Why not just get your agent to write LaTeX? Use the beamer template and get a much more professional, customisable, git-commitable and transferable presentation.
I prefer to get bleeding edge LLMs to use old, time-hardened technologies. They are more stable, I trust they're not going to change out from under me, and it is more transferable. The ergonomics of the actual markup don't matter to me, since I'm not writing it, so I'd rather the more powerful, more training data option.
Just one example: Every university and academic publication in the world give LaTeX formatting templates, examples courses etc.
Ergonomics matter even for LLMs. Also, compile speed, quality of error messages, how easy it is to get something right (especially that LLMs are "blind" unless you want to burn a lots of tokens, so they sort of just reason without the visual output) absolutely matter and can turn a one-shot LLM prompt into a multi-turn agent churning wasting tokens for no good reason.
I'm not saying that it's necessarily the case here, but typst is past the very early phase and LLMs are decent at generating it. Also, many formatting templates have been converted to typst as well.
Yeah, the concept and execution are really impressive but lack of any import is a real blocker I think. Especially in case of spreadsheets - quite often they are not started from scratch, but require previous data or data from other sources etc.
can I do a PR for live collaboration? to be transparent, I'm going to use my own project puter.js for webrtc, but the project is open-source (https://github.com/HeyPuter/puter)
As a kid in the late 1980s, my dad got me a PC (an Olivetti IBM clone to be precise).
I distinctly remember that there were separate apps for spreadsheets (Lotus 1-2-3 [0]), charts (Harvard Graphics [1]) and word processing (Chi Writer [2]).
And now we just have a basic office suite for free in one website.
I was musing on something like this. There is non zero chance that some US administration disables GSuite / Office365 for all of us living in shithole countries.
Open street map servers aren't really meant to be used in end user applications. They are primarily to support the editing of the map data. Usage like this should be with a provider like Mapbox, or your own supplied tiles.
I've recently found out that this is not quite true - while the primary purpose of their tile servers is mapping and editing applications, they do allow third party apps to use the servers as long as they follow usage policies: https://operations.osmfoundation.org/policies/tiles/
If folks like this kind of thing, I discovered https://bento.page/ from a HN post and have been using it off and on for the past few months. It has a much larger featureset but is specifically focused on presentations (for now!). Unlike this project, the editor and content embed themselves in an html file, so you aren't dependent on the host site existing to serve the editor.
Maybe I'm a 2026 luddite but I can't bring myself to read about let alone try using products whose landing page is such a cookie cutter AI design. At least put some effort in to sand off the horrible unnecessary gloss and choose different fonts, colour schemes and layouts from every other project.
I am gradually becoming resigned to the fact that I'll have to get over this prejudice, though I'm not sure how I will.
I agree with you, but it is so funny that three years ago, a website that look like this would probably make the front page for its excellent design alone. Now, it gives off similar energy to a wix website.
You are right, then again: I think three years ago someone would have commented about the gloss in the background as well, pointing out that it’s unhelpful, distracting and resource intensive.
Landing pages are always a proxy for quality. Quality in implementation, quality in domain knowledge, in communication…
At a time when building such a page was hard, it communicated quality in implementation. Now it fakes that, and that is deceptive and therefore met with negative reviews reaction.
Counterexample: highly skilled low level engineers would end up with a crappy front page most of the time, but still the compiler/system/kernel whatever could be high quality.
That depends on how you look at it. These front pages are made in a certain way that tells me "low level engineer".
I could make other examples: websites of online shops that have been around for a while and that are just medium-sized, local businesses, tend to look much worse than your average influencer webshop. But the medium-sized local business tends to have the much better value for money ratio, and I can tell it from their website.
Design says something. AI is trying to imitate that very thing that a specific design used to tell you, and that imitation annoys people, rightfully so. That has little to do with AI per se and everything with the immitation the human behind it does.
it's a fair criticism though. learning and eventually relying on new tools is an investment. it helps to know the developer is invested too, not just moonlighting.
Not only the design but the text. Like it says so much and so little at the same time and constantly repeats the same cliches like "the file is the app" or "the api is the x". Unmemorable, unfocused and boring is what it comes across. I know devs don't care about marketing copy but it does make a difference on how you're presenting your app to first time potential users.
We had few submissions and links in comments this week to some very "young" projects and their home pages looked like pressed from the same machine.
But if it works for people who just want to share their software and avoid working around the packaging and presentation, then I guess "it is what it is".
> I can't bring myself to read about let alone try using products whose landing page is such a cookie cutter AI design.
My issue isn't cookie-cutter, but if by gloss you mean unnecessarily animated, then yeah. But I've had this issue for over a decade. AI without direction is just mimicking popular site designs that were doing this same crap before.
> At least put some effort in to sand off the horrible unnecessary gloss and choose different fonts, colour schemes and layouts from every other project.
