A Rust Belt Native and Technologist Re-Thinking American Manufacturing
medium.com
medium.com
Like Plethora, eMachineShop has a CAD system which can tell you what's manufacturable and how much it will cost. The main difference is that Plethora uses a plug-in to Inventor or Solidworks, so they're aimed at pros who have those expensive packages. eMachineShop has a free downloadable CAD program, which understands what their processes can do and will calculate pricing. It's an impressive program, one which will prevent you from designing unbuildable parts and warn you if you're forcing an unnecessarily expensive operation.
Plethora has the San Francisco location, the cool web site with no pricing info, the guy with the neckbeard, the funding to lose money on the first order, and the emphasis on onboarding potential customers for marketing purposes. Maybe that will help them scale.
Our goal at Plethora is to make hardware as easy as software. One question we ask ourselves is why can a small team of software devs push out an app in a few days / weeks when a hardware product takes a sizable kickstarter/VC round and months / years? It comes down to:
- great tools
- fast iteration
- easy deployment
- straightforward scalability
To achieve this kind of agility with hardware, this means removing the friction at every step of the product lifecycle.
One half is in the design & ordering of custom parts. CAD packages today are powerful and it’s very easy to design beautiful parts that are completely unmanufacturable. The Plethora add-in analyzes parts in a few seconds, provides helpful hints on manufacturability, instant pricing and ordering. We went with a plugin for existing CAD programs because getting people to switch CAD programs is like asking a programmer to switch code editors - extremely difficult unless the value of the new tool is orders of magnitude greater than that of the incumbent.
The other half of the friction in manufacturing is when the part enters production. Plethora isn’t just a pretty face on a traditional machine shop - we’ve been working for years to automate the extremely manual process of converting a 3D model into instructions for a machine to produce. This automation enables quick turnaround and no minimum order quantities - which allows teams to have more iterations and faster deployment of their products.
Perhaps related, any plans for Blender plugin?
Thanks!
It's mostly driven by what our customers have asked for. Subtractive processes like CNC have a high barrier to entry (compared to additive), so we started there to make it easier for everyone to get quality parts easily and quickly. Aluminum is a great material that you see a lot in both prototype and final products across industries.
The CAD integrations, materials, and part complexity we support are constantly improving - much more to come in 2017!
Aluminium is already the material of choice for a lot of prototyping and general purpose widget making, and for good reason. It's relatively cheap, easy to machine, lightweight, strong enough, and corrosion resistant.
Automating the design-to-CNC-instructions process is a pretty logical next step.
Comparing "manufacturing" software products to manufacturing hardware products is as nonsensical as it can get. One has nothing whatsoever to do with the other.
Source: Me.
Background: I've been developing software and hardware for over thirty years. I go back to the days of the 8080 processor. Embedded, system, real time, Windows, Linux, OSX and iOS software development over the years. Commercial, industrial, consumer and aerospace environments.
At the same time I have been designing and manufacturing the physical side of said products for an equally long period of time. CNC, sheet metal, through hole and SMT electronics, TIG and MIG weldments, plastic and metal 3D printing, composites, injection molded plastics. All in markets from consumer to aerospace.
Heck I have a Haas VF2-SS VMC, Bridgeport manual mill, Hardinge lathe, various 3D printers, routers, table saw, half a dozen other saws and metal working equipment and this is only in my house for my hobby and R&D stuff.
I've been running AutoCAD since inception, Solidworks since inception, CAMWorks, SolidCAM and MasterCAM for some time as well as Siemens NX (exclusively for aerospace stuff).
So, yeah, I'll pat my own back and say I know manufacturing. Please stop this Silicon Valley "we are going to disrupt manufacturing by making CNC machining easier" nonsense.
Again, I apologize for being a bit harsh but I have to say you don't have a clue.
OK, here it goes. I can beat anything you can possible put together. How? I can send a model to China and have part back in three days flat. 5 axis. Complex. Detailed. Perfectly finished. And reasonably priced. I don't even have to know how they make it. If you work with reputable contract manufacturers out there you get great product at fair prices. It's isn't that hard. I truly don't give a shit what software they use or what machines they might own. There are great CM's which which you can do $200 to million dollar orders and they treat you just the same.
Between that and Proto Labs, Protocase, Emachineshop and others the problem of CNC machining parts and making plastic or sheet metal rapid prototypes is pretty much solved.
Now, if you want to compete with Proto Labs or Emachineshop and come up with a slicker path to CNC machined parts. By all means, go for it. But please don't say you are going to disrupt manufacturing. CNC machining is a small, very small part of most manufactured products.
Oh, yeah, did some aluminum die casting, custom extrusions and casting as well.
Sorry, I'm pretty tired of the active rejection of experience, existing industry knowledge and existing optimized processes by these Silicon Valley types who are clueless and think everything can be hammered into a software development model. Not everyone is out there trying to get people to click on yet another call to action button.
Here, I'll give you a project: Go design and manufacture a motorcycle from scratch. I mean a real product, not a one-off garage-built prototype. A real product that can be manufactured at a reasonable scale with decent DFM, documentation, certifications, component and system level test engineering, qualification, sourcing, etc.
