PHASE 1 PREVIEW / PRIVATE AI INFRASTRUCTURE

OWN THE
INTELLIGENCE.

Verité AI Core is in Phase 1 product development. The configurations below are design targets for private AI systems running local models, governed agents, scheduled work, and professional workloads.

192–576GBaggregate GPU capacity, topology disclosed
Source-pinnedagent and connector contracts
Customer-controlledlocal inference infrastructure
AI CORE SYSTEMS

Purpose-built for serious local inference.

Phase 1 status. AI Core hardware has not yet been built and is not currently available to purchase or ship. Configurations and pricing are planning targets subject to prototype, supply-chain, performance, safety, and manufacturing validation.

CORE X2

Two brains. One machine.

$49,995 planning target
192GB aggregate GPU memory

Professional multi-GPU platform, ECC system memory, high-speed NVMe storage, and CORE OS readiness.

CORE ULTRA

The datacenter comes home.

$159,995 planning target
576GB aggregate GPU memory

Professional multi-GPU platform, ECC system memory, high-speed NVMe storage, and CORE OS readiness.

GPU memory figures are aggregate installed capacity. Effective model parallelism depends on topology and inference-backend support.

CORE OS

See the whole system. Keep the boundaries.

One tenant-scoped surface for machines, GPUs, models, agents, schedules, memory indexes, connectors, policies, usage, and receipts.

Enter the control-plane preview →
Integration discoverySOURCE-PINNED
Tenant data boundaryENFORCED
Production adaptersGATED
AI-CORE workspace runtimeISX BLOCKED
Hardware telemetryNOT CONNECTED
Control-plane modeDEVELOPMENT
AGENT COMPUTE

A private computer for every governed agent.

AI Core-managed agents must execute in ISX. Execution is currently unavailable because the internal ISX source does not expose the required workspace lifecycle transport; no alternate runtime is used.

Readiness: ISX_LIFECYCLE_CONTRACT_UNAVAILABLE. This AI-CORE policy does not change the runtimes used by other Verité systems.

ISOLATEDWorkspace boundary by policy
PERSISTENTIdentity separated from ephemeral state
CONTROLLEDScoped tools, networks, and credentials
AUDITABLEReceipts for privileged operations
PROFESSIONAL WORKLOADS

Infrastructure for work that carries consequences.

01

Legal intelligence

Keep sensitive analysis close while governing research, evidence, and approved capabilities.

02

Private engineering

Route coding and review workloads to local models with explicit repository and network policy.

03

Executive operations

Coordinate persistent agents, scheduled work, and durable history from one accountable control plane.

AUTOMATION & PERSISTENCE

Own the compute. Subscribe to the operating layer.

Hardware remains usable without a subscription. Plans add managed scheduling, persistence, history, backups, integrations, and support according to product policy.

CORE Assist

$299/mo
  • 5,000 workflow runs
  • 25 persistent agents
  • 3 users

CORE Enterprise

$2,499/mo
  • 100,000 workflow runs
  • 500 persistent agents
  • 50 users

Software plan names, limits, and prices are Phase 1 planning targets and are not currently offered for purchase.

PRIVATE DEMO

Design your AI infrastructure.

Tell us your workloads, users, and privacy requirements. We’ll help size the system and software layer.

Submitting this form does not create a professional engagement or authorize access to any system.