What Verificate does

Verificate Pty Ltd is an Australian sovereign AI infrastructure company founded in 2025 by Craig Atkinson and headquartered in Sydney, NSW. The company builds three core products on a unified substrate: HELIX, a CPU-native inference engine that runs any open model (Granite 4.0, Gemma 4, GPT-OSS 120B) on standard AMD, Intel, IBM Power, or Apple Silicon hardware with inference-level hallucination prevention via the proprietary CORTEX per-token truthfulness manifold; a Decision Transformer (EAV-DT) that owns deterministic decisions via argmax over complete forward-projected state with bit-exact reproducibility and per-decision audit lineage; and a governed MCP validation server that gates AI output against reality across generation, planning, and analysis workflows. The platform is designed for zero network egress, running entirely inside customer infrastructure, making it viable for HIPAA, PCI-DSS, GDPR, and defence-grade air-gapped environments.

The technology stack is built from first principles on llama.cpp + GGML with a C++/ONNX production engine, NUMA-aware threading, and architecture-native SIMD kernels that deliver 12.9x TTFT advantage over stock Ollama on commodity CPUs. Verificate generates revenue through two primary streams: quote-based HELIX pod licensing for enterprise and government sovereign deployments, and a $30/month self-serve MCP server subscription (with 30-day free trial). The go-to-market combines founder-led direct sales into regulated industries, government, and academia with an IBM Business Partner channel relationship. The company has one named foundation customer (UNSW Sydney, in production on Katana HPC since May 2026), minor deployments at University of Queensland, and operates with a 1-10 person team without disclosed external funding. SOC 2 Type II attestation is in progress.

Name

Verificate

Legal name

Verificate Pty Ltd

Website

https://verificate.ai

Company type

Private

Founded year

2025

Operating status

Operating

Headcount range

1–10 employees

Short description

Verificate Pty Ltd builds sovereign AI infrastructure (HELIX CPU inference, Decision Transformer, MCP validation server) for regulated enterprises, government, and research institutions requiring zero-egress, deterministic AI on commodity hardware.

Ownership category

akta.pro rank

Product category

Sovereign AI Infrastructure

NAICS

Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (5182), Software Publishers (513210), Computer Systems Design Services (541512), Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518)

SIC

Services-Prepackaged Software (7372), Services-Computer Integrated Systems Design (7373)

akta.pro primary industry

Model Hosting, Serving & Inference Platforms (HDAAACAB)

akta.pro secondary industries

Model Deployment, Serving & Inference Platforms (HDAAABAF), Confidential AI & Privacy-Preserving ML (federated learning, MPC, HE, TEEs) (HDAAAKAI), On-Device Inference Runtimes & SDKs (mobile/embedded) (HDAAAJAB)

Keywords

• Sovereign AI infrastructure
• CPU inference engine
• Deterministic decision transformer
• AI hallucination prevention
• Governed MCP validation

Headquarters

HQ city

Sydney

HQ country

Australia

HQ region

Oceania

Offices2 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Technology or R&D, Personnel, Marketing or Sales, Operations

Revenue model

1. HELIX Pod Licensing: Licensed inference pods deployed inside customer infrastructure. Standard economics: a single commodity CPU node serves complete, schema-valid transactions at a fraction of per-call margin of a public API. No egress, no recurring cloud costs. Licensed for academic or commercial deployment in customer's own environment.
2. MCP Server Subscription: Monthly subscription for governed MCP validation server access after 30-day free trial. Single-user key, rate-limited and abuse-protected. Volume and academic pricing available on request.
3. Academic and Volume Licensing: Academic and volume pricing available on request for universities, research institutions, and organizations requiring multiple deployments or special licensing terms.

Pricing tiers

Go-to-market motion3 records

Distribution channels3 records

Marketing channels6 records

Core offering

Verificate builds and sells a sovereign AI infrastructure platform composed of three products — HELIX (a CPU-native inference engine with built-in hallucination prevention), Decision Transformer (deterministic, auditable decisioning over a full forward-projected state), and Verificate MCP (a governed validation server that gates AI output against reality). The products are licensed as on-premise pods deployed inside customer infrastructure (Kubernetes, OpenShift, bare VM, or air-gapped) with zero network egress, and the MCP server is additionally offered as a $30/month subscription with a 30-day free trial. Customers include regulated enterprises, government, and universities such as UNSW Sydney.

Product overview

Verificate offers a unified sovereign AI infrastructure platform comprising three core products: HELIX (CPU inference engine with built-in hallucination prevention), Decision Transformer (deterministic, auditable decisioning over complete forward-projected state), and MCP (governed validation server that gates AI output against reality). All three products share a sovereign, zero-egress architecture that runs entirely inside the customer's boundary on their own CPUs — no GPU procurement, no third-party cloud dependency. The Decision Transformer serves as the deterministic decision-making engine while HELIX provides the inference substrate and MCP adds a governance validation layer.