Do this and I'll bet your view of someone asking "why can a small team of software devs push out an app in a few days / weeks when a hardware product takes a sizable kickstarter/VC round and months / years" will land you right-smack where I started this comment:
Are you fucking kidding me?
Want a business that's machining-based but needs better software? Design and make injection molding molds automatically. Injection molding is incredibly cheap per unit, but the molds cost too much (average is around $12,000) and may not work right the first time. Mold design requires highly experienced mold designers.
Make that process easier. Be able to start from 3D printer type models. Advise the naive user interactively on how to modify their part for easier injection molding. Figure out all the flow and cooling and release stuff automatically.
Love the Bridgeport. I have somehow resisted the temptation to convert it to CNC.
Another reason I love it is that I use my garage shop in mentoring a group of kids at the local high school robotics team. Nothing like making parts on a manual mill and lathe to learn about the real world of "just 'cause you can draw it in Solidworks it doesn't mean you can make it that way". One of my favorite moments to witness with the kids is when the make two parts that are to be bolted to each other and they can't. The holes a don't line-up and the hole diameters are not what they expected. What follows is a chat about Geometric Dimensioning and Tolerancing (Light Edition). Fun stuff.
That said, back when I was still bootstrapping my business in the garage I reflowed SMT boards with everything from toaster ovens to hot plates. I have to say that my favorite budget reflow tool back then was a hot plate with a really thick top. There are some that have a metal cover box that will make a little oven-like environment. None of these are 100% reliable, of course, and a lot depends on the nature of the board, size, type and mass of components, etc. What you do get is precise temperature set points.
I'll see if I can ask our manufacturing engineers tomorrow and see if there is such a thing as a cheap lab-class reflow oven that does a decent job. It isn't going to be $200, I can tell you that much.
I don't have nearly as much experience as you, but I think you make a great point. I tried Plethora about a year ago and was disappointed it couldn't make a single part I needed, and that it had such limited materials.
Although I wish Plethora the best, they do seem to have fallen into the mindset of thinking they're different when they appear to be reinventing the wheel (and making it worse). My guess is Animats has it right and they're hoping to get acquired by Autodesk. I'm surprised how many startups today are designed solely to be acquired (I doubt Plethora started with this mindset), and if they fail at that then they have zero chance of going public as they have no chance of sustaining profits.
Of course, maybe I'm wrong and they know something I don't. I always love to be proven wrong by someone succeeding.
The solutions to these problems are not going to come from sipping latte's in San Francisco while asking why manufacturing can't be more like software development.
That's misguided Silicon Valley hubris. It's an echo chamber that funds and produces tons of bullshit startups and sometimes gets lucky. When it comes to hardware they almost always seem to be getting it wrong.
This is probably because they insist on rejecting experience and filling their ranks with fresh grads who know nothing about nothing and can, within their version of reality, conclude that software will solve all problems.
Here's a hint: Want to make products that truly serve humanity (whether it's manufacturing or something else) rather than better and more clever ways to get people to click on buttons, get addicted to stupid games or social platforms?
Simple. Hire or fund people older than, say, 40 years of age who come armed with experience and common sense. Yes, give them a team of young hot-shots with the crazy ideas to try and find some of out-of-the-box stuff. Just don't setup startup after startup that looks more like a remake of Lord of the Flies rather than a real business.
You have companies like Google and Facebook who brilliantly showcase how horrible these young, socially untrained, ethically inept and often maladjusted people can be. They think it's OK to do such things as to use an algorithm to completely shutdown someone's business (Google AdSense and FB Ads) and, at the same time, provide exactly zero customer service or the ability to have a business-like conversation to try to rectify whatever might have triggered the issue. This is deranged lunatic territory and it seems to permeate some of the culture coming out of these startups. They know everything and they don't want to hear from anyone who might actually know something. What a waste of human talent.
Our supply chain is long and expensive. Start there. And by this I mean everything, from raw materials, components and sub-assemblies to tooling, equipment and consumables. The length and complexity of our supply pipe can easily double or triple a manufacturer's costs and impose very high inventory costs.
Our regulations are crippling. For example, try to get a steel weldment porcelain enamel coated. In China, no problem. In the US. Nearly impossible and definitely not even in the realm of being competitive.
Our unions have done of good job of helping kill-off industries. Union leaders (not union members, leaders) succeeded at pressing companies so hard without regards for long term viability that they eventually forced some of them out of business or out of the country.
Taxes are ridiculous. One way to look at it is how much of the year is devoted to, effectively, working for the government. In other words, in order to earn the tax money paid to the government you have to work.
Well, at a 39% corporate tax rate the entire company is working for the federal government for approximately the first 4.7 months of the year. After that they get to keep their profits.
In Ireland, with a rate of 12.5%, the people in a company work for the government for 1.5 months and the rest of the year they get to keep their profits.
So, taxes in the US means you are working 5 months to pay them vs. less than 2 in Ireland. That is horrific.
Liability and tort reform is a huge deal. As a manufacturer you are incredibly exposed to being sued out of existence (or out of being able to compete due to financially crippling lawsuits).
