The NeuroForge vision
Intelligence that
keeps growing.
I want capable AI to become something people can build on, extend and afford to use.
NeuroForge is working toward systems that accumulate knowledge and skills, carry experience forward and choose how much computation a problem deserves. That is the reason for ERAIS, Fracture and the experiments around them.
Connect.
Keep growing.Knowledge · Perception · Memory · Action
Useful capabilities should become lasting assets.
The central idea
Use the expertise
the problem calls for.
A system may need a large repertoire of skills. Any one task draws on a particular part of that repertoire. ERAIS is built around that distinction: stored capability, selected execution and an explicit budget for the work.
Biology gives us useful principles to investigate. Perception, memory, regulation and action work together under tight energy constraints. We test those ideas as computational mechanisms, alongside components learned by artificial neural networks.
Fracture provides a way to reuse existing learning. It converts supported model components into native ERAIS execution, then checks their outputs and state against the source. Each successful conversion expands what we can build with.
Why it matters commercially
Change what people
can afford to build.
Training, deployment and integration costs shape who gets to use advanced AI. These are the opportunities driving our development.
01Researchers and independent labsRun the experiment your budget usually rules out.
The constraint
Model size, hardware access and the cost of repeated training limit which ideas get tested.
The opportunity
Reuse trained components, choose the computation a task needs and run reproducible experiments on accessible hardware.
The commercial consequence is a lower cost of finding out whether an idea works.
Talk through your research02Model builders and product teamsAdd the capability your product needs.
The constraint
Each new requirement can trigger another model, another serving stack and another round of integration.
The opportunity
Build around reusable capabilities with explicit interfaces, retained state and tests for the behaviour customers depend on.
The aim is to protect the value of existing training and make upgrades smaller, faster and easier to inspect.
Discuss your product idea03Communities and their moderatorsPut your community’s knowledge to work.
The constraint
Useful answers are spread across rules, FAQs, conversations and the people who keep explaining them.
The opportunity
Make approved knowledge available through a community AI that can answer naturally and help moderators handle repeated questions.
The Discord product is in development. We’re testing how this research can become a useful community service.
Explore the Discord project04Industrial and field teamsCarry expertise into the field.
The constraint
Power, connectivity, privacy and equipment constraints shape what an AI system can do on site.
The opportunity
Develop task-specific systems that can combine relevant vision, language, memory and control capabilities within a defined hardware budget.
This is a longer-term deployment goal. It connects the architecture to the constraints of real equipment and real work.
Discuss your hardware and taskOur product pages explain current availability. For a model or research problem, tell us what you need to achieve.
The work underway
Build it.
Put it through its paces.
We have native training paths, source-model conversions and embodied systems running. The next job is to improve their quality, cost and reliability together. Our results identify the model and test conditions so the progress is inspectable.
Read the current evidence- ArchitectureExplore ERAIS
How capabilities, memory and computation fit together.
- ConversionInspect Fracture
Supported sources and measured preservation checks.
- ExperimentsSee the systems
Language, speech, vision and agents acting in a world.
Lloyd Handyside / founder
The ambition is to change
what intelligence can do,
and who gets to use it.
I bring biology, physics, data and industrial experience to this work. I use AI throughout the research and engineering process. Every useful result gives us something stronger to build the next iteration with.
How I workBuild with us
Bring a hard problem
and a reason to solve it.
We’re looking for model builders, researchers and partners who see the value of reusable intelligence and want to test it on a real task.