Hi, I'm the creator of Bento. Just wanted to share a bit more about how I created it beyond what's in Github.
The file contains more or less two sections. There is a plain block of JSON near the top of the file which is the slide data. You can read, grep, or point a harness at it. The app itself is in a base64 blob that loads through a small shim which deflates in the browser with DecompressionStream, which keeps the package small and so that we don't need to fetch any external files at runtime. File System Access API is used for JSON writeback into the same file, which falls back to a plian download and all updates are ECDSA signed.
I started with reveal.js which served as a base, and incorporated GSAP/Flip to handle the animation. Then I added charting with echarts, but due to a number of issues with size and how the worked, I ended up re-implementing both.
What I'm most pleased about is how seamless the CRDT works. The blind relay is a small file that runs on Cloudflare Durable Objects, and all it sees are the encrypted data from the clients. Collab is off until we turn it on in the app (by starting a session by sharing by invite file), and then some sets of keys are generated. Access is handled by user key, and you can have read-only users, and also revoke live collab access by user as well.
The term "powerpoint" has been genericized in the common vernacular to mean any electronic slide show, like kleenex for tissues and sellotape/scotch tape for transparent adhesive tape.
It might not have reached legal generic status but just about any working professional will understand what you mean if you refer to "making a powerpoint deck" even if you don't use the actual Microslop product.
This was my biggest question. Thanks for preemptively answering (and being proud!).
When I was playing with this, I immediately thought of TiddlyWikis, which I've always been fond of, but they never took for me. However, an easily editable presentation -- that's a use case I can understand.
I couldn't see how there was a CDRT when I'm just working with a single file. I was trying to figure out where the server was. So -- this is just on for everyone? Are you concerned that your Cloudflare account will be hammered with this?
My project for the past few years is called AppDeck, collection of self-contained HTML files that know how to save themselves back to the AppDeck server, which is just WebDAV. AI has been enormously helpful in building out a suite of apps that can do this. So far I have a diagram editor that can produce SVGs, a multi-document editor that backs everything with markdown, but supports Kandan boards, documents, logs, and calendars. I also have a work flowy clone that I've really been enjoying organizing ideas with at work. There's a gallery app that inlines all of the images and can display them full screen in a light box and with a timed slideshow that also supports captions and encryption.
All of it just runs out of a folder on disk and can either be deployed locally or on a server behind a login. I intend to open source it, but I'm not quite proud enough yet of some of the implementation aspects to do so. Hope to do it this summer though!
You can play around in the guestbook. I'm in it right now and you can see me editing things.
Cloudflare Durable Objects are pretty cheap and I'm using it like a very basic pub/sub with each deck having it's own DO. I did some calculations and it should be well within my spend limits.
Something super small that I've noticed in the guestbook: since there're a lot of people connected, I noticed that every time I received an update, I lost the focus of my text element, so it was a little bit difficult to write something.
It may be annoying if two users are working at the same time in the same project.
Yeah, upgrades are pretty seamless so far as the JSON format for the slide data is pretty forgiving. You can also export and reimport the data. It's in the split pill next to Save.
This is amazing. I also think this will become more common. I feel there is a lot of software that can just be served locally via html/typescript/react etc and even have local state. There's just not been an economic incentive till now.
I am personally using kimi k3 to create a mobile friendly set of games, starting from math like linear algebra to quantum physics - neurodivergent friendly, completely local state, served from github pages - to get someone interested in the field. This needs an economic incentive to exist without LLMs, because of so much different expertise needed, and time.
if anyone's interested in the games let me know - i plan to play it myself, improve it, remove bugs etc before publishing on HN.
Yeah, I was just discussing with a co-worker the concept of a hybrid document where editing controls and the content all exist in a single HTML file (prior implementations include various wikis such as: https://feather.wiki/ ) --- makes one wonder what an SVG or spreadsheet implementation might look like.
My M1 Mac froze in the end (had to hard-reboot), but that was so much fun! I guess there’s a reason Figma uses WASM and a custom renderer, though I guess your implementation should work fine for most cases (i.e. not when all of HN are trying to edit everything simultaneously :^)
Two notes:
1. It would be nice if another person changing something unrelated to what I’m doing didn’t reset the focus for me.
2. WAY too llm-y copy both on the landing page and the example deck. Your HN submission text is way easier and more pleasant to read.
Thanks! I had to reset the guestbook because it's being spammed. It's on epoch 4 now, hopefully people will be better behaved moving forward!