Labor force education and availability is becoming a greater issue every year. Skills development has stagnated over the years. We don't have a modern tech savvy workforce. Schools don't teach any of that stuff any more. As a software guy the "hour of code" is great but as a hardware guy I think the "hour of drilling a hole and cutting wood" is equally important.
So, again, there is so much more to making a product beyond rapid prototyping that at some point you have to wonder if people understand that there are far more pressing areas to be optimized, areas with far more significant impact on the bigger picture.
Elon Musk seems to have taken many ideas from the software industry to manufacturing at Tesla and SpaceX. My understanding is that manufacturing uses more of what we software developers would call a waterfall method of development. Long upfront phase with research and detailed specifications before manufacture. Elon Musk seems to have favored a more software like approach of rapid iterations and feedback. E.g. trying out more risky things, see it break and go back to the drawing board. They also seems to have borrowed the idea of platforms from the software business. Where you gradually refine an established system. E.g. I read that in comparison to other auto manufacturers they create detailed specs for all their parts and fan out the production to all sorts of contractors. These guys might not get used in the next round, in the next model. So you can't e.g. gradually refine and upgrade the software systems of the car in the same manner in traditional manufacturing. Mind you I am not doing a great job of explaining these details, if you don't know what I am talking about, better to read up other accounts of the differences between Tesla and other auto manufacturers as well as SpaceX and other rocket makers.
To be fair this isn't the same as discussed in this article which is more about what happens at the part level. What Tesla and SpaceX has changed is what happens at the over organizational level. Doing more manufacturing in house and relying on off the shelf parts rather than getting them made according to some spec. E.g. SpaceX uses seatbelts manufactured for race car drivers as their astronaut seatbelts rather than using magnitudes more expensive and custom made astronaut seatbelts.
Let me just say this. There are certain realities in every business. If you are baking cookies in your kitchen you can get away with doing things a certain way. You scoff at the guys making cookies using contract kitchens. Your ways are better. Until you hit a certain scale. Once you hit that scale you'll be the one talking to the contract kitchens and you'll have to learn and accept the way they work.
Now you laugh at the folks making cookies in their kitchens and still think the large multinational operations are archaic, old-school, slow and inefficient.
Until you reach a new scale. And now you have to make things in the millions. And the consequences of making mistakes can end your business. So you seek to learn and adopt the process the large organizations use to reliably manufacture process at a scale you never dreamed of.
And that's when you finally understand and realize things were not really as bad as you thought when you were in the kitchen. There's a "form follows function" reality to manufacturing at different scales.
Tesla can do things the way they do because their scale allows them to operate this way. And also because they are making electric cars, which are a LOT easier than gas powered cars. By switching to an electric motor the complexity of the manufacturing process changes to the extent that eliminating a couple of thousand mechanical components allows. Electric motors are simple and manufacturing electronic motor controllers is far more efficient than assembling and testing internal combustion engines.
Yet, if Tesla ever achieves the same scale as a Ford, GM or others it is very likely they will have to operate almost exactly the same way as the large companies do. You can't iterate fast when you are making a million units of a design per year. Making mistakes can cost your company. And, when that's the context, you go slow, you plan everything, create detailed specifications, recruit and verify suppliers, order things a year in advance, etc.
With regards to such things as racing seat belts vs. aerospace seat belts, well, it's a great way to feel "rogue" and fantastic for press releases. In the context of a twenty five million dollar rocket and one where the consequence of being "hip" and a maverick is that people die, things, I'll be you, are far more like traditional aerospace than not. Things in aerospace aren't expensive just to be expensive. No, they cost money because the are generally unique, they have to meet certain specification and, in some cases, people's lives depend on them. Just like in Medical.
Bottom line: You can be a maverick and cut corners when you are in the garage. As things get "real" you will end-up making pretty much all of the same decisions everyone before you made before you thought everyone was stupid and incompetent.
And that's the problem I have with a lot of these things. These ridiculous echo chambers starts off assuming everyone before them was stupid and didn't know how to innovate (or didn't care to). And that's patently false. Innovation is part of every day engineering. But it has to be constrained by limitations of the realities within which they exist.
A book titled "The innovator's dilemma" is a great read that covers part of the reason companies are not able to destroy it all and do something different. Hint: Good management is the reason companies don't adopt every possible shift in technology.
If these guys were talking about making metal-based 3D printing simple and inexpensive and that's all they focused on I'd be singing their praises. If I could buy a metal 3D printer today for, say $50K to $100K, with better throughput, reliability, accuracy and operating cost than compared to a VMC (and the ability to work with a dozen different materials) I'd be first in line buying half a dozen of them. What they are talking about instead is just silly.
No - we aren't fucking kidding you, RebootTheSystem - we actually are making it a lot easier! ;-)
We usually find the most resistance to our ideas from within the industry itself and among people with experience, which is why it's not changing. It's like the cab drivers complaining about Uber.
We've got a ton of great customers (world class names, to people in their garage) who love us and our revenue is growing like crazy. That's how I know we're on to something.
If making hardware was solved, why isn't everyone able to execute on it? Why is iterating physical products so slow and expensive? We were able to easily raise money because VCs saw the hardware startups struggling.