I knew a web dev who exclusively made WordPress websites for local small businesses. They always looked great. The themes were put together by hand with care from a designer, and the developer/end user filled them and tweaked them *purposefully*.
Sure if you have some web dev experience you start spotting them, but that never turned me off from a business in the same way that the above page does. I expected every business to have a website, but not for every business to commission a wholly bespoke design system. The themes available were numerous and of high enough quality that they were fit for purpose and didn't have this nauseating effect, at least for me.
same, i've been using it extensively and it's very convenient having the ability to let Codex/ClaudeCode generate a whole deck, and then do the last edits myself in the browser.
Interesting, though on initial scan I could not easily determine who made it, why, what's the future for it, any terms or licenses, etc. With complex html and downloads, I have no idea what else may be sneaking in and/or phoning home :-/
The only bug I found is that the presentation previews in dark mode have the text as black on a dark background, when it should be white (matching the canvas text).
I’m waiting until somebody vibe code a proper desktop apps to provide more diversity than MS vs Open Office. just like I see options blossomed in the “Creative Suite” field.
But maybe I’m just one of those who don’t like web office apps and/or google docs related workflows.
I'm intrigued about the "Trip Planner" being a top-level app. Only tried it in a very surface level way, but the recent trips I've planned (Ottawa for 6 people, Cancun for ~20) have ended up as an ugly mishmash of spreadsheets + google maps saved locations list + random chatgpt sessions.
It feels like there's an unmet need there for managing something more flexible than a minute by minute itinerary, maybe like a list of must-do and want-do activities and dining options, with the system able to figure out what is open when and has other constraints like the weather, costs, what activities pair well together, which group members some things are more or less suited to, what the contact details are for things like a private tour, etc etc.
Definitely there is a need, as you say, but the Internet lacks a cross-site (DB-like) transaction capability. Hotel bookings and flight leg bookings are interdependent, and you will want to lock them in (transaction "COMMIT") once the complete path has been validated as available; if, in contrast, one flight leg or one hotel night are unavailable, the whole plan path may have to be cancelled.
Such a tool would be enormously helpful even for individuals traveling (e.g. business people or scientists visiting customers or trade fairs), which is important as this massively increases market size/potential user base scale.
At least for myself, business travel remains relatively simple: go to Perk or whatever the business-supplied travel portal is, put in the dates, book whatever flight and hotel are in-policy, when I get there report to the conference/customer/whatever, and buy whatever food is available when I get hungry.
It's travel for pleasure (especially with groups) where all the complexity comes in: there's a lot more stuff to schedule, more slop in what exactly happens when, more competing preferences, more need to have multiple things on deck so you can react to changing energy levels, etc.
Not too hard to see why a lot of people just give up and book an all-inclusive where every day is the same and everyone can just "do whatever", but with better tools, maybe rolling your own vacation could be made more practical and coherent for a normal busy person to put together.
Holding flights and hotel bookings like that is possible although any kind of online travel agent doing it at scale will not last very long. Travel is still pretty relationship focused when it comes to inventory and if an OTA tries to do things that create churn on that inventory then it will get flagged pretty quickly.
An aggregator or AI assistant like Muse might have better luck although the mechanism for those will require charging whatever credit card you give it. For flights, it should work just fine in the US. There is a mandatory fully-refundable 24 hour cancellation period on all flights booked more than 7 days in advance through US point-of-sale, or they have to honor a 24 hour hold without payment but not both. Hotels get a bit dicier since the policy there will be dependent on each hotel.
You would rather trust a package installer and/or binary executable that might inject malware like a keyboard logger, bitcoin miner, back orifice, etc?
I would prefer using sources and making that mandatory.
But I prefer trusted binaries over "private" web app in open internet. And even locally - html and all layers of http stack is inferior to pixels. Or vectors.
183 comments
[ 0.26 ms ] story [ 12.8 ms ] threadEdit: realized it looks like the data is stored in the url, which is a neat approach but I guess its not really that close.
Permalinks are needed to share documents, which would be something like a link shortener, with the target updating as the document is updated.
Here's a deck I had it generate about how the URL format works: https://bit.ly/4yRBc1v
Who cares?
https://postimg.cc/0r4jrF59
Personally I like to keep my LaTeX tooling local so it can integrate with everything I already use, and I own simple plaintext files.
Just one example: Every university and academic publication in the world give LaTeX formatting templates, examples courses etc.
I'm not saying that it's necessarily the case here, but typst is past the very early phase and LLMs are decent at generating it. Also, many formatting templates have been converted to typst as well.
That last slide with a QR code that purports to embed the entire document/deck is intriguing.
I presume there are practical limits to the file size here?
And images inside slides must be impossible to fit it right away?
It's a great idea, but this limits the size of the serialized document to 2048 kiB for Chrome, and 64 kiB for Firefox.
