SnapDiagram – Instantly Convert Hand-Drawn Diagrams to Digital with AI | Hacker News Reader

SnapDiagram – Instantly Convert Hand-Drawn Diagrams to Digital with AI

snapdiagram.xyz

108 pointstompreneur71 comments

Hey Hacker News,

I'm excited to introduce SnapDiagram, an AI-powered web app that allows users to seamlessly transform hand-crafted diagrams into polished, digital visuals. Whether you're sketching ideas during a brainstorming session or mapping out a quick process flow on paper, SnapDiagram can help you bring those rough drafts into the digital world, instantly and effortlessly.

How it works:

- Snap a photo of your hand-drawn diagram using your phone or upload an image from your computer. - Our AI processes the image to recognize shapes, text, and connections. - Get a clean, editable digital version that you can export in various formats, share, or enhance further.

Checkout how it works on this video https://www.youtube.com/watch?v=c2cq8B7vTXw

Congrats on building this into a working app!

But as far as I can tell, this is simply a workflow that e.g. ChatGPT and Claude have supported for quite a while now. (It's what I do with my whiteboard diagrams)

Which means you'll probably need to think of a value add to help it gain traction. I'm curious to see where you'll be taking it!

The people willing to pay 20$ to learn to use ChatGPT/Claude to accomplish this probably doesn't have much overlap with the people willing to pay x$ to have a clean web interface to do just this one task.

IDK - the "help me get stuff done" crowd seems to use LLMs fairly frequently. If I'm wrong and they're disjoint markets, more power to tompreneur. But even then, it means his moat is very thin.

(It might have been a different proposition when doing this took major prompting gyrations, but multimodal LLMs have gotten pretty amazing in the last few months)

Of course, if it's just a hobby project, it doesn't matter. But if this is in search of traction, it very much does.

You can actually do this now - I've taken a great deal of my hand written workflow diagrams, snapped a picture of them, and asked chatGPT to convert them to mermaid UML.

It was one of the more impressive demonstrations of multimodal AI.

You can also import mermaid into Google's Draw.io app if you want to reflow or change the theme.

I did something really similar by converting handwritten math to LaTeX.

This use case (and yours as well) can be really useful and works well just by using ChatGPT. In many cases people are just building wrappers of gpt-4o, but I can see some room for improvement by building on top of OpenAI’s APIs.

Even better - Excalidraw supports importing mermaid - and it has a built in AI tool where you can bring your own LLM via an API to generate diagram without even leaving the app.

an FYI-You-should-too -- I am a huge fan of telling claude to give me mermaids and swim diagrams of most of the logic and code I am iterating with.

I like to do the following:

"Review this code and give me a readme, include mermaid, swim diagrams to draw the logic."

Claude is great because it renders the mermaid in the artifact (most times).

then I can copy the mermaid into a mermaidlive (I still havent found the best mermaid site because some of them choke on the mmd output of claude and I dont know how to fix it sometimes...

for example - here is a mermaid where I was looking into how the universities, professors and families of SBF, Holmes and Ellison were tied:

https://i.imgur.com/njFKvmJ.png

Here is a mermaid of who sits on the agra committees, who funds their campaigns, and who did they make investments into wrt to the sugar lobby:

https://i.imgur.com/FGGcvxL.png

Having chatbots do mermaids is fn fantastic

I feel like we are building a very large and complex interface between human beings and the internet, so that we are becoming more like mechanical nodes on the network, where slowly the idiosyncracies of individuality are being erased. That's not unique to AI, but AI is a major component of it now. I can imagine future scenarios where every part of our communication is passed through an AI filter until we are all perfectly tuned to be mindless consumers of information.

This SnapDiagram does it for sketches, and LLMs do it for text. Soon everyone will sound very similar and the system will be able to remove any aspect of individuality that is troublesome to it with a mere parameter tuning.

The little picture is just a conversion to a more readable diagram. The big picture is the slow erasing of the human spirit.

Damn, well put!