There is so much room to go in making manufacturing / physical engineering easier - Moore's Law has barely made a dent in manufacturing. I'm excited for our journey ahead!
Because you are moving molecules, not electrons and the cost of mistakes can range from financially crippling to causing a company to go bankrupt to killing people.
Example: Takada Airbags.
Do you really think it matters one iota if they had better rapid prototyping or a better version of Proto Labs or a better CAM software package?
Or, how about manufacturing a VW car? How many billions are they losing because of software? How significant do you think eliminating the machinist (or whatever) might have been in this case?
Or how about these issues:
The resistance you are seeing is from people who exist well outside the echo chamber you are in. I mean, listen to yourself, you are comparing Moore's Law to manufacturing? Again. Are you fucking kidding me? Please get that out of your head.
Funding? Funding means nothing. Funding doesn't mean an idea or a product is good or will be successful. It just means you were good at selling what you wanted to do to investors. And, in some cases, absolutely clueless investors (not saying that's the case here). That's it.
Again, you are in an echo chamber. Funding means nothing. Fan boys mean nothing. Here's a list for you:
https://www.cbinsights.com/blog/biggest-startup-failures/
Many more like that one.
Look, where you lose someone like me is when you start to spew this nonsense about Moore's Law and manufacturing or the software engineering process and manufacturing.
> It's like the cab drivers complaining about Uber.
Another nonsensical example.
And, yet again, this betrays your lack of understanding of the problems faced in real product manufacturing. And your other comments showcase your disdain towards those who, unlike you, have been busy making real progress in a range of industries, from consumer electronics, to brick manufacturing to medical, aerospace and everything in between.
Manufacturing is always progressing because it is often the largest portion of the cost of getting a product to market. And so, the motivation to optimize manufacturing for cost, speed, efficiency, reliability, throughput and quality has been there for centuries. One could not be competitive to day with the processes and tools used 20 years ago.
Yet you somehow reduce the totality of the manufacturing problem to "we need to CNC machine prototypes as fast as software teams iterate code". Who gives a fuck? Really? That problem is solved, to a large extent, by both technologies and companies.
I can 3D print complex parts in metal while I sleep. I can have a complex 5-axis parts machined perfectly and in my hands in three days and at a reasonable cost. You are creating a solution for a non-existing problem.
I just don't see it.
Want to solve a real problem? Solve all the nagging friggin problems in 3D printing plastics. From speed to quality to dimensional accuracy and repeatability and more. There are real problems that need solving in that industry. The other stuff, what you are talking about, is a waste of time and money.
Anyhow, we obviously disagree. I wish you the best.
Hardware isn't hard because it's physical. It's hard because there's still a lot more progress left to go. There are no physics limits to making it orders of magnitude better on many dimensions. We just need to revitalize a stagnant industry.
"Moore's Law to manufacturing" means taking the gains of computers and applying them to manufacturing where they're vastly under applied. Visit IMTS or Fabtech and you'll see how behind everything is - machines and software designed 10-30 years ago.
Plethora is more than putting CNC machining on the internet. It's a new way of thinking about manufacturing. About taking the analogies of software and porting them to the manufacturing world.
We've made a debugger, compiler, simulator, etc. for manufacturing CNC machined parts, but that's 1% of what we'll do.
In the end, I suspect you're reacting to our optimism about radically changing a system - which I find funny for a user with the handle "RebootTheSystem" :-) Have fun out there.
I visit IMTS every year. Progress is significant and steady.
Many years ago I tried to lead a "movement" to stop using G-code and go with something that made much more sense to me. I wanted something closer to a modern programming language, something that, in my mind, worked better.
In the end I failed because of a simple reason: I wanted a solution to a non-existing problem. Yeah, G-code is "unrefined" and cave-man-ish but, in the greater context of things, G-code is an insignificant part of the equation. It's a solved problem. And I was trying to solve it all over again.
That's just one example.
You are beating your head against a wall.
It's really funny to read your comments about machines and software designed 10 to 30 years ago.
First of all, you are wrong. These systems are always being improved. Every year. The problem you don't seem to understand is that manufacturing isn't like buying a bunch of PC's and upgrading them every year.
I can't throw away a dozen Haas CNC machines of different types every two years to adopt someone's hair-brained scheme. I can't dispose of my very expensive automated SMT assembly and optical inspection lines because some kid in Silicon Valley decided they want manufacturing to be hip.
No sir, this isn't a hobby. If you want me or anyone else to shift into new equipment just because the UI is cooler, well you lost right there.
If you, on the other hand, came up with, as you said, "orders of magnitude" better manufacturing, well maybe you have something.
You do know that an order of magnitude is 10 times better and "orders of magnitude" is at least 100 times better, right? See, this is where you lose me. The only way I can put such a statement is that is it ignorant of reality to an incredible degree. Naive beyond description.
Just what do you think people have been doing for the last 30 years? Waiting for you to come along while using stone-age tools to manufacture product so you could make things 100 times better?
Again, echo chamber. I don't think you understand just how incredibly stupid some of these things sound to someone who lives in the real world making real products with real modern manufacturing. You really don't have a clue, do you?
OK, here's another challenge for you. Since you seem to be focusing so much on IMTS and arcane CNC machines.
