Sovereign AI, built in Australia — and running in production.
Verificate builds the execution substrate for governed, deterministic decision intelligence — so any organisation can run capable AI on the infrastructure it already owns, without sending a single byte to a third party.
The best AI is the AI you fully control.
Cloud inference means scarce GPUs, unpredictable per-token bills, and sensitive data leaving your boundary. Meanwhile every enterprise already owns racks of capable CPUs. The gap was never the hardware — it was making a serious model run precisely on it, and making its decisions governed enough to trust.
That is the whole of Verificate: sovereign execution, and deterministic, auditable decisions — on your terms.
Building the inference layer for sovereign AI.
Craig founded Verificate to close the gap between what enterprises are allowed to run and what they can actually deploy. His work spans AI systems engineering, enterprise architecture and applied machine learning — with a single throughline: making capable models run precisely on the CPUs organisations already own.
He designed and built both the HELIX inference runtime and the Decision Transformer decision layer from first principles, took them through production validation on IBM Fusion infrastructure, and delivered the UNSW deployment that is now a live foundation customer. He is the author of Verificate’s two published research papers on inference-level hallucination prevention and typed-state Decision Transformers.
Verificate is founder-led and scaling deliberately — which is why customers get direct access to the person who built the system, not a sales layer in front of it.
From first commit to foundation customer.
As an IBM Business Partner, Verificate has access to IBM Fusion HPC infrastructure and the Granite model family. HELIX runs any open model — Granite, Gemma and larger open weights among them — and sits alongside existing watsonx.ai workloads rather than replacing them.
The University of New South Wales runs Verificate HELIX in production on its Katana HPC — offloading inference to idle CPU capacity and freeing GPUs for research. Read the case study →
Global reach, sovereign by design.
Verificate ships as licensed pods that run entirely inside a customer’s boundary — which makes it deployable in any jurisdiction, on any sovereign cloud, fully air-gapped. We are building the partner network to take that everywhere it is needed.
We’re building the team behind sovereign AI.
Verificate is at the point where the technology is proven and the demand is real. We are growing deliberately, and we want people who want to own hard problems end-to-end — not a cog in a large machine. If that is you, these are the areas we are investing in:
Low-latency inference runtimes, state projection engines, CPU performance work.
Decision Transformers, typed state representations, inference-level safety.
Sovereign, regulated and government accounts — from first call to production.
Direct briefings, no sales layer.
Enterprise evaluation, sovereign procurement, investor diligence or a strategic partnership — you talk to the people building the product.
- Enterprise evaluation & deployment architecture
- Government / sovereign AI procurement
- Technology & distribution partnerships
- Investor diligence