So all the highly unique, contextually relevant and aesthetically pleasing AI art that I am seeing pop up in blogs, chats, documentation, presentations, etc where before there would have been nothing or maaaybe some ultra-generic clipart, this explosion of creativity is slowly erasing the human spirit? Nah. Looks like the opposite, if anything.

"It all looks the same" -> there's plenty of room for better prompting and models. Just look at civit.ai -- the "sameness" problem is just low skill with the new tools. The floor used to be "looks like a kid drew it," now the floor is "looks like the default StableDiffusion style." Replace with whatever domain you are interested in, I am using art as an example because it is comparatively further down the maturation curve. In any case, the ceiling is high and it still takes effort to reach, but the bar is raised everywhere.

> So all the highly unique, contextually relevant and aesthetically pleasing AI art that I am seeing pop up in blogs, chats, documentation, presentations, etc where before there would have been nothing or maaaybe some ultra-generic clipart, this explosion of creativity is slowly erasing the human spirit? Nah. Looks like the opposite, if anything.

None of that art represents the individual experience of a person any more. It's all a psychotic averaging. I've seen that art too and I find it disgusting.

I don't see anything people are producing with AI as representing the human spirit. I see it as humans being plugged into a machine pushed to create and create and for what? What's the next step? It's horrific and should be destroyed.

So someone who has lost the use of their voice and uses a text to voice program is disgusting and has no human spirit? I would disagree. In the same way new AI tools allow people who could not express themselves otherwise to express themselves. Does it mean they have good taste? Not necessarily, but it is authentic communication.

I’d argue that the person who lost its voice have no choice, the people choosing to draw by AI to express themselves do it by sheer laziness and to get quick gratification, but that seems what technology is mostly about, putting any effort in any pursuit is so old school.

> So someone who has lost the use of their voice and uses a text to voice program is disgusting and has no human spirit?

You are twisting my argument. I never said a PERSON using AI doesn't have human spirit. Rather, that in GENERAL, the content generated by AI does not have the subtelty of human expression compared to the human-made stuff. Of course, the person that lost their voice naturally loses some expression by using an artificially-generated one.

That doesn't mean that they can't express themselves in other ways.

Come on, these counterarguments are too easy to refute...

Adding an image does not automatically make a post better. I'd rather just not have any image at all than something mindlessly thrown in just for its own sake, regardless of whether it's generic or generated.

I’d argue the opposite in fact. It’s more often than not making it worse. It adds nothing and it’s a waste of bandwidth

> this explosion of creativity

But is it? It's fancier clip art, no?

It is in my opinion because it has that extremely generic feel, even more than clip art because at least the clip art had some character from the artist.

Those AI photos at the start of blogs often make me say "fuck it" and go back.

It screams "I rushed this out, I'm probably regurgitating something I just learned".

I rather a dorky stock photo than some nonsense AI image puking out some combination of "cyber", "tech", "hacker", and "circuit".

It's unserious at best, scammy at its worst.

Exactly. I stop reading any article as soon as I see an AI image.

I cringe when I see one, but sometimes the articles are good. Apparently, some people have great writing skills but suck at visual taste. Though, arguably, they aught to know that about themselves.

True, but I believe taking a strong unilateral stand is worth the sacrifice of not reading such articles.

Is that a little like cutting off one's nose to spite one's face? Good to have principles that propel forwards and not become entrenched by them, wouldn't you agree? That said, you will save, or at least not waste, an awful lot of time - there was already too much and now a lot more tripe out there since so-called a.i..

> Good to have principles that propel forwards and not become entrenched by them, wouldn't you agree? That said, you will save, or at least not waste, an awful lot of time - there was already too much and now a lot more tripe out there since so-called a.i..

In this case, it's not really entrenchment, because there are viable alternatives.

That an interpretation. The other interpretation is "Hey I'm not a talented artist. This tool allows me to express an idea I couldn't otherwise." Yes AI art is used in scams, and AI doesn't give you good taste. By I don't think it indicates laziness.

If your blog isn’t about art, why does it need the entire above-the-fold taken up with generated ‘art’?