Show up at the next IMTS with a next generation CNC vertical machining center and lathe of your design. Demonstrate how your hardware and software combination is "orders of magnitude" better than prior approaches. I'll let you slide and say, just show me an improvement of 10x. That's it. Show me how you can make things 10x better by designing new machines with new software.
Your statement was "you'll see how behind everything is - machines and software designed 10-30 years ago.". OK, great. Show-up at the next IMTS and be the beacon of light with machines and process that is at least 10x better.
Of course, I fully expect the exact opposite to be the case. If you attempt to do such a thing you will soon discover just how little you know about manufacturing and how much more you have to learn. I feel sorry for your investors. As for your "growing customer list", it'd be very interesting to learn who they might be. I have a feeling I know.
As for "rebootthesystem". Well, you are reading too much into that. It's just a name, not a cult. What you have is a cult. I am all for driving progress but it has to be in the realm of the plausible. One requirement there is to create attainable solutions for real problems.
This isn't about old school clashing with new ideas. I know you and your team are painting my feedback as "typical old-fart who doesn't understand". Have a good time laughing at my expense. No problem. I have exactly zero problem with developing ground breaking modern solutions and questioning everything in the process. That is exactly what I do each day. And, guess what, manufacturing isn't where the problems are. Could it be better? Sure. 10x better. That is impossible. 100x better? C'mon, get real.
You understand real hardcore manufacturing and have a lot of experience in that realm. Plethora seems to be targeting people that don't know anything about manufacturing but want an easy way to build parts in small quantities. I don't know how they plan to scale to full manufacturing, and I don't know their numbers (the real question is retention, how many customers are coming back with larger order numbers), they may have found a niche by targeting customers that just don't know any better. I assume this market is small and I am skeptical of its value at all, since I know it's easier to just hire someone internally or just outsource it to a competent shop, but something Silicon Valley can do like nobody else is hype itself up. What that means is a lot of things, but it includes pumping up valuations for an eventual acquisition.
Is Plethora creating real value? I'm extremely skeptical. Are they attempting to create a company that can be sold for a profit? That seems pretty likely to me, and that's what the overwhelming majority of investors care about.
I think part of the problem is Nick sounds like he's pitching investors in this thread, when he's talking to people that are experienced in manufacturing. You're not his target audience. Then again, I doubt he's planning on showing this to investors, so I'm not really sure what his goal is here (unless he's just defensive, which is a really bad sign for him being able to see the forest for the trees).
While, admittedly, my tone has been rather aggressive it has not been my intention to aggressively do anything than to say "you are making a big mistake". Making an "orders of magnitude" claim in a field like manufacturing can only happen if your customers are utterly ignorant.
Find a tribe deep in some unknown forest who knows nothing about modern technology. Show up with a box of matches. You've made fire-starting "orders of magnitude" easier.
Today I had a meeting with my local Haas representative. Yes, it's Sunday. I'm buying another machine and he is very accommodating due to year-end needs. We had an interesting conversation about improvements in the industry. Haas introduced an updated control about a year ago. If you look around on the 'net there are people who hate it. Some even talk about having returned machines due to the new control (and buying used machines with the old software).
Are the controls that different? No, not really. However, to an experienced machinist who has been running these machines daily for years their "muscle memory" gets them into trouble when using the new control. I would compare it to switching between Mac and PC keyboards. I do that all the time and it can be a pain in the ass. When developing iOS code I have to use a Mac while all other tools are on our PC's including such things as Photoshop. My solution was to install a piece of software that allows me to remap the keyboard on the Mac to match a PC with sufficient fidelity to eliminate the cognitive load of using both platforms back-to-back.
Imagine now having a shop with half a million dollars of machines you know how to use like you are playing a piano and the "new and improved" version of the machine has you making mistakes and feeling like you don't know what you are doing. It isn't that bad but there's enough of a cognitive load for some (not all, a few) of those who have to deal with this to not be happy with Haas.
Another example is when MS switched-up the UI from Office 2003 to 2007. I felt like moron for a few weeks.
And then we have someone come along and suggest they are going to make things more like software development and achieve an "orders of magnitude" improvement in manufacturing. While, in the real world, a highly experienced, well respected, leading-edge manufacturing company has to take baby steps to improve the UI and control of their $100K to $500K machines. The reality of manufacturing is that people need to make things, not play with toys (hardware or software) every time a new shiny thing is put on the table.
I wish these guys the best but their messaging is horrible. To anyone who knows manufacturing it screams "I don't know what I am talking about". I am sure there are a bunch of Kickstarter projects that could use help. If they can build a business around that, fantastic.
Our core customers are actually experienced engineers, but our publicly exposed product (CAD Add-in, site, etc.) isn't the offering where we're currently making most of our revenue, but in Q1 it will be. We'll announce the specifics publicly, so stay tuned. I suspect you'll be happy with what this has. Unless you have a exotic needs, we'll be able to make it.
We essentially decided that it's more important to satisfy our early enterprise customers using a lot of hand-selling until the CAD Add-in is fully there - which is the slow part because the computational geometry is quite tricky to get the experience you want for instant quotes / feedback that are meaningful.
