Crossing the Chasm with Autodesk Forge

Forge is the first of its kind. Let's hope Autodesk does it justice and quickly evolves a coherent business model.

Having now returned from Autodesk University 2017 in Las Vegas, I wanted to lay out some thoughts and observations on Autodesk Forge from the perspective of a startup looking to develop on the platform.

This was my second visit to AU, the first being when I gave my own class in 2015 about C360, an iLogic-driven configurator that I had implemented in a manufacturing business I was running at the time. Forge was then barely a twinkle in the eye of Autodesk's fortunes. 

Two years later, it is clear that much has changed. Autodesk is evidently betting heavily on its cloud-based platform, which comes at a fortuitous time for Unifize, a start-up that I co-founded at the beginning of 2017 to solve problems in CAD-based engineering collaboration.

What is Forge?

"CAD has always been about the what. What do I want to build? What [Forge] allows us to do is to build workflows that are more about the who, the when and the how much." -- Brian Mathews, Vice President of Cloud Platforms at the Forge DevCon 2017 Fireside Chat .

Computer Aided Design (CAD) is the latest in the long list of technology to be swept up by the tide of cloud collaboration. Until only a few years ago, cloud-based CAD tools were out of reach, primarily because CAD files themselves are often large and unwieldy, requiring relatively high internet bandwidth and processing speeds.  However, with average internet speeds increasing and the advent of on-demand cloud computing platforms, this dynamic is changing quickly.

The backbone of Autodesk's cloud technology is built on AWS and drives their front-end products such as Fusion360 and A360.  A few years ago, someone at Autodesk had the bright idea to expose this functionality to developers via various APIs and SDKs to create a platform they called Forge.  This is essentially a place in the cloud where you can store, view, manage, manipulate and retrieve CAD data.  It's format-neutral, meaning it can not only handle Autodesk's native CAD files, but also files from SolidWorks, Creo, Siemens, Onshape and others.

Autodesk's Forge strategy is to achieve world domination by encouraging system integrators, enterprises and startups to build on its platform, paying Autodesk a fee for storage and consumption of its web services.  The devil lies in the detail of this, as we shall see later.

"Forge is about collaboration that creates insights.  And from insights come competitive advantage, come business success." -- Jim Quanci, Senior Director of the Autodesk Developer Network at the Forge DevCon 2017 .

The Good

At Unifize, we have been prototyping on Forge for several months.  Our most important piece of positive feedback is that it actually works.

While there is certainly a learning curve, Autodesk has invested substantial resources in both its documentation and developer support.  The Forge team is highly responsive to development questions and often pre-empts our needs by pushing reusable code to GitHub. They have also organised various accelerator programs around the world, including one that we attended ourselves in Bangalore.

The result of this is that we have been able to store and convert 2D and 3D files of any format using the Data Management and Model Derivative APIs, and access these files in the Viewer embedded in both a mobile and desktop environment with access control, layering, 3D markups and commenting.

It might seem simple, but this functionality is far ahead of anything else available today.  Without Forge, you need to build code to manage storage and file conversion, along with your own wrapper on a geometric modelling kernel (that isn't designed for mobile).  This is highly complex development, and is way out of the budget of most startups or systems integrators.

Beyond this, Forge also provides developers with the ability to access vast amounts of CAD data already stored on the Autodesk cloud from platforms like A360 and Fusion 360, and to extract its geometry for analysis.  In theory, for example, you could conduct computation fluid dynamics on a model without ever having to download it, or create catalogs and comparisons of models on the cloud.  

Forge also offers a number of other existing and planned APIs that are being continuously developed and improved, including:

• Reality Capture API: enables construction companies to create 3D models of outdoor landscapes.
• Design Automation API: allows companies to conduct virtualized design operations, which until now was only available in a desktop software. This will enable automated operations in CAD on a vast scale.
• Other APIs which will get mature over time (Generative Design, AI, IOT, etc.)

Finally, there are already 1,000 in-house Autodesk developers working on the platform.  Unlike other Autodesk products and initiatives that don’t pass muster, shutting down Forge would be like conducting a brain transplant.

The Bad

While Autodesk is clearly trying to invest in Forge developer support and resources, they could be doing a lot more to build out the ecosystem.  Many of the features we needed (eg 3D markup on a mobile) had to be built ourselves from scratch.  It would make more sense if Autodesk were to invest more in providing building blocks to developers on the Forge Platform. As an example of how this might work, the Salesforce AppExchange allows users to install to tools that solve small problems within their larger CRM application.

The large existing Autodesk Developer Network seems unprepared for the world of Forge.  While knowledgeable about the API features, they seem unaware of the functional possibilities. Since Forge intends to build a world of end-user applications, it may be somewhat strange for a network of developers who, until now, have been trying to automate or integrate existing Autodesk software into business processes or enterprise tools.