And if it is about art, you should be showing art.

Right, because non creatives can't have nice things, and there was never a thing called conceptual art.

To each there own — I loved the AI art for about one month but since then it’s become insanely triggering. It’s everywhere now and it all has this same “feel”.

What used to just be an unsplash image that I never gave any thought to is now a serious detractor that can often overshadow the content (for me)

I’ve clicked out of articles and videos many times because of the AI art. I agree with OP that it feels like it’s washing out a lot of individuality at the moment.

I prefer a stock image as well. At least it was crafted by a human being behind a camera. And as a pro photographer, and someone who takes thousands of photos a month and views thousands more, there is something indescribably human about human works that are lacking in AI.

There is no way you could tell the difference between stock photo and AI generated art. Stock photos are created to be generic and soulless. That allows them to have the most broad applicability.

I'm extremely skeptical of that claim. Got any examples you think would stump us?

Seriously? Most AI-generated images look much faker than real photos. Real photos often have some flaws and some visual inconsistency about them.

AI image generation has taught me two things: never had it been more obvious that taste is only cultivated through a careful, studied engagement with human arts and culture and never has this been more irrelevant to the majority of people.

The American slop factory is the predominant cultural idiom and like this country’s factory farming is now an entire automated machinery for mass producing and force feeding everyone poison. I’m hoping some subset of countries will go AI art Galapagos to preserve a trace of our humanity and aspire to something beyond the median of a Google image search.

No, the human spirit does not lie in the identical LucidChart diagram I would have made in order to transcribe a whiteboard photo without a tool like this. It's a direct transcription, the exact class of AI work that is irrelevant to your broad point about the human spirit. No choice or decision is being delegated to statistical noise here. Failing to make that distinction weakens your message imo.

I disagree really, actually, because the tool offers large-scale mechanization and uniformization of human activity with relatively low cost, which is the proto form of more advanced AI.

Disagree. There are different use cases for diagrams. The first is communication. My people are poor at drawing diagrams. That means our speech in the medium is limited. Tools like the expand our ability to communicate. A second use case is artistic expression. I can't imagine why you would use such a tool for artistic expression. Nothing is being lost here.

I never said that THIS tool would replace artistic expression. But the spirit of it and all of its implications will very likely result in tools that will replace it. One cannot pick and choose when it comes to technology -- the prisoner's dilemma will result in a net loss of human expression.

I don’t see why you need a seperate app / subscription for this. Claude does this very well, as do local VLMs such as InternVLM. They generate mermaid graphs which you can use directly or import to excalidraw.

Do you think average joe knows about let alone knows how to use Claude for this?

I mean you just type “create this diagram”…

Microsoft offers "certification" for Copilot usage.

I stopped offering to build custom demos for AGI pipelines last year because people just didn't get it. April Dunford's "Obviously Awesome" pretty simply covers why, because as very much not a product-person, I frequently fall prey to "if you build it, they will come" mentality.

@Tompreneur,

I would like to learn more about the features:

- What types of diagrams does this support? E.g., could the diagrams include some cartoons, artwork, icons, besides generic shapes like ellipses and arrows like shown in the video?

- What editable formats are supported? SVG?

- Privacy policy stuff. Would my diagrams be stored in the account? (I prefer not. I like it stateless.)

I am happy to try to find above answers myself, however, do not want to create a login just for trying. Also, I prefer site-specific login systems as opppsed to BigCo login IDs like Google mentioned.

I need to digitise about 15-20 diagrams. If it would work for these, I am ready to pay $10 for these. (I've seen the pricing at $3 for more diagrams than what I need.)

Hi alok-g, thanks for your interest in SnapDiagram!

The application focuses mainly on converting simple diagrams of connected shapes with text. May work with some simple icon, but probably won't work for cartoons or complex artwork.

The editable file is compatible with https://app.diagrams.net. If you need an SVG, you can import your editable file there and export it as SVG.

The application does not store your diagrams permanently. These have a max TTL of 12hs.