I distinctly remember that there were separate apps for spreadsheets (Lotus 1-2-3 [0]), charts (Harvard Graphics [1]) and word processing (Chi Writer [2]).
And now we just have a basic office suite for free in one website.
0 - https://en.wikipedia.org/wiki/Lotus_1-2-3
1 - https://en.wikipedia.org/wiki/Harvard_Graphics
2 - https://en.wikipedia.org/wiki/ChiWriter
This recently came across my radar as Home Assistant has migrated to the vector tiles hosted directly by OSMF for all deployments: https://github.com/home-assistant/frontend/pull/53816
For my own needs I usually just go for https://openfreemap.org/ though, I like the default style, and it feels maybe a little faster.
Maybe I'm a 2026 luddite but I can't bring myself to read about let alone try using products whose landing page is such a cookie cutter AI design. At least put some effort in to sand off the horrible unnecessary gloss and choose different fonts, colour schemes and layouts from every other project.
I am gradually becoming resigned to the fact that I'll have to get over this prejudice, though I'm not sure how I will.
At a time when building such a page was hard, it communicated quality in implementation. Now it fakes that, and that is deceptive and therefore met with negative reviews reaction.
I find it completely logical.
Always is a very strong claim.
Counterexample: highly skilled low level engineers would end up with a crappy front page most of the time, but still the compiler/system/kernel whatever could be high quality.
I could make other examples: websites of online shops that have been around for a while and that are just medium-sized, local businesses, tend to look much worse than your average influencer webshop. But the medium-sized local business tends to have the much better value for money ratio, and I can tell it from their website.
Design says something. AI is trying to imitate that very thing that a specific design used to tell you, and that imitation annoys people, rightfully so. That has little to do with AI per se and everything with the immitation the human behind it does.
Now I see your point. Fair enough
It could be the case that marketing is high quality, but the actual project low (tho, of course, this doesn't have to be the case).
And vice versa with low quality (looking) landing page.
1. https://www.cosmos.so/alfirous/websites/landing
It makes me kinda tense up - I've become almost allergic to it. Kinda interesting really.
But if it works for people who just want to share their software and avoid working around the packaging and presentation, then I guess "it is what it is".
My issue isn't cookie-cutter, but if by gloss you mean unnecessarily animated, then yeah. But I've had this issue for over a decade. AI without direction is just mimicking popular site designs that were doing this same crap before.
> At least put some effort in to sand off the horrible unnecessary gloss and choose different fonts, colour schemes and layouts from every other project.
How do you feel about WordPress / Hugo themes?
I knew a web dev who exclusively made WordPress websites for local small businesses. They always looked great. The themes were put together by hand with care from a designer, and the developer/end user filled them and tweaked them *purposefully*.
Sure if you have some web dev experience you start spotting them, but that never turned me off from a business in the same way that the above page does. I expected every business to have a website, but not for every business to commission a wholly bespoke design system. The themes available were numerous and of high enough quality that they were fit for purpose and didn't have this nauseating effect, at least for me.
and the functionality has been available already for decades. solved problems, kiddos
How is it easy to share megabytes encoded in a link?
The only bug I found is that the presentation previews in dark mode have the text as black on a dark background, when it should be white (matching the canvas text).
But maybe I’m just one of those who don’t like web office apps and/or google docs related workflows.
It feels like there's an unmet need there for managing something more flexible than a minute by minute itinerary, maybe like a list of must-do and want-do activities and dining options, with the system able to figure out what is open when and has other constraints like the weather, costs, what activities pair well together, which group members some things are more or less suited to, what the contact details are for things like a private tour, etc etc.
Such a tool would be enormously helpful even for individuals traveling (e.g. business people or scientists visiting customers or trade fairs), which is important as this massively increases market size/potential user base scale.
It's travel for pleasure (especially with groups) where all the complexity comes in: there's a lot more stuff to schedule, more slop in what exactly happens when, more competing preferences, more need to have multiple things on deck so you can react to changing energy levels, etc.
Not too hard to see why a lot of people just give up and book an all-inclusive where every day is the same and everyone can just "do whatever", but with better tools, maybe rolling your own vacation could be made more practical and coherent for a normal busy person to put together.
An aggregator or AI assistant like Muse might have better luck although the mechanism for those will require charging whatever credit card you give it. For flights, it should work just fine in the US. There is a mandatory fully-refundable 24 hour cancellation period on all flights booked more than 7 days in advance through US point-of-sale, or they have to honor a 24 hour hold without payment but not both. Hotels get a bit dicier since the policy there will be dependent on each hotel.
But I prefer trusted binaries over "private" web app in open internet. And even locally - html and all layers of http stack is inferior to pixels. Or vectors.
https://aboard.com/introducing-tinylittleoffice-com/
It seems they had the idea originally 6 years ago and did versions of it, but
> We moved on, and sadly, those sites fell off the web, and the old links all broke.