We are NOT building this to flip. There are far easier companies to start if you're looking to make a quick buck, and Founders Fund / Lux don't invest in companies with a small vision. We haven't talked a lot publicly about our long-term vision, but it's a core piece of our DNA.
Happy to answer more questions.
Other than that, you are more expensive than hiring a machinist the old-fashioned way for one-off parts. Apparently I just don't understand your business, because I'm clearly not your customer.
Also please don't name drop your investors - nobody cares. Funding is a vanity metric. Focus on how you can be profitable when your business is no longer subsidized by external funding - that's what will make you successful. Growth is great but it doesn't help if you can't ever sell at a profit.
Anyway, while I may not understand your business, I wish you luck. If you can truly revolutionize manufacturing the way you say that would be great.
From our price studies, we're on par with other fast turn shops and faster than all of them with big high mix orders. You can get cheaper parts slowly, but we focus on people with more money than time.
Ha, funding isn't a vanity metric when your company needs financing. If you're doing a really big project in a high-growth competitive market, you'll very often need to go the VC route. Bootstrapping doesn't work for this.
If you've ever done it, you know that raising angel/VC money is hard work. I respect people who are able to raise funding, and far more I respect when they turn that into great products that help a lot of people.
Yea, I think making manufacturing better is a big deal, and there is much work to do. Thanks for all of your feedback!
Your studies are cherry-picked to make you look good then, as that just isn't true.
"funding isn't a vanity metric when your company needs financing"
Needing to raise money and bragging about it as a sign of success are very different things.
Turn off the pitch, dude. I've already told you several times I'm not your customer. I'm bowing out of this conversation now.
Few non-trivial products use a single manufacturer over their lifetime, let alone for a single production run, so having your own CAD system is a liability not a competitive advantage. I don't see how neckbeards, location, and VC funding are relevant in the face of such a glaring oversight.
Autodesk Inventor's sheet metal design tools will let you design things in sheet metal which are not makeable with standard machinery. It will show you the unbent flat form of the part, and allow for the metal thickness at the bends, but there may be no sequence of bending operations that works because the previously bent part gets in the way of the tools. eMachineShop understands the limitations of bending.
While definitely a good point (restrictive CAD keeps you out of trouble) the issue you have is that most companies want canonical CAD files living inside Solidworks, etc. and don't want to go back and forth. It's a lot of extra work, and people aren't switching to just using their CAD package because it's not feature complete like a commercial package.
Haha - I appreciate the neck beard call out John - well trolled ;-)
My reaction to this is essentially: "Why would Google make another search engine if Yahoo already did it?" which is a pretty common trope for most startups to hear. We're not worried about it.
I tend to think companies that don't change much in ten years are more victim to their lack of execution/ambition than actually hitting some external roadblock (competition, etc.).
The SF location is less for the good coffee and more about being close to some of the most progressive customers, employees, and investors. The Bay Area overall has been so good for us, and we're excited to keep growing locally and beyond.
The guild/apprenticeship model in the 20-21st century has always felt to me like a thin veil over a seniority-based structure that places strong barriers of entry on newcomers to protect the jobs of the senior staff. Want to do that job? Sorry, you can't, kid, until you "pay your dues". Want more money? That's based on how old you are. It's pretty backward and broken for a capitalist system that should pay market rates for providing value regardless of age or guild-proven experience. These types of things irked me as a child.
How much innovation would we have in technology if you had to spend five years on a starter Internet, or were forced to just use phones from the start? What if you were only allowed to program in certain languages and weren't admitted to download a specific compiler?
It seems realistic to want to prevent beginners from touching expensive machinery that they could break through incorrect use.
Based on my experience, equipment at Techshop is, on average, an order of magnitude or two cheaper and a decade or two older than what you'd see at a commercial manufacturer.
Both of these are small machines. Larger machines can easily be over 100,000.00
You want your large machines to always be working, that means you want someone who can set it up right and fast. That takes experience.
The real shame is when businesses let those people retire without bothering to adequately train a replacement.
Also 5 years to be a machinist, I don't think that is true anywhere. I don't know the German system in detail, but I live in Norway which has a similar system. You take 2 years of high school to prepare for say a machinist and then 2 years as an apprentice. That seems totally reasonable to me. That means vocational training is only 1 year longer than somebody taking 3-year high school to prepare for university study.
In Norway we have seen what has happened with the professions not being as regulated, like construction work. We have vocational training for this field but the applications to this study has fallen through the floor. Years of influx of polish construction workers have pushed the wages, benefits and status down. Initially the polish immigration was god send because we had serious shortage of construction workers. However now it turns out this was almost like peeing in your pants to keep warm. Should polish workers not stay, we are screwed. We are losing our native competency in this area and we have a lost a whole generation of native Norwegians.
More regulated professions like plumbers still see strong recruitment. They got higher wages and status. That means schools retain teachers, equipment etc. Getting rid of the guild mentality might make sense in a pure economic sense. But only in the short term IMHO. In the long run you shoot yourself in the foot, because you reduce the attractiveness of the profession. Remember the main problem in the west today isn't that people aren't able to train for these professions or that the barriers are too high. The problem is that too many chose to not get vocational training because it doesn't have high enough status. People want to be college educated.
