Robots will be built like software.

Coding has flipped from writing to directing. The best engineers now spend more time steering systems than typing lines of code. The interface has moved from IDEs to intent. 

Now apply that same lens to the physical world…and turn the dial up.

Full-stack robots should be designed the same way: hardware, data, and software developed together, iterated rapidly, deployed continuously. We’re nowhere near that. 

Not because the intelligence piece is missing, but because the hardware manufacturing can’t keep up.

The real gap

If you compare a world in which robots augment human work wherever it makes economic sense with where we are today, the gap is staggering. The question isn’t if we get there, it’s what’s slowing us down?

The answer isn’t intelligence — it’s manufacturing.

Robotics today feels like early 3D printing in the 1990s; technically viable, proven in labs and niche applications, but constrained by cost, complexity, and production speed. Once machines got cheaper and design software matured, the industry unlocked mass adoption. 

Robotics is at a similar inflection point, but the upside is far larger. Software stacks are evolving quickly: simulation, data pipelines, vision-language-action models, retraining loops. The software side is moving fast.

The hardware side is not. 

The hardest hardware problem 

Building a humanoid robot means compressing millions of years of biological optimisation into a machine. 

If you want to understand why this is hard, start with the human hand.   

Hands aren’t a “gripper problem”. They’re the general interface to the world: strength and softness, precision and abuse tolerance, high-bandwidth control in an insanely compact form. Human hands have 27 degrees of freedom, powered by tendons pulled from the forearm like a puppet. 

That’s why Elon Musk said bluntly about Optimus:

“To create a general humanoid, you must solve the hand problem.”

Robotics has been trying to do this for decades. The result is usually one of two compromises:

• Extremely capable hands that cost $25k-$100k+ and take weeks to assemble, or

• Cheap hands that are too simplistic and lose what makes hands useful in the first place.

And even when a team builds something impressive, the next bottleneck is manufacturing:

Can you reproduce it thousands of times?

Can you iterate without burning a year of the team’s time?

Can you scale without exploding cost and failure rates?

That’s the wall robotics keeps hitting. 

Why Allonic

Allonic starts exactly where robotics currently breaks: manufacturing. 

Instead of assembling hundreds of delicate components by (human) hand, Allonic’s 3D tissue braiding technology weaves fine fibres around a skeletal core. Tendons, joints, and load-bearing structures are formed together as a single integrated body. 

Inspired by biology and nature, but built like software. 

Its platform takes high-level digital designs and translates them directly into production code. Similar to how 3D printing turns models into toolpaths. 

Mechanisms that once took days or weeks of skilled assembly can now be produced in under an hour, with orders of magnitude fewer components and a fraction of the cost. 

The collapse in iteration cost is the gamechanger. And Allonic is pushing this further by integrating multiple materials on its platform and embedding wiring and sensing during production. This eliminates late-stage integration debt that kills reliability when teams try to scale.

So this isn’t about creating a better ‘gripper’. It’s actually about a new manufacturing layer for tendon-driven robotics.