Tools & services ERAIS Experiments Evidence Company Search site Contact

Browse all pages

Founder / NeuroForge

Lloyd Handyside

I started NeuroForge to build AI that keeps learning and uses the hardware we can afford. Here’s what I’m working on and why.

Read my story Get in touch

A note from the founder

Why I started NeuroForge.

What I’m building with ERAIS, and why I think AI should be cheaper and easier to keep improving.

1 min 34 secEnglish captionsTranscript below

Accessible alternativeVideo transcript

Download plain text

Hi, I'm Lloyd Handyside, founder of NeuroForge.

I started ERAIS from a fairly simple question. Why do we treat brute-force compute and throwing away already-trained models as inevitable parts of advancing AI?

ERAIS, the Efficient Routed Adaptive Inference System, E-R-A-I-S, is a novel sparse and adaptive AI architecture designed to use computation selectively rather than activating everything uniformly.

We're developing two paths into the same architecture. New models can be trained natively as ERAIS, and the second path is to convert existing dense models into ERAIS, with the goal of preserving useful capability while reducing the active compute needed to run, update, and continue training them.

Supporting that architecture is a rigorous research and analysis stack: exact checkpoint recovery, model and data lineage, controlled evaluation, the ERAIS Observatory, and evidence controls that keep measured results separate from projections and unproven claims.

We're already seeing strong, bounded engineering results across CPU and GPU systems, but the bigger point is how we do the research. Negative results stay in the record, and claims move forward only when the evidence does.

The goal is straightforward: make advanced AI more efficient, more accessible, and easier to keep improving without throwing away what already works.

That's what we're building at NeuroForge.

On this page

I started NeuroForge because I think we can build intelligence in a much more useful form: a system that keeps learning, adds capabilities and puts the right amount of effort into the problem in front of it.

That is the ambition behind ERAIS. Fracture is our way of bringing what supported models have learned into ERAIS. The value of that training should carry forward.

How I approach the work

My background crosses biology, physics, chemistry, big data, programming, industrial work and academic research exposure. Those fields give me different ways to look at a problem. I use whichever helps explain what is actually happening.

I left my previous work to build NeuroForge full time. I work on the systems, experiments, code and infrastructure, as well as turning that work into products and services. That lets me follow an idea through to what it actually does.

My working principle is:

I do the impossible because most things called impossible are just incorrectly defined.

For me, that means finding the real constraint: the hardware, the representation, the measurement or an assumption nobody has revisited. Once I know what’s getting in the way, I can work on it.

Working with AI

I use AI throughout the day to research, write code, test, analyse and keep track of a large body of work. I set the direction, design the experiments and make the decisions about what the results justify.

AI helps me cover more ground. I still check which code ran, what changed and whether the results hold up.

The team I want to build works in the same way. Researchers and engineers with real depth in their fields, using AI to extend the amount and quality of work they can do.

What I want NeuroForge to achieve

I want useful learning to carry forward. A team should be able to reuse it, add to it and improve one part of a system. The hardware budget needs to be part of the plan from the start.

For a small team, that could mean keeping what a model has learned, trying ideas at a lower cost and making upgrades easier. Our tools and services give us a way to learn from real work: RKC for finding sources, Sinter for community records and handovers, and SubjectNest for learning resources.

There is substantial engineering ahead. We have working systems, recorded conversion results and live experiments to build on. The next steps are better quality, better efficiency and dependable operation on real tasks.

Explore ERAIS, Fracture, the current results or our funding and research priorities.

The company

NeuroforgeIO Pty Ltd is a founder-led, pre-revenue Australian company. Its ACN is 701 314 329 and ABN is 87 701 314 329. Company information and contact details are on the NeuroForge page.