If you kill the guild like approach, you also kill the status. If anybody can do it, without any papers, then the profession commands no respect.
One of the key reasons Germans kick the shit out of the UK in manufacturing is that being a craftsmen is highly respected in Germany, while it is not in the UK. Perhaps not an accident that this is because the medieval guild system lasted much longer in Germany than in Britain.
> By the way, I actually think that computer programmers are not going to be a white collar job in the long run. Machinists were the programmers of 100 years ago. It was a highly paid job, and can still be one even today. My grandfather had a great job. He was almost a petty bourgeois guy after being a prototype machinist. That’s not true for that job anymore. I actually think software programming will go the same way.
My father trained / apprenticed as a tool & die maker in Germany. When he moved to North America (Alberta, Canada) this was not really a thing he could do where he moved. So he became a machinist.
Machining where we lived was/is all about the oil and gas sector and by and large not large unionized shops and not great pay. Mostly sweat shops intense work, lots of underpaid immigrant labour, and my father suffered long bouts of unemployment or unstable work through the 80s and early 90s. So yeah, formerly a 'skilled trade' and then very working class.
In the end things turned out well for him as he ended up landing into a great job teaching machining at a technical school.
He applied for a shop-assistant position at the school, after years of soul crushing unemployment or marginal employment.
They interviewed him and said he wasn't qualified to be the shop assistant -- but they wanted to offer him a position... as a teacher.
He got a full teaching job with benefits, a pension, summers off. A very happy ending.
The same is true for software. But there, the strongest employers are much farther ahead of their struggling peers than a manufacturer of goods could ever be. Google is putting whole industries out of business, at global scale, with a tiny number of employees compared to what they are replacing (and they probably could even shrink their workforce to a fraction of what they have today and still do roughly the same). In consequence of that it seems quite reasonable to expect that the wage difference between the wealthiest and the poorest employers will likely stay much higher than it has ever been for machinists.
Low wage programmers might find themselves in a proud "temporarily embarrassed millionaire"-trap for a very long time. (if it is a trap, "high-paying job at a low-paying company" could also be a path to zen happiness for some)
Building the systems to make the tools to automate that away? Those will still be white collar jobs. Pushing bits is easy, converting an abstract vision into a scalable, maintainable system is hard.
The best programmers are always learning new languages, programs, frameworks etc. Avg programers get good at 1 thing and stay with it.
The real difference is that the first group is training their brain how to learn and adapt quickly - that is the skill that will keep them employed long past the days of simply writing code.
Writing python or C is not the skill. The skill is how quickly can you deal with a new problem you have never seen before.
It already did, it is called globalization and outsourcing.
Business love the wonders of Internet and remote working, paying a dime per hour that they would pay back home, without any of those worker rights guys nagging them.
Specially in any business domain that doesn't focus on selling software.
Consulting, architecture (domain, technical, business), integration and customer management is were software development is going to.
Pure code writing has become like factory work on a distant country, connected by an IP address.
There is also a project management overhead cost to outsourcing that must not be ignored and is prohibitive for smaller companies.
If we want to get anecdotal, across many different companies the only place I've seen outsourcing work is in a huge enterprise and only with several expensive and highly trained local devs cleaning up the crap. Otherwise it's just not worth it. AI will take our jobs long before outsourcing does.
I used to be very critic of the quality, specially since I happen to do reviews of delivered code, but then I also have seen the quality of some internal developers at those customers.
All because selling software isn't their core business, so as long as whatever is running on their servers, desktops, tablets, phones, helps to sell their core products (not software), no one cares one single second about code quality and how it is written.
Just like 99% of people don't care how the goods at their local stores are produced.
The minimum price for a machined part (a 2" x 3" x 0.5" piece of 6061 aluminum with no other machining) is about $95. The soonest available delivery date is 7 days away. The price goes down slightly if you select a date further in the future.
The plug-in will detect and report un-machineable situations, such as square-sided pockets and narrow slots into which a tool could not reach. Holes cost about a dollar each, and a machined pocket costs about $10-$20. Adding a pocket on the reverse side of the part, which requires a second setup, added about as much cost as a front side pocket, which leads me to suspect that the program doesn't figure out how many setups are needed.
Chamfering a top edge added another $100. So curved surfaces really cost. The plug-in doesn't give any hints on cost reduction, unlike eMachineShop. You don't get told the operation sequence; that's a black box to the user.
Overall, it seems like a useful basic machining back-end program. Has the feel of eMachineShop circa 2003.
Seeed Studio in Shentzen finally got the assembly thing working. They make blank boards and will put parts on them if you pick the parts entirely from their Common Parts Library. You can make most simple digital electronics projects using that library - analog and power, not so much. Seeed Studio has good prices but requires you to release the rights to your design, so others can copy it. Sort of like Github for hardware.
For good reason. If we are talking real production, not 3D printing a bust of Beethoven for fun of buying a Shapeoko to cut a little bracket. If that's what we are talking about, well, you are not going to take someone who hasn't proven themselves trough years of training and learning on the job to run something like my Haas VF2-SS, VF3-SS or VF4-SS.
