It's an exciting time to be a developer working with Autodesk technology.

For my entire career at Autodesk, I've been a platform guy at a product company: while much of AutoCAD's success, over the years, can be indirectly attributed to the existence and use of a rich set of APIs, Autodesk has largely (and this is of course a gross generalisation) approached the exposure of APIs as an afterthought, or at least a secondary activity. Some teams have been better than others at building a platform and then a product based on top of it – a shining example is the Object Modeling Framework, upon which AutoCAD Architecture and MEP are based – but there are many cases where APIs have simply had to be retrofitted. Revit's a great example of a product that fits into this category: it was definitely not architected as a platform when the company was acquired by Autodesk, but over the years the Revit team has done an amazing job of turning that around, building a robust platform for external developers.

But the times they are a changin'. When the Forge initiative was announced at AU2015, I won't say I was skeptical, but there was a certain caution to my optimism. Was it even possible for a "big" company like Autodesk – with the inertia that comes with having 10,000+ employees, worldwide – to make the shift to focus on delivering a new web-based platform and a set of products based on it? And manage to keep delivering value in our desktop software, in the meantime?

There have been examples of big companies making such radical shifts. Apple's possibly one example, although their shift relates to going from selling computers to mainly selling music players and mobile phones. Amazon is of course a more relevant example: evolving beyond a marketplace for books (and now more or less anything) to providing the world's most popular cloud computing platform. Amazon's shift was in many ways driven by a top-down mandate… in late 2011, Steve Yegge – an ex-employee of Amazon who was trying to get colleagues at his then (current?) employer, Google, to buy into a holistic platform strategy – accidentally posted a hilarious rant for the whole world to read. It describes how – while, in his opinion, Amazon gets a lot wrong – one thing Jeff Bezos got exactly right was his 2002 mandate for the engineering staff to focus on delivering platform capabilities. This was the moment of conception for AWS. (Looking back at my posts from 2011, it turns out I wrote about this at the time.)

But does Autodesk – which doesn't have this kind of top-down culture – have the chops to do something similar?

In recent weeks I've become convinced it does. Part of this is seeing the energy the company is putting into organising the upcoming Forge DevCon, but a larger part is seeing how our various engineering teams are getting behind the Forge initiative. And not only to deliver services through it, but to use it for their own products.

Something you may not be aware of: Autodesk employees see more services when they log into forge.autodesk.com than external developers do. The delta will hopefully disappear (or at least reduce), over time, but for now there are quite a few services that haven't yet gone through the process required to make them fully public.

Let's take a look at the APIs I see when I log in using my Gmail address:

Forge (external)

And here's the set I see when I log in with my Autodesk email address (I've blurred the specific APIs, but this should give you a sense for the fact there's lots of good stuff coming down the pike).

Forge (internal)

Autodesk is really "all in" on Forge. There's a long way to go, but I do believe that by the time of AU2016 we'll really have a lot more to talk about (beyond what we'll be talking about in 6 weeks in San Francisco).

So yes, it's a great time to be a platform guy at Autodesk, or to be a developer working with Autodesk technology. Forge on!

10 responses to “Stoking the Forge”

  1. Kean, I've always said I care more about autocad as my operating system than windows (or DOS back in the day...). Here is something I am wondering though: Autodesk has only developed one big product - AutoCAD - that I can think of. All the others like Revit, LandDesktop (for civils), navisworks, and so on, have been purchased. My observation is that Autodesk does much better by buying things already molded by an industry, then polishing them. When they buy ideas, then do the implementation, things have not gone so well. Civil3D is the one our industry struggles with, now they jump to infraworks as if the problem was the interface. Its not, the problem is program objects that do not contain our design data. They got the "mp3" standard half wrong, and will not go back and fix it. So much money left on the table. The forge effort you are describing is sounding like Autodesk is deciding on the implementation of an idea. I wonder if they will also change their current practice of not maintaining consistent relationships with key industry people, in the specification writing stage of their products. Its sad, as there is so much talent, but hopefully we do not learn of new great products as they are 90% coded. Just being honest.

    1. As ever I appreciate your thoughts, James.

      Some might argue that large companies innovate by promoting a healthy ecosystem and engaging in effective M&A activities. That's certainly one way to go, and Autodesk continues to acquire technology and expertise on a regular basis from outside. Some of our acquisitions have been great, others less so: you have to tolerate losses along with the wins.

