Current product
PulSen Craft
The private beta is PulSen Craft Desktop and PulSen Craft Fleet on hardware you operate. Self-improving agents and a training feature are not in this beta.
See the productFoundation / long-term vision
Australia needs more than access to models. It needs control of the work around them.
This is a proposal for an Australian-controlled layer that can run, govern, evaluate and orchestrate AI. It can use local or hosted models. The choice is explicit: where execution happens, what data is shared, who can act, and what evidence remains.
The proposition
Control lives in the layers between a prompt and an outcome.
Buying access to a model is useful; it does not by itself decide where sensitive work runs, which tools an agent may use, whether a result is checked, or how a decision can be explained later. Those are product and infrastructure decisions. The thesis is that Australia should be able to make and operate more of those decisions on terms it controls.
01Compute and modelsLocal, Australian-hosted or selected external providers
02Routing and orchestrationFit the model and execution place to the task
03Identity and permissionsSpecify what an agent may see and do
04Evaluation and evidenceTest behaviour, record outcomes, enable review
05Applications and operatorsPut people in the control loop where they matter
Status, without shortcuts
What exists today; what does not.
Pilot work
Operational proof
A pilot would define its own workload, data boundaries, controls and measurable acceptance criteria. No customer deployment or independent security certification is implied here.
Explore a scoped pilotResearch and vision
Australian secure-AI layer
Shared governance, evidence standards, cross-organisation operation and Australian infrastructure partnerships need design, testing and capital. This is a direction to investigate, not a deployed national system.
Follow the evidenceWho it needs next
The next step is a programme of proof.
Technical and research collaborators
Help test permission boundaries, model evaluation, auditability, failure handling and deployment patterns with evidence that others can inspect. University, infrastructure and industry perspectives are welcome.
Investors and strategic partners
Capital would support engineering capacity, reproducible evaluations, pilot delivery and independent review. Funding scope, ownership and milestones belong in a proper diligence conversation; no return or market-size claim is made here.