Finally, we went to Adam Nagy's class on High Frequency Data Management (HFDM).  It was unclear when this important feature would be ready for general use.  Without it, managing multiple small design changes will be virtually impossible, especially considering Autodesk's current pricing model (see below).

The Ugly

At Unifize, we have interviewed multiple prospective early adopters who are generating more than 10,000 models / drawings and have margins of less than USD 20k per month.  Even after the recently announced pricing of 20 cents per call of the Model Derivative API, these numbers will mean the platform is unaffordable for such customers, even if the majority of their drawings are small 3D part and 2D drawing files.

A back-of-the-envelope calculation shows that Autodesk is likely to be marking up its AWS costs for files at between 5-6 times on file sizes of around 5 MB.  The consumption-based pricing means that the cost remains fixed whether or not the file size is decreased to 500 KB or less, as is often the case with smaller .ipt, .dxf and .dwg files.

While this is a significant improvement on their earlier pricing, the Forge business model remains terrifying for startups like Unifize.  Anyone building on the platform with a business strategy that involves uploading and displaying large quantities of .ipt, .dxf and .dwg files to Forge must either hope that their end-users are building products out of gold, or they will be out of business very quickly.

At AU 2017, we hired a booth in the Exhibit Hall to help us validate the Unifize business model, and randomly whinge about the Forge business model to any unsuspecting Forge Expert we encountered.  We were surprised that, more than once, we received the response that “you don't have to pay for the Model Derivative API if you are already using Fusion360 or A360”.  This implies a conflicted internal understanding about the purpose of Forge as a platform and how it should be used by Autodesk partners.  Is this really a 'platform' that exclusively targets existing Autodesk users?  If so, why waste time developing functionality to bring in any kind of CAD file?

There also seems to be a fundamental failure to distinguish between startups, systems integrators and enterprises building on Forge. A system integrator, for instance, will have far fewer pricing concerns when they use Forge to build a customized solution because they can pass the cloud-credit cost on to their own customers.  A startup trying to offer a fixed monthly user fee will have to bear the entire risk of its end-users' use of Forge.

The result of all this uncertainty leads us to question whether Forge is a customisable Autodesk product, or a true platform?

This assessment is likely to be skewed by Unifize’s specific business model, and that the economics of developing on Forge may improve depending upon the end-use case.  As the overall quantity of CAD files decreases (or their relative size increases), so too do the associated costs.  Obviously, the value of the material and the design intrinsic to these CAD files also plays a role in determining Forge's viability.  Furthermore, this scenario will certainly improve over time, especially after design files are stored on large, multi-tenant databases, which appears to be the direction Forge is heading with HFDM.

Nevertheless, until these Forge business model issues are resolved, developers need to be wary about building a strategy blindly around Forge, and they may need to deploy scarce resources into building defensive technical options to mitigate the risks.

Potential Solutions

Autodesk is clearly too focused on pricing simplicity at the expense of the vast range of potential Forge use-cases.  Given this, there are a number of alternatives that it could explore:

1. Create an ISV program and enter into specific arrangements with partners based on their business model and the underlying costs, and not try to squeeze everyone into a one size fits all, consumption-based pricing.
2. Adopt a win-most, lose-some pricing strategy.  This may upset the pricing hawks, but they could possibly look to support from the much-vaunted USD 100m Forge fund to absorb some of the risk until the platform use cases and costs become clearer.
3. Provide the Model Derivative API on a processing hour basis.
4. Offer an alternative, subscription-based pricing for startups.
5. Publish volume discounts on cloud credits to give startups visibility on their potential future costs.

Ultimately, this boils down to re-balancing Autodesk's stated desire for scale and adoption with its need for instant profitability.

It doesn’t require an enormous leap of faith to imagine how a focus on scale and adoption might yield better results for Autodesk in the medium-to-long term.  At Unifize, for example, our product will drive users to upload models and drawings to the Forge platform in large quantities; users that could then be upsold other Autodesk products and services.

Conclusion

One of our advisory board members calls us Autodesk groupies, an accusation we would find hard to deny if there were any other options available to us for achieving our goals. There aren't.

Forge is the best by virtue of it being the first and only solution of its kind.  This situation will change sooner or later, and the opportunity for Autodesk to do for CAD what Salesforce did for CRM is Autodesk's to lose or win.

“The two most important requirements for major success are: first, being in the right place at the right time, and second, doing something about it.” - Ray Kroc

Ben Merton is a Co-Founder of Unifize, a technology platform that brings engineering communication into one place. He is a contributor for various publications on business, technology and entrepreneurship, including the Wall Street Journal, the Financial Times and Business Standard. © Ben Merton 2017