      But anyway - I digress. Autodesk has shifted gears in recent years: we've gone from being a "fast follower" to (arguably) a technology leader. Which means our R&D investment is going to be different.

      We'll see how Forge plays out. One thing I see is that Autodesk believes it critical to the company's future. Which is great for us developers.

      Kean

  2. friendfromarea51 Avatar
    friendfromarea51

    Hey Kean, Is Forge strictly built for online technology, or the API in general? I'd hate to see Autodesk create an entire system for cloud programming while ignoring the majority of us who develop for other reasons, sending us to random forums.

    FYI: not sure Apple or Amazon were good comparisons...both sale products, many that can be returned when defective.

    I do hope the programs success.

    1. Hi Brian,

      It's the online platform we're building. We're still building non-cloud solutions, but over time you're going to see more and more capability delivered online.

      I agree that Apple isn't the best choice. But I'm not talking about Amazon as a whole, rather Amazon Web Services (AWS).

      Cheers,

      Kean

      1. friendfromarea51 Avatar
        friendfromarea51

        Glad to know the off-the-grid(ers) aren't abandoned (yet)...excited to see what we can do though! Thanks a bunch 🙂

  3. Looking at the pics, this reminds me of the spark project thing. I'm still not getting what problem autodesk is trying to solve on all this. The idea of sharing 3d data in remote servers is good, but the creation of that data in autocad, inventor, 3dsmax, maya...is all desktop stuff. On top of that, 3d printing is still polarized to hobby fdm or expensive glued powder or laser cured resin. Sending stl files does not need a fancy cloud interface. The big thing coming is VR viewing of the models. I guess I am saying Autodesk has already solved the hard part - making the models, though the progs run local. Can you comment on what the gaps are Autodesk wants to fill?

    1. I expect the same people are involved in the branding and web-sites for both platforms, hence the similarity.

      I think there are actually lots of "big things coming". Like all technology companies, Autodesk makes investments in various areas. The logic of these "bets" (let's call them that) isn't always obvious (even in hindsight, as some of them don't pay off).

      On the specific case of 3D printing, do you really think consumer 3D printing middleware is as good as it can be? How often do you see failed prints, for one reason or another? How good are the standard supports - do you ever break your model trying to remove them? That's without even getting started on lock-in to 3D printing consumables...

      Kean

      1. James Maeding Avatar
        James Maeding

        I have built three fdm style 3d printers, after the prusa I3 design, but with several mods. IMO, the problems to be solved on these stepper motor driven printers are 1) no motor position feedback incorporated into the system. 2) The geared extruder design should be belt. 3) the bed leveling must be attended to by hand, as software leveling degrades quality. 4) use of pla printer parts is no good for an enclosed printing area that you heat for abs. 5) the hot ends are hard to inspect if you suspect a jam. 6) there is no precise flow control for the hot plastic, it oozes a bit always. Several more minor ones, but I would claim this: If you do a good job on a stiff frame, rail alignment, and level bed, you will get very good prints. Even the cheapest arduino/RAMPs bord, steppers, rails, bearings...from ebay work incredibly well. What I see commercial printers doing is making cute cases, with minimal improvement in real design. I think everyone that buys these things quickly realizes they all work the same, but the parts on commercial ones are only available from the vendor. I don't think there is hope of a consumer fdm printer that is "press a button and forget" You have to constantly keep an eye on cleaning the bed from last print, leveling, and plastic flow as the print starts. Anything odd, and you stop it and fix, start again...There is a huge community doing thses printers, yet I have seen little Autodesk participation, or anything offered from them to help so far. I even had to make my own civil3d surface to stl prog since the civil3d surface to solid chokes on large surfaces. Anyway, I wish they were more involved, as I'm sure they could offer a lot.

  4. Hey Kean, can I query a model using Forge? For example, I have set up a client including a 3D-view of a specific model using the Data View API. I want to be able to fetch the GUID:s of the selected model elements in the Revit model. Also, I am interested in the geometry of a model element. Is this information possible to get somehow?
    Best regards, Joel

    1. Kean Walmsley Avatar

      Hi Joel,

      I don't know if it's currently possible to query the model data outside of a viewing session... you might want to check in with the ADN team (via the Cloud & Mobile DevBlog) to see if they can explain how.

      Regards,

      Kean

Leave a Reply

Your email address will not be published. Required fields are marked *