One mistake could cost tens of thousands of dollars in damage to the machine and untold losses due to disruption to production. This can include damage to very expensive tools (we are not talking about $0.50 Home depot drill bits here). And then there are such things as optimizing tool wear and throughput.
Despite what is said in this article real machinists don't sit there guessing at what to do. They know because they studied, learned and made enough mistakes.
Fields such as manufacturing highly accurate medical, optical or aerospace parts require real skills.
Can all of this be automated to the point of eliminating both human CAM toolpath programming and the machinist? I'd say in a very limited set of cases and with restrictions that do not make for efficient manufacturing.
This is one of the problems with Proto Labs. They can machine quite a range of parts but they are very, very expensive for what you get. Sure, for quick turn "I just need a part" manufacturing (and, to be fair, more than that) they work fine. For anything else, no way.
Here's an example from a couple of weeks ago. I needed a half dozen copies of a part made out of mild steel. We needed to send it out because our machines are busy. Proto Labs could make them for $300 a piece. I called a couple of local shops. I had them made for $20 a piece and delivered to us overnight (one of the guys from the shop drove them over). They used their waterjet machine to cut the basic shape quickly and then threw it on a VMC to square-up and finish all the other features. Twenty bucks.
Yes, we will eventually succeed at automating the CAD-to-Machined-Part process to a significant degree and with good results. This will require AI, Machine Learning and a lot of the knowledge accumulated out there encoded into these systems. You need to be able to make real parts with real manufacturing problems. Any part. Some parts require designing new parts to make them, whole machines even (fixtures, jigs, clamping and alignment systems). It isn't a simple problem.
And, even as these problems are solved the totality of manufacturing isn't CNC maching aluminum. Not even close.
I've taught a number of people to use 3-axis CNC equipment more or less safely if not capably in several evenings, including Solidworks introductions. I don't have patience for people telling me how hard it is to learn to use machinery well.
I have less patience for the guild system. It takes something pretty straight forward and whines about how hard things are to keep prices high.
The best tools for learning to make your own electronics, machined parts, code, and actual products is the right attitude, some intelligence, and the right tools that don't require months of intense tolerance to bad tool design that require you to have months of knowledge to make basic models or parts. It takes a while to learn to machine well, but not years for crying out loud.
Rather than expounding on all this experience you have and all the effort and blood and sweat and tears it took you to gain those skills, it might be better to acknowledge that a lot of the hurdles are bad systems. Myself and a lot of other people have gained these skills with a lot less frustration than perhaps many others. There are a lot of smart people out there. They don't need Proto labs. They won't need this offering really, either. But some will, and a broader swath of the product development public will have greater access to more options with good free CAD, automated post processing, or better/easier PC layout tools.
The app kiddies aren't going to take the manufacturing world by storm. But the guilds are basically watching themselves get eaten alive, and that is a good thing. Let product developers do this stuff themselves. The mythology of it being so hard is a fantasy, and a lot of people are starting to realize it. Any young ME who doesn't have strong EE chops and vic versa already has a harder time making a career these days, and frankly, any engineer without strong fab skills now is definitely less valuable.
Example: TIG and MIG welding. I have both. I can make decent welds. They look pretty. Learned on my own just futzing around here and there. Yet, I would never, in a million years, trust my welding to, say, build even a light aircraft. I, objectively speaking, don't know what I am doing when compared to the aerospace certified welders I walk by every day.
> The mythology of it being so hard is a fantasy
I disagree. Perhaps scale and application is the difference here.
A gadget for a Kickstarter campaign? Sure, anyone can screw around at a maker space and get one or a few done.
Something like a Dyson vacuum cleaner that needs to be designed and manufactured at scale and be safe and durable in the field? No way. This requires experienced engineering specializations from EE, ME design to manufacturing engineers, materials engineers, test engineers, machinists, mold makers, process engineers, etc. Way different reality.
Move up another level to Medical or Aerospace and things get even more interesting.
And some industries are just awe-inspiring in terms of their manufacturing process. One such product are LCD displays. A typical factory for the LCD displays on your computer costs somewhere in the range of two billion dollars to build. Two billion. It's a massive undertaking. Automation, clean rooms, robotics and all kinds of specialized processes designed to optimize quality, cost and yields. To compare that to the software "manufacturing" process is as silly as one can get.
Will AI help streamline some of these processes over time? Sure. Of course. You are still going to need a two billion dollar plant to make LCD's and similarly scaled facilities and equipment to make cars, radios, batteries, ovens, etc.
Look at making cars as another example. Even if the CAD-to-CNC process was completely automated and perfectly executed it would still be less than a rounding error in the context of the cost, processes and complexity of manufacturing a car.
UK: HPGL and a bit of autocad. The lads (and ladesses) with the utility trousers and steel toecapped boots round here (motor vehicle supply chain) all know that. What else can you show us?
(Here's an example of Hypermill at work.[1] This is what the high end can do.)
If I remember the results of googling 'Creo vs Solidworks' correctly, Creo is better at handling really big assemblies and is used extensively in automotive industry. Most of my peers prefer Solidworks.
I don't know what SolidWorks charges now, but I heard that the price is considerably more than $4k.