The A.R.I. Lab — Artificial Relative Intelligence: private AI infrastructure, enterprise automation, healthcare AI agents
AI Evaluation • Integration • Private Deployment

Find where AI creates value.
Then make it work in production.

NEC Consult helps financial, pharmaceutical, healthcare, government, and enterprise organizations evaluate AI, select the right models and architecture, integrate solutions with existing systems, and move from experimentation into production.

When privacy, control, customization, or economics require dedicated infrastructure, we build and test solutions in the A.R.I. Lab—our private NVIDIA Blackwell AI engineering environment.

Book a 20-Minute AI Fit Call → 🎙 Talk to Dr. Franklin — Live Explore the A.R.I. Lab

Bring one workflow or AI question. We'll tell you whether it is technically practical, what architecture makes sense, and whether NEC Consult is the right partner.

4 RTX PRO 6000 Blackwell GPUs 384 GB dedicated GDDR7 GPU memory 2 NVIDIA DGX Spark nodes 256 GB unified memory 100 Gb/s internal fabric

Up to 18 PFLOPS FP4 sparse vendor-rated peak · 640 GB total accelerator-accessible memory across the cluster · full specifications and methodology

Live from the A.R.I. Lab

Meet Benjamin Franklin.
Ask him about the lab.

This is not a prerecorded animation. Benjamin hears your question, processes it through a private AI pipeline running on A.R.I. Lab hardware, generates a response, and speaks back in real time through a synchronized 3D avatar.

Prefer not to use your microphone? Text mode is built in — just type. Benjamin is an AI-generated historical interpretation: his appearance, voice, and responses are synthetic and are not authentic recordings or statements by Benjamin Franklin. The live avatar pipeline is operated from the actual A.R.I. Lab near West Chester, Pennsylvania.

1 · You speak or type

Live microphone or text input in the browser.

2 · The lab listens & thinks

Real-time speech recognition and private local model reasoning on A.R.I. Lab hardware — no public AI model APIs.

3 · He answers, live

Synthetic voice generation with real-time lip-sync and 3D animation, streamed back to your browser.

The Story Behind A.R.I.

A.R.I. stands for Artificial Relative Intelligence. The lab was named in memory of Michael Friedberg's brother, Ari, whose life and work inspired its creation. An optional memorial experience — an authorized AI representation created from family-owned recordings — tells the story in his own voice.

Hear the Story of A.R.I. from Ari →

Services

From AI opportunity to production system

We help organizations determine what AI should do, which technology should be used, how it should connect to existing systems, and what is required to operate it reliably in production.

Discover

AI Evaluation & Use-Case Discovery

Identify where AI can create measurable value, assess feasibility and risk, and prioritize the right first projects.

Evaluate

Model & Architecture Benchmarking

Test frontier, open-source, private, and hybrid approaches against the organization's actual data, workflows, security requirements, and operating constraints.

Connect

Integration & Automation

Connect AI to existing applications, documents, data sources, APIs, workflows, and human-review processes.

Operate

Production AI Engineering

Design for security, scalability, monitoring, failure recovery, performance, governance, and ongoing operations.

Industries

AI engineering for data-sensitive and operationally complex organizations

Financial Services

Sensitive financial environments

Private AI, model evaluation, document intelligence, workflow automation, and integration for sensitive financial environments.

Pharma & Life Sciences

Regulated research & documentation

AI for proprietary research, regulated documentation, internal knowledge, operational workflows, and controlled deployment.

Healthcare

Operations & back office

AI agents and automation for documentation, compliance, intake, EVV, billing, onboarding, and back-office operations.

Enterprise

Enterprise AI integration

Use-case discovery, model benchmarking, RAG, agents, system integration, and production architecture for organizations adopting AI.

Engineering pedigree

AI is new. Mission-critical engineering isn't.

NEC Consult's engineering leadership brings nearly three decades of experience designing, scaling, administering, and supporting consequential enterprise systems across financial markets, pharmaceuticals, healthcare, Fortune 500 companies, and the U.S. federal government.