>> converting simple diagrams of connected shapes with text

All 10+ AI-based diagramming solutions I see on the web are like that.

I was searching something like this to convert my maps, charts drawings in iPad to Obsidian. But 1 diagram/month is so not enough. Yes 3$ for 50 diagram/month is looks like a better option compared to the free but it's still a paid option.

It is cumbersome for me to start another paid service let alone another risk of trust I am putting myself in.

Make it free option 15 diagram/month and let people try your site fully so that who uses it that 15 diagram can switch to paid option.

Hey adamsocrat, thanks for your interest in SnapDiagram! If you tried it and fulfills your needs, we can provide you with one month free of charge of the basic plan. So you can decide if it is worth it, and help us improve it.

Curious, would you be more likely to pay for a service like this if you could pay per use with a very low friction checkout?

Counterpoint: no matter what product you ship, someone is always going to tell you that you should have a free plan generous enough that they won't need to pay for it, and if you listen to them you'll go out of business. The object of this game is to be creating enough value that people who love your product willingly pay for it, not to create just enough value that people who dabble in your product use it once in awhile.

The problem is here not free plan generous enough, free plan that lets you enough taste from the desert that you want to have the whole piece.

I mean sure I want people to clean my house and mow my lawn for me for free too but services have costs...

Nice job! This is something I would use. However, the pricing plans are not friendly for occasional users. I maybe use it 10 times a year, but the only option is a monthly subscription. I wish they also had pay-per-use plans.

This illustrates the annoying trend for software products to insist on monthly subscriptions. I understand that businesses want recurring revenue, but many users would rather pay upfront or per use.

I get your point but its $3/mo or $18/yr. If a diagram isnt worth 1.80 to you then you probably wouldnt have bought it even with pay as you go

I have always wondered if projects like this would be better off having a pay-as-you-go-model. One example:

  - Minimum account balance to start: $10.
  - $1/use 
  - Then top-off as you need 
  - no balance expiry (outside service going bust)

A credit based model with minimum add amounts like that does sound like a great alternative if people prefer usage based pricing. From what Ive seen usage pricing actually drives a lot of anxiety even if the end cost is lower

Auphonic has this model and I’ve been a happy on/off customer for 10+ years now. (They do audio processing.)

This is an outstanding first release. I hope stylizing the diagrams automatically for visual impact is on the roadmap.

Thanks for the support jaysonelliot! Yes! the first iteration of this new feature would allow the user choose from certain predefined styles.

Oh man this is a great idea. I can't count the number of times I've spent an hour in a meeting diagraming stuff on a whiteboard and then taken a picture of the whiteboard on the way out. This would be super useful for documenting the results of meetings like this.

Cool idea and implementation. Great job.

I personally hand draw diagrams and then recreate them in Balsamiq (old school).

I, myself, won't use it because there's a lot of creation and innovation that happens in the transition phase when going from hand-drawn sketches to creating the diagrams.

I want to go in the other direction: Take computer-generated math animation and make it look like Walt Disney was in his office "relaxing" while fifty artists drew each frame in a better version of my artist's hand.

This is a cool idea

I have a need for something like this. I want to convert hand written / drawn stuff to SVG. Not to a defined diagram schema though, just freeform SVG.

Any suggestions on alternative tools? For pricing reference, I would be incorporating this into my app, and users might convert 1 page a day. For what I would earn, your pricing model would be completely unusable.

I will have to build either a CV pipeline or use NNs, I imagine. Not sure which is the better approach for my task.

Credit card sign up not working. Would prefer to pay via Amazon Pay, Google Pay or Stripe given past history with Paypal.

What’s the “Editable file” format?

very cool and a very useful tool to integrate with drawio and lucidchards or visio maybe ?

Can't you do this with https://flowchart.fun?

No, you can’t upload an image- it’s just via prompt. SnapDiagram looks really useful!

One diagram per month for free.

No privacy policy link on your landing page is a red flag and will discourage users.

very cool tool.

Reply on news.ycombinator.com