Differentiator

Problem solved

Functional benefit

Brands

• HELIX: Sovereign CPU inference engine with hallucination prevention built into the inference loop. Runs any open model on standard CPUs.
• Decision Transformer
• MCP (Model Context Protocol)

Products and services

• HELIX HELIX is a hardware-software platform that runs any open model (Granite 4.0, Gemma 4, GPT-OSS 120B) on standard CPUs (AMD EPYC, Intel, IBM Power, Apple Silicon) with hallucination prevention built into the inference loop. It features NUMA pinning, architecture-native SIMD kernels, grammar-masked JSON sampling for 100% schema-valid output, and is licensed as pods for on-premise deployment with zero network egress.
• Decision Transformer Decision Transformer is a domain-scale decisioning system that owns the decision deterministically via argmax over the complete forward-projected state. It is reproducible to the bit with per-decision audit lineage, supports cascading O(changed) recomputation, and self-improves in seconds on CPU. The LLM interprets and narrates but never decides.
• Verificate MCP Verificate MCP is a governed Model Context Protocol server that gates AI output against reality with four tools: generate_code, validate_plan, analyze_code, and validate_ai_output. It catches hallucinated code, ungrounded plans, and integrity failures before they reach production. Available via a 30-day free trial and $30/month subscription, with drop-in MCP config for Claude, Cursor, Windsurf, and any MCP-compatible client.

Quantifiable outcome

• GPU CAPEX avoidance of $1.5–4M per deployment (UNSW case study: 25–40 A100/H100-class GPUs deferred)
• +13 more outcomes

Named customers1 record

Segments4 records

Ideal customer profiles5 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration8 records

AI capability9 records

Feature12 records

Partnerships

Two partnerships are on record, tiered core.

• University of New South Wales (UNSW)coreTechnology or Integration · 1 January 2026UNSW Sydney is a foundation customer running Verificate HELIX in production on its Katana HPC. Pilot began February 2026, converting to production in May 2026. Deployment on Intel-based Katana HPC offloads inference to idle CPU capacity, freeing GPUs for research. Published case study with validated outcomes.
• IBMcoreTechnology or Integration · 1 December 2025Verificate is an IBM Business Partner with access to IBM Fusion HPC infrastructure and Granite model family integration. HELIX runs any open model including Granite 4.0, Gemma 4, and GPT-OSS 120B. Two live HELIX v1.7 pods running on IBM Fusion OpenShift (AMD EPYC) at University of Queensland. Sits alongside existing watsonx.ai workloads rather than replacing them.

Scale indicators18 records

Recent moves6 records

Expansion highlights9 records

Emerging players

• Anyscale: Ray-based AI compute and inference platform addressing enterprise-scale LLM serving; comparable to Verificate on inference cost and performance optimization.
• Together AI: Cloud inference provider for open models with emphasis on cost-efficient serving; competes for the same open-model inference workloads Verificate targets on-prem.
• Modular: AI inference platform (Mojo / MAX) pursuing high-performance CPU and GPU serving; competes with HELIX on throughput, latency, and developer experience for open models.

Broad incumbents

• Hugging Face: Broader AI/ML platform with model hosting, inference endpoints, and open-model ecosystem; overlaps Verificate in serving open models to enterprises.
• Cohere: Enterprise AI platform with private deployment options and deterministic, compliance-focused offerings that overlap Verificate's regulated-industry positioning.
• IBM watsonx.ai: IBM's enterprise AI platform that Verificate complements rather than replaces via the IBM Business Partner relationship; both target sovereign, regulated AI deployments.

Direct peers

• Ollama: Open-source local LLM runner that Verificate benchmarks against directly; both target running models locally on commodity hardware with similar developer ergonomics.
• llama.cpp: Foundational C++ CPU inference project on which HELIX is built; both serve the same CPU-native LLM inference niche and share GGML underpinnings.
• vLLM: High-throughput LLM inference engine whose GuideLLM benchmark harness Verificate uses to validate HELIX; both compete on inference latency and throughput for open models.
• LM Studio: Desktop application for running local LLMs with OpenAI-compatible APIs; overlaps HELIX's local/sovereign inference positioning for developers and small teams.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat6 records

Key risks7 records

Key highlights7 records

Customer concentration

• LinkedInlinkedin.com/company/verificate

Compliance5 records

Revenue estimate

Valuation estimate

Number of profiles

Profiles1 record

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Verificate