1. Yes, that's a good point that somebody else raised also. It went pass testing as I was only testing with 1 or 2 other people at once. Will fix this tonight.
2. Yeah, agreed and fair point. But I don't write well and I don't have time between day job and being a dad. Will try to fix this gradually.
1. I’ve added the ⌘V remark on the first slide (it was the only way I could add anything in the previous epoch at all!). Hope it helps others have fun in the meanwhile.
2. Absolutely fair point, and it’s okay – the demo speaks for itself :-) Although if you wrote that submission text yourself, you do write pretty well! But obviously no pressure here.
mdwiki [0] would probably qualify. Disclaimer, I created+released it in 2013. I always jokingly called it the first true single-page application - as it was a single, offline-capable .html file with all images, icons, styles etc. inlined in the same .html file.
Ilograph diagrams can be exported to standalone HTML files (like this example: [0]). Coincidentally, they also have slides (kind of) and clock in at around 560KB.
Thanks for raising that. It might be in the standard cloudflare/actors SDK I used for Cloudflare Durable Objects. I'll try to disable this behavour and if we couldn't, we'll rip out DO and replace it with something that doesn't phone home. Apologies I wasn't thorough enough here.
Just an update to this phone home issue. I haven't found anything showing that this should be normal behaviour. I'm thinking this may be due to cloudflare challenges due to the traffic spike. In addition to the DO, the whole website sits behind a Cloudflare proxy.
Will keep chipping away at this and update as I learn more.
Because they didn't write all the code by hand? Seems like an odd measurement for determining if a project is ambitious or impressive. Clearly this one is based on the replies here.
Looks very good and I'm excited to try it out properly. It's a shame about the name considering a lot of self-hosters will be using BentoPDF. But thats one of two hard things in Computer Science.
I've been reading HN much longer than I've had an account, and I have never once left a comment or interacted with any post. Had to break the seal to say great job with this. Really fantastic out of the box, and timely thinking. I'll be using this for sure.
Was using Reveal.js which worked well because it can use any HTML or any plugin.
Bento works with a predefined schema, so you cant use any HTML library (Such as Mermaid.js) if you need it in your presentation.
But then again, often you dont need it.
Just created a presentation with Bento... I do think Reveal.js gives you more raw power. But I like bentos single html page design, and you can easily edit it manually. You cant do that with Reveal.js
Yup, I'm trying to see if we can add arbitrary JS. I think it may be possible, but would take a bit longer to test out different approaches. However I guess the main target audience for this is coders that interact with non-coders when doing presentations, and I think currently this approach is sufficient.
I recently did something similar with just asking claude to create me a presentation with GSAP animations. With a some nice features to embed apps, switch the language while presenting, etc. - it works like a charm: https://eos-lci.gitlab.io/eos-lci-presentations/klimabuendni... (all open source)
I like it. I used Claude and Qwen to build single file HTML decks in the last few months. The nicest thing with HTML decks is that you can not only ship the decks but integrate your prototype/demos in the same file integrated into the flow of your deck. Would love to see a feature like that.
Thanks. That would be a cool feature, and definitely possible. Currently it's already got a feature where you can export and import it's data JSON. We just need to find a way to squeeze a whole frontend in.
This blew my socks off. Theres something so satisfying about knowing everything I need is in a single file. I know the caveat is *1. Have a modern browser*, but this is my go to format when trying out random ideas. I even have a macro called `try` that sends me directly to a new folder called `scratch/YYYYMMDD_<try_bento>`.
Great work! I hope to see more and more apps like this in the spirit of "Toss it all into context"
Very cool project! It looks nice and the technical aspect of how sharing/CRDT is implemented is a really great idea! I would definitely be interested in the ability to change the relay; ideally to something self-hosted.
Thanks! That's a really good point. I used Cloudflare DO as it was elegant and convenient, but no reason why we can't run a small daemon to handle the same. I'll work on this later in the week.
This is so cool! When I initially saw the title I was wondering how this is different from SlideJS or similar, but this is really cool and now I'm wondering if this could (at least in theory) be integrated into something like Pandoc as an output option?
Thanks! I'm not familiar with pandoc but I think you can get some data out of the Bento JSON but is not going to be very rich. May be easier to render it to PDF (which works well in Chrome) first.
Thanks! I guess there are some drawback to this approach, large companies especially would be completely adverse to it. But it's cool and I'm hoping people would find it useful!
