
comment in response to
post
Change of personality from The Mandalorian, or from the original trilogy? He wasn’t shown much in the original trilogy.
comment in response to
post
There needs to be an algo that has a toggle for American politics. It should be off by default if it detects your IP is outside America.
It’s crazy how much non-Americans have to put up with on platforms that should be global.
comment in response to
post
👋 Would love to join! I’m building promptbible.ai with @ideosyncretic.com
comment in response to
post
Found this: @bookmarkit.bsky.social
comment in response to
post
🙏