IQStack

Start with One Spark. Scale When You're Ready.

Always-on AI agents running on your desk. Scale to a cluster when you're ready. Powered by NVIDIA DGX Spark.

Start with Spark Talk to Us

The Problem

You Should Control Where Your AI Runs

Your documents are everywhere — except where you need them
Most AI platforms force you into cloud-only — no choice, no control
Your data shouldn't leave your building for AI
Enterprise AI platforms cost millions — and take months

The Journey

From Desk to Data Center

A natural growth path. Start small, prove the value, then expand.

01

SPARK

Put AI on Your Desk

1× DGX Spark — 128GB unified memory, 1 PFLOP FP4

Knowledge Assistant — ask your documents anything in Thai + English
Local AI Engine — everything runs on your hardware, zero cloud dependency

1 agent · up to 3 data sources · Thai + English bilingual

Start a Pilot

02

CLUSTER

Give AI to Your Teams

2–4× DGX Spark clustered — 256–512GB unified memory

Everything in Spark, plus:

Agentic Data Mining — AI agents explore your data autonomously
AI Eval — monitor quality, detect hallucinations, enforce guardrails

Multiple agents · up to 10 data sources · custom fine-tuning · 405B parameter models

Scale With Us

03

ENTERPRISE

Run Your Business on AI

GPU cluster, DGX Station, or cloud deployment

Everything in Cluster, plus:

Multi-Agent Orchestration — agent swarms coordinating across departments
Custom Compliance Agents — industry-specific regulatory monitoring
24/7 Managed Operations — dedicated engineer, SLA-backed support

Contact Us

What's Inside — Spark Starting Point

Open-Source AI Stack on DGX Spark

Start here — one device on your desk. Prove the value with real data before scaling to a GPU cluster. This is not a fixed architecture — every component can be swapped, added, or replaced to fit your needs.

Layer

Component

Role

Hardware

DGX Spark

GB10 Superchip, 128GB, 1 PFLOP

Security

NemoClaw

Enterprise-ready guardrails, audit, and access control

Models

Your choice

Runs any open model locally, zero per-query cost

Data

PostgreSQL + pgvector

Vector search and structured data storage

Knowledge

Document AI

Reads your documents and finds answers

AI Engine

Fast AI

Makes AI respond quickly and efficiently

Dashboard

Grafana

Monitoring and observability

Interface

Open WebUI

Chat UI for end users

Agents

OpenClaw via NemoClaw

Always-on, self-evolving agents

Total software license cost: ฿0 — all open source

Outgrown Spark? Scale to GPU Cluster or Cloud.

Same software stack, more compute. Multi-GPU clusters, 405B+ parameter models, unlimited agents — the architecture stays the same, only the hardware grows.


Capability	Spark	Cluster	Enterprise
AI Agents	1	Multiple	Unlimited
Data Sources	Up to 3	Up to 10	Unlimited
Custom Fine-Tuning	—	✅	✅
Model Size	Up to 70B	Up to 405B	405B+
Knowledge Assistant	✅	✅	✅
Local AI Engine	✅	✅	✅
Agentic Data Mining	—	✅	✅
AI Eval & Guardrails	—	✅	✅
Multi-Agent Orchestration	—	—	✅
Custom Compliance Agents	—	—	✅
24/7 Managed Operations	—	—	✅
Enterprise SLA	—	—	✅

Optional: Cloud Intelligence Layer

100% Local. Cloud When You Choose.

Your platform runs entirely on your hardware. Add Claude for the tasks that need advanced reasoning — routed through NemoClaw's privacy layer.

Your query enters NemoClaw Privacy Router

90% of queries

Handled locally by Nemotron

Free · Instant · Fully private

10% of queries

Sensitive data stripped, routed to Claude

Advanced reasoning · Logged · Auditable

→ Your data policy controls the cloud

→ Every request in audit log

→ Turn off anytime

Basic

Standard reasoning

Pro

Advanced reasoning

Builder

Code + API access

Add Claude Intelligence

Optional: Enterprise Extensions

Ready to Extend Your IQStack?

Add IBM watsonx to any stage. Your DGX Spark keeps running as the local foundation — IBM extends it with enterprise-grade governance, data integration, process automation, and optimization.

AI Models

watsonx.ai

Governance

watsonx.governance

Data Integration

watsonx.data

Automation

watsonx Orchestrate

Optimization

IBM CPLEX

Add IBM watsonx

How We Deliver

The INFOZENSE Method — Applied to Your IQStack

The same proven methodology that powers our consulting — now applied to your on-premise AI platform.

1

Assess

Evaluate your data readiness and identify the highest-value use case

Strategic Consulting →

2

Build

Deploy DGX Spark, install the stack, fine-tune models on your data, build custom skills

Solution Development →

3

Measure

Monitor agent quality with AI Eval, track ROI, prove business impact

Performance Analytics →

4

Scale

Add Sparks, expand agents, optimize continuously

Operational Excellence →