For people clicking on a link to some slides that someone shared, it seems like starting out in viewing mode with a prominent edit button might be better than starting out in edit mode?
Some unexpected behavior: I reloaded the page, double-tapped on the first slide to go back to it, then hit slideshow. Instead of starting out on the slide like I chose, it went to a slide further in othe presentation. I think it’s a bit too persistent about resuming where you left off?
Also, swiping left to get back to the first slide puts me back in the editor (I went too far), and then hitting slideshow puts me back where I started so I have to swipe left again.
Also, viewing slides on a tablet, I don’t see a way to open speaker notes since I can’t “press the S key.” These instructions should probably be different if you’re not on desktop.
Maybe display speaker notes by default if the screen is in portrait mode?
The slideshow button is a split pill, you can open it there where there is this tiny up arrow that opens up a drop down menu. But on tablet it probably doesn't do what you want it to as it was built for multi-screen use on the desktop.
I'll have a think on this tonight and I'm sure we can do a tablet speaker notes mode.
I’ve honestly been kind of surprised that large companies software hasn’t been recreated for free en masse from the use of AI in a weekend. I’ve definitely seen some cases but it’s not as common as I had personally expected.
Thanks! I started on this last week in my spare time and I used Claude, all my commit history is on Github. I did most of the coding via Claude Code, no fancy system or exotic harness.
I would love to do more by hand, but I look after a tech team inside a renewables firm, so I don't have much time between job and my preschooler.
this project is a monument to the HUGE failure of the Workspace team at Google to enable agent editing in a way that makes sense.
Google is really shitting the bed in all the way that matters on GenAI and being an ex employee of Cloud I am not in the least surprised about it, just really sad.
Why does the homepage mention AI over 20 times? Why does it mention AI at all? Why is most of it absurdly long and obfuscated base 64, which looks be binary data?
It's really to solve a use case where there are a lot of these cool web-based presentations being created by AI, but they are all disposable things that non-technies can't work with easily.
Base64 is paired with compression to keep the size small so you can share it easily. The original version was around 1.3MB and maybe we can have a switch to output without compression. The source that generates it is all in Github.
Because whatever you like or not, AI support these days is a must. Nobody who makes PPT does everything by hand nowadays. And the project is on GitHub, you can get the full code there if you must...
Thanks! It's a project I did in my spare time and it's made with a lot of MIT licensed JS libraries, so it's natural to make this MIT licensed as well. I don't know even know how anybody would start to monetise something local-first like this in this day and age.
Did you chew on monetization at all? I'm building something similar but for bio-imaging scientists, where I've packed everything I can into a single HTML (and even some modern segmentation models like Cellpose3 and StarDist that fit under Cloudflare's 25MB free tier limit haha.
Maybe I'm thinking too far ahead but I'm hoping maybe I can cover the (future?) compute costs by charging for the heavy duty ML that would have to be run non-locally but IDK. This isn't critical yet but would love to hear your musings.
If you have a backend that does processing, I'm sure you can monetise it. But for Bento, the whole idea is to forgo the backend, so I didn't really think about it that much. If an enterprise version comes, I'm sure there is a way to integrate stuff like SSO that you can charge for.
The file contains more or less two sections. There is a plain block of JSON near the top of the file which is the slide data. You can read, grep, or point a harness at it. The app itself is in a base64 blob that loads through a small shim which deflates in the browser with DecompressionStream, which keeps the package small and so that we don't need to fetch any external files at runtime. File System Access API is used for JSON writeback into the same file, which falls back to a plian download and all updates are ECDSA signed.
I started with reveal.js which served as a base, and incorporated GSAP/Flip to handle the animation. Then I added charting with echarts, but due to a number of issues with size and how the worked, I ended up re-implementing both.
What I'm most pleased about is how seamless the CRDT works. The blind relay is a small file that runs on Cloudflare Durable Objects, and all it sees are the encrypted data from the clients. Collab is off until we turn it on in the app (by starting a session by sharing by invite file), and then some sets of keys are generated. Access is handled by user key, and you can have read-only users, and also revoke live collab access by user as well.
Hopefully simpler formats like this will become the norm and the old ones can die.
It might not have reached legal generic status but just about any working professional will understand what you mean if you refer to "making a powerpoint deck" even if you don't use the actual Microslop product.
When I was playing with this, I immediately thought of TiddlyWikis, which I've always been fond of, but they never took for me. However, an easily editable presentation -- that's a use case I can understand.
