Apply HN: Cadwolf – Intelligent Engineering | Hacker News Reader

Apply HN: Cadwolf – Intelligent Engineering

15 pointstheuttick10 comments

www.cadwolf.com

When an engineer/scientist/worker solves a problem, they generally use Excel, Matlab, MathCAD, or python to do their work. They then document their results in Word and email that to their boss. This causes problems in that documentation and analysis almost always get out of sync, and bosses/customers will have a license to check your work.

CADWOLF solves these problems by doing all of these things in the browser. Users create web pages called "documents" and add blocks to the document through a gui. These blocks can be text, tables, plots, equations, etc. Equations solve top to bottom in the document and update in real time.

In short, it replaces all of the above platforms with a web page - with nothing to download or install.

For engineers, the platform does considerably more. Whenever a large structure - like a space vehicle - is built, every part is given a unique number through a part numbering scheme. Our part tree module lets the user create this scheme and then add systems and components to the structure.

This gives us two big functionalities : the ability to query the structure for items like "weight" in real time; and the mathematics will update along with design changes.

What would we do with the grant:

Complete an AWS server side to solve large problems, publicize the site, and begin working on a CAD system to go along with the mathematics. This will be unique in that it will be browser based and linked to a backside mathematical model. As the design changes, the math and the CAD will update as well.

This will drastically reduce the time and manpower it takes to design cars, space vehicles, planes, etc.

The system is in beta now and you can see several videos and demos on the main page.

Short intro video here - (2 min) https://www.youtube.com/watch?v=pdFf7jfA5EE

Longer video giving full explanation (15 min) https://www.youtube.com/watch?v=MkdSZhrsxrU

Thanks to everyone that has asked a question. I wanted to add a little more info on how this will speed up engineering in the future.

A "fitting" is a small component that usually attaches a rod to another item. The fitting consists of a base plate and two vertical extrusions. There is a bolt at each corner of the base plate connecting it on that end and an eye bolt sits between the two vertical extrusions and screws into the rod.

I CADWOLF, you will be able write a single web page or "document" that takes in the load on the fitting through the rod and then calculate the base plate size, the thickness, the size of the vertical extrusions, and the sizes of the bolts needed. These items will populate your part tree automatically. When the load changes, everything updates automatically. If you change this document, the design updates.

Now imagine that you are building a large structure with dozens of these fittings and any engineer can simply place the fitting into their structure in the same way that you call a function. You don't have to worry about whether or not the engineer did the design properly as everyone is using the company standard analysis and the engineer can go about doing the more complicated things.

Now imagine that CADWOLF provides a large array of built in structures like this where a single engineer can design large and complicated structures faster than teams of engineers do today.

I'm no domain expert here, just a random HN user. But I think this looks cool though, is trying to solve a real problem.

My first thought: This is a very big problem space to be tackling all at once. Massively huge.

I could see it taking a lot of engineering time to build out a featured product, which has parity with the incumbents at the things they do well. Also your target customers are very process oriented, so you've a big challenge to sell to them. For any given engineering application, you're probably going to be missing some features or feature-richness they consider 'key', for at least the medium term.

Unless something has changed that means they can achieve their needs with a lot less functionality? E.g. Spreadsheet software, by virtue of the inherent power and simplicity of the metaphor, turned out to be good at a lot of things, and killed a lot of special purpose apps. Even then it took a lot of resources to write an early spreadsheet package. And its really difficult to build horizontally successful products like that.

So the common pattern to tackle this is to pick a really narrow niche you can specialize in and dominate, within the wider market. Ideally a new or underserved niche, or one that has changed recently in some important way, which means it isn't being served optimally by existing solutions. And getting enough better than an incumbent in just that niche that you can get some traction, get funding, bootstrap your development, and [>5-10 years later] eventually build the more general product.

Have you thought about doing that? Its not always the best thing to do, but it should be given a lot of thought?

You are very right in that a product in this space has to meet a very high bar before it is useful compared to other things like Matlab.

I have functionality to do things like differential equations, linear algebra, FFTs, but not really hard math things like symbolic integration. So, right now my targets are users that need something more powerful than excel but don't need the higher level features of matlab or python. To help acquire those people, I am writing a server side on AWS to let them solve large problems and simply pass the cost they incur on to them (for now). In this way, I can target small companies and let them solve some large problems without having to buy large servers.

I am hoping that this side can pay for the development of the CAD system.

How does this integrate with existing tools?

> This will drastically reduce the time and manpower it takes to design cars, space vehicles, planes, etc.

I don't understand how this follows. Can you give an example of the savings your software will offer?

How well do you understand this market? Have you worked in it before?

To me, it's obvious that writing things in a word doc and emailing it to my boss sucks. But do you know why it's done this way? Are there better options without traction? Is there something about accountability that I'm missing?

Hi,

I am an aerospace engineer that worked on the Orion capsule in Houston as part of the constellation program. I have a BS and MS in aerospace engineering from the University of Texas at Austin.

We use MS Word mainly because it's there, everyone has a license for it, and everyone just accepts that we use one tool for documentation and others for analysis. We can change that.

The time savings comes this way:

Right now, you design your components in Excel, Matlab, etc and then document them. When a requirement or neighboring system changes, each engineer must then go back and make the necessary changes to their math, their documentation, and their CAD. This can take weeks or even months for small changes as each engineer has to go back through their work.

With our system, the user designs his component as part of the whole system. When a requirement or neighbor changes, the software filters those changes out and the components redesign themselves. This will be true for the CAD as well. Weeks or months of work for dozens of engineers or more then takes seconds.

There is no integration with existing tools. There are some very good CAD systems out there - Catia, ProE, etc. However, for security reasons, there is no way I could link up to that software on the server.

so, you can't import g files or excel files or word files? and Have to design everything from within this new framework?

The platform has 4 components. One component mimics a folder tree and shows all of you documents and subfolders and stored images.

The second is the documents themselves. Here, you can create a document and add text, equations, plots, etc.

The third is what we call "datasets." A dataset lets you import an excel file, a web page, a text file, or just paste data and store the parsed results. You can then import that dataset as a variable into a document.

The fourth is the part tree module.

Everything other than the datasets is done from within the framework. You can paste text into a document if you like and work with it like a word document.

Interesting concept. The aggregating of documentation and analysis together reminds me of a Jupyter notebook. If you add in a fully-featured CAD plugin and connect everything to AWS or whatever, does that approximate your product plan?

Pretty much. The CAD will be fully browser based and connected to the documentation and analysis side as well.

There are a lot of things that are somewhat similar at first glance. However, many of those require the user to write a document of some kind in a text editor and then upload it to a server where it is viewed through a URL as a web page. With CADWOLF, the user creates a new page and edits it in a WYSIWYG format.

Reply on news.ycombinator.com