Prior systems experience includes work supporting organizations such as Susquehanna, CooperNeff/BNP Paribas, Merrill Lynch, Wachovia, Merck, SmithKline Beecham, McNeil Consumer & Specialty Pharmaceuticals, and the U.S. Department of Veterans Affairs.

Today, that same production discipline is applied to AI evaluation, integration, automation, and deployment.

Results

We validate against real workflows—not hypothetical demos.

Selected healthcare automations already operate against real business workflows. Model, retrieval, and agent systems are evaluated using defined benchmarks, test suites, and representative workloads before we recommend production deployment.

Regulated Healthcare Operations

AI-assisted EVV exception detection, compliance workflow automation, and administrative process automation developed against real Pennsylvania home-care operations.

Private Knowledge Systems

Document intelligence grounded in private organizational data, with source attribution and automated verification designed to reduce unsupported answers.

Domain Model Engineering

In-house fine-tuning, evaluation, benchmarking, and deployment on private NVIDIA Blackwell infrastructure.

AI Agents & Automation

Multi-agent systems that automate document-heavy, compliance-heavy, and operational workflows while routing exceptions and consequential decisions to human review.

The infrastructure

Real compute. Honest numbers.

Vendor-rated peak
Up to 18 PFLOPS
FP4 sparse vendor-rated peak — NVIDIA published ratings for 4× RTX PRO 6000 Blackwell + 2× DGX Spark (GB10).
Measured in our cluster
2.5 PFLOPS
Measured FP8 dense cluster throughput, benchmarked end-to-end on the actual cluster.
Note: these figures represent different precisions and sparsity assumptions and are not directly comparable. Full methodology and per-node results are available on the technical detail page.
The A.R.I. Lab

Inside the A.R.I. Lab

A private cluster, built and operated in-house.

A.R.I. — Artificial Relative Intelligence — is the private AI engineering environment behind NEC Consult. It provides the infrastructure for model evaluation, private inference, fine-tuning, retrieval systems, voice AI, and agentic workflow development.

Actual A.R.I. Lab private AI command center near West Chester, Pennsylvania

Inside the A.R.I. Lab — actual private AI engineering environment operated by NEC Consult near West Chester, Pennsylvania.

Designed for Sensitive Data Environments

Private model execution, isolated networking, controlled data flows, encrypted storage, access controls, audit logging, and human-review paths can be designed around the requirements of each engagement.

A Partner, Not a Portal

Clients work directly with the engineer designing and operating the system—from architecture and evaluation through integration and production deployment.

Private and Hybrid Options

Use public frontier models, open-source models, private infrastructure, or hybrid architectures based on the workload—not a predetermined technology stack.

Predictable Capacity

Where appropriate, infrastructure capacity can be allocated under predictable engagement terms instead of uncontrolled per-token consumption.

About

Nearly three decades of enterprise engineering—now applied to AI

Michael Friedberg has spent nearly three decades designing, building, scaling, administering, and supporting mission-critical enterprise systems across financial markets, pharmaceuticals, healthcare, Fortune 500 organizations, and the U.S. federal government.

Through NEC Consult and the A.R.I. Lab, he now applies that production engineering experience to AI evaluation, integration, automation, and deployment.

The lab's name carries its origin. A.R.I. — Artificial Relative Intelligence — is named in memory of Michael's brother, Ari. Michael built the lab in his memory, and its earliest production systems now serve and protect the company Ari founded.

The technology changed. The engineering requirements did not: scalability, reliability, security, data integrity, performance, integration, monitoring, and production operations.

Get started

Start with one real workflow

Bring one process, one AI question, or one integration challenge. We'll help determine whether the opportunity is technically practical, which architecture makes sense, and what a focused pilot should prove.

Prefer email? mj@necconsult.ai  ·  (484) 467-7660
Or explore a focused 30-day pilot, scoped to your workflow.