I couldn't see how there was a CDRT when I'm just working with a single file. I was trying to figure out where the server was. So -- this is just on for everyone? Are you concerned that your Cloudflare account will be hammered with this?
All of it just runs out of a folder on disk and can either be deployed locally or on a server behind a login. I intend to open source it, but I'm not quite proud enough yet of some of the implementation aspects to do so. Hope to do it this summer though!
Cloudflare Durable Objects are pretty cheap and I'm using it like a very basic pub/sub with each deck having it's own DO. I did some calculations and it should be well within my spend limits.
It may be annoying if two users are working at the same time in the same project.
Again, very cool project! :)
Can I click upgrade and get the JSON slide deck data exported into a v2 shell?
I am personally using kimi k3 to create a mobile friendly set of games, starting from math like linear algebra to quantum physics - neurodivergent friendly, completely local state, served from github pages - to get someone interested in the field. This needs an economic incentive to exist without LLMs, because of so much different expertise needed, and time.
if anyone's interested in the games let me know - i plan to play it myself, improve it, remove bugs etc before publishing on HN.
https://phaseworld.play.naigap.com/
https://slope-rider.play.naigap.com
Three so far Repos -
https://github.com/praveer13/vecto
https://github.com/praveer13/slope-rider
https://github.com/praveer13/phase-world
My M1 Mac froze in the end (had to hard-reboot), but that was so much fun! I guess there’s a reason Figma uses WASM and a custom renderer, though I guess your implementation should work fine for most cases (i.e. not when all of HN are trying to edit everything simultaneously :^)
Two notes:
1. It would be nice if another person changing something unrelated to what I’m doing didn’t reset the focus for me.
2. WAY too llm-y copy both on the landing page and the example deck. Your HN submission text is way easier and more pleasant to read.
And huge kudos for releasing this as FOSS!
1. Yes, that's a good point that somebody else raised also. It went pass testing as I was only testing with 1 or 2 other people at once. Will fix this tonight.
2. Yeah, agreed and fair point. But I don't write well and I don't have time between day job and being a dad. Will try to fix this gradually.
2. Absolutely fair point, and it’s okay – the demo speaks for itself :-) Although if you wrote that submission text yourself, you do write pretty well! But obviously no pressure here.
I don't think WP is the right place for this. It is supposed to be a tertiary source, not a place to promote concepts you want to see established.
[0]: https://mdwiki.info
[0] https://www.ilograph.com/demo/intelligentDocumentProcessing....
One note: The home page says "Nothing phones home," but the Bento files I've examined include a cloudflareinsights.com beacon.
Will keep chipping away at this and update as I learn more.
Because while cool this is clearly heavily vibe coded.
Bento works with a predefined schema, so you cant use any HTML library (Such as Mermaid.js) if you need it in your presentation.
But then again, often you dont need it.
Just created a presentation with Bento... I do think Reveal.js gives you more raw power. But I like bentos single html page design, and you can easily edit it manually. You cant do that with Reveal.js
Great work! I hope to see more and more apps like this in the spirit of "Toss it all into context"
https://en.wikipedia.org/wiki/Bento_(database)
https://docs.pbs.org/space/B3/4751362/Bento+3.0
For people clicking on a link to some slides that someone shared, it seems like starting out in viewing mode with a prominent edit button might be better than starting out in edit mode?
Also, swiping left to get back to the first slide puts me back in the editor (I went too far), and then hitting slideshow puts me back where I started so I have to swipe left again.
The swipe left on first slide is a feature though, its to help mobile users get back to the editor easily.
Maybe display speaker notes by default if the screen is in portrait mode?
I'll have a think on this tonight and I'm sure we can do a tablet speaker notes mode.
- Even works well on mobile.
- Embedding images and videos also works flawlessly.
This is the best feature.
I would love to do more by hand, but I look after a tech team inside a renewables firm, so I don't have much time between job and my preschooler.
Google is really shitting the bed in all the way that matters on GenAI and being an ex employee of Cloud I am not in the least surprised about it, just really sad.
Base64 is paired with compression to keep the size small so you can share it easily. The original version was around 1.3MB and maybe we can have a switch to output without compression. The source that generates it is all in Github.
Maybe I'm thinking too far ahead but I'm hoping maybe I can cover the (future?) compute costs by charging for the heavy duty ML that would have to be run non-locally but IDK. This isn't critical yet but would love to hear your musings.