Automating design for the physical world | Drones and 3D printing | Building @kyrall_
München, Bayern
Joined July 2023
- Tweets443
- Following891
- Followers773
- Likes494
Pinned Tweet
Today, we're officially launching Kyrall
Until now, AI models couldn't capture design intent in engineering.
We're changing that.
Upload specification documents, sizing tool results, requirements, and past designs, and Kyrall builds a Knowledge Graph, so it not only understands what you work on, but how you work.
You simply describe what you want to design, and Kyrall generates a fully parametric assembly, grounded in your engineering knowledge.
Not a mesh. Not a rendering. Real, editable, parametric 3D models.
We need to start seeing software as a “calculator”, the real value of the calculator is the work done with it and not the fact it can add numbers easily.
claude.com/resources/article…
1
84
Who is buying this data?
If these prices keep going up, at some point it will make sense to design and manufacture hardware just to sell the data
Are you crazy Natalie? No we are not selling our data. Data is what we use to make our robots smart. I am worried small mom and pops shop would fall for this though. Getting one of these almost every day now. Who is behind this?
90
People underestimate how big of a difference a seemles UX does. If you think about it, you (both the consumer and the business) make purchase decisions most of the time based on the UX you had rather than pure product performance/quality
90
I talk about design automation a lot, here is another one:
With design automation getting more mainstream, we will start seeing more bootstrapped hardware companies, it will start with simple products, maybe even consumer parts, but it will very fast move more towards enterprise and B2B.
But we still need on demand, 48 hour delivery, low volume manufacturing in Europe to do that...
116
If building proper software for the hardware industry is a 10 step process, the first one is showing a nice demo and the 9 others getting all the nuances taken care off.
the 80/20 split doesnt work here, either you build a good software that fullfills all the requirements or you have nothing. It is really binary and many Software devs coming into this space need to understand this.
As one of our users put it "if it doesnt reach 99% accuracy we dont see a clear value".
Happy to say we are even higher than that for most cases!
You see a lot of people posting something like “I will never trust a plane/car/etc designed by AI”
Such a statement obviously shows lack of understanding of how AI will be used.
No is saying that we need to eliminate testing, simulation, and other validation methods.
A plane designed with AI will still go through all the compliance and certification gates existing today.
Using AI will do two main things:
1. Eliminate manual work
2. Connect CAX tools together
It is as simple as that, yet will profoundly change how we design and manufacture physical objects
The team that designed grippers with AI won the hackathon!
We recently sposored a Hackathon in Munich were the teams had to design and manufacture three different grippers to move three different bodies.
1. A ripe tomato
2. A heavy metallic cylinder
3. A flat slippery plate
the teams used AI to design the gripper, simulate collision path, simulate the kinematics of the grippers and much more. Most of the team members had limited knowledge in hardware design and never operated a robot arm before.
here is what happened:
While these methods seem simple and they are, they were developed by people who dont work on design on a regular basis and with limited knowledge in developing mechanisms and in a short period of time.
This showcases in what direction this tech is moving.
As a bonus, the team also developed a tool changer to simplify switching grippers and they were working on automatic tool change when the time was up.
1
15
The developed grippers were built with 3D printed parts and laser cut wood. Simple prototypes, but developed in a very limited time and with limited knowledge.
The team needed a simulation tool to make sure the mechanism works well, they vibe coded that. Also they vibe coded simple FEM tools on the spot.
The future of design will a place where engineers are orchestrators of the AI as it design, simulates, develop custom tools, and calls exisiting tools.
Soon, no manual work anymore!
11
AI isn't at the level of a senior mechanical engineer yet. True.
But calling it useless because of that is like saying in 1920: "Cars are slow and expensive, so they're useless."
5
8
683
If AI can design cars, drones, planes, etc. what will the defensibility of automotive, aerospace and robotics companies be?
Could the same happen to them as what is now happening to SaaS?
Is the MOAT then the manufacturing facility? The network effect? The brand name? The vertical integration?
I guess taste will actually become more of a moat with time.
The bitter lesson is coming for us all
Fable 5.2 shows substantially improved skills in controlling physical robots, even on tasks it has never explicitly been trained to do.
On RoboDojo, a benchmark with a simulated two-arm robot, it outperformed other models and most AI systems built to control robots.
7
496
Some think that AI will one day one shot CAD and other CAX software
While that will probably be possible one day, it isn’t what “should” be built
The future is a new way of performing design.
The current separation between CAX software is because of inefficiency and lack of automation, once that is solved they will collapse into one software/one interface