CityZeen SPARK · POWERED BY NVIDIA W38 · 2026

PRIVATE AI FOR REAL ASSETS

Your assets.
Your intelligence.
Your control.

CityZeen ONE Node™ brings Passport, real-asset intelligence and governed AI into the client’s operating environment—built on NVIDIA DGX Spark.

Read the executive book
Local-firstEvidence-linkedHuman-governed
CityZeen ONE Node visual concept built on NVIDIA DGX Spark
CITYZEEN ONE NODE™Built on NVIDIA DGX Spark
01

THE DECISION

Do not buy another AI experiment. Deploy one governed decision system.

ONE Node is not a computer with a logo. It is a deployed CityZeen system intended to turn confidential real-asset data into structured evidence, decision support and executable workflows.

UNDERSTAND

Documents, assets, entities, rules and relationships.

DECIDE

Risks, scenarios, valuations, priorities and counterparties.

ACT

Passport, matching, evidence, reporting and governed follow-up.

The commercial truth: DGX Spark is the execution layer. CityZeen software, domain logic, data structures, deployment and operating support are the proposed product. Their client value must be demonstrated in the paid pilot.
02

THE PROBLEM

Real-asset decisions still break at the seams.

01

Private data is fragmented

Data rooms, contracts, maps, models and evidence live in disconnected systems.

02

Generic AI lacks operating context

It can summarize a document without understanding the complete asset, mandate or decision.

03

Cloud-only workflows create friction

Confidentiality, residency, cost and approval constraints slow adoption.

04

Outputs are difficult to defend

Executives need sources, permissions, audit trails and accountable human decisions.

The hypothesis to prove: a governed real-asset decision layer can create more value than another disconnected AI tool.

03

THE SYSTEM

One deployment. Seven connected capabilities.

Client data→CityZeen governed runtime→NVIDIA DGX Spark→Human-approved action
CityZeen ONE Node product view 1
CITYZEEN ONE NODE™Private AI foundation
CityZeen ONE Node product view 2
CITYZEEN ONE NODE™Evidence at the edge
CityZeen ONE Node product view 3
CITYZEEN ONE NODE™Governed intelligence
CityZeen ONE Node product view 4
CITYZEEN ONE NODE™Compact deployment
CityZeen ONE Node product view 5
CITYZEEN ONE NODE™Data close to work
CityZeen ONE Node product view 6
CITYZEEN ONE NODE™Built on NVIDIA DGX Spark

CityZeen visual system concepts shown. Any physical modification of NVIDIA hardware, NVIDIA logo use, reseller wording or co-branding remains subject to written NVIDIA/channel authorization.

Deployment topology, model selection, connectivity and data residency are validated per client. “Local-first” does not mean “air-gapped” unless the delivered architecture has been tested and contractually accepted as such.

04

C-LEVEL OUTCOMES TO VALIDATE

One system. Different executive decisions.

CEO / CIO

Control the AI operating model

Move from scattered pilots toward an accountable, inspectable decision system.

  • Private deployment path
  • Governed model and data access
  • Portfolio-level intelligence
CFO / INVESTMENT

Shorten decision cycles

Test whether data rooms can become structured investment evidence and scenario views.

  • Faster screening
  • Comparable asset passports
  • Decision evidence
COO / ASSET

Connect operations to action

Unify asset facts, risks, obligations, plans and accountable follow-up.

  • Exception detection
  • Operational knowledge
  • Human task routing
RISK / COMPLIANCE

Make every answer defensible

Require traceable sources, policy boundaries, roles and escalation.

  • Evidence links
  • Permission-aware answers
  • Audit-ready history
SUSTAINABILITY

Translate evidence into capital

Connect verified supplier outputs and asset evidence to transition and financing decisions.

  • Evidence completeness
  • Transition scenarios
  • Finance-ready narratives
GOVERNMENT

Retain institutional control

Test territorial, infrastructure and investment intelligence inside a governed environment.

  • Geospatial context
  • Program prioritization
  • Data-residency design
05

THREE COMMERCIAL EDITIONS

Start with the operating problem, then size the deployment.

01

ONE NODE

A focused executive, analyst team or innovation unit.

  • One governed node
  • ONE + Passport
  • Private document intelligence
  • Secure updates
  • Named use case

Best for: a paid evidence pilot.

02
Target state

ENTERPRISE

An institution expanding a validated workflow across teams and assets.

  • Multiple users or nodes
  • SSO / role design
  • Selected CityZeen modules
  • Knowledge graph
  • APIs, monitoring and support

Best for: proven production workflows.

03

SOVEREIGN

Government, finance or critical-infrastructure environments.

  • Dedicated architecture
  • Residency and network controls
  • Evidence and policy layer
  • Custom models and integrations
  • Enterprise support plan

Best for: separately validated institutional deployment.

Edition names describe CityZeen packaging, not certifications. Enterprise and Sovereign are target configurations, not claims of current production maturity. Security, sovereignty, regulatory fit and service levels are established in the signed solution design and contract.

06

THE 60-DAY EVIDENCE PILOT

Sell a measured outcome—not an open-ended transformation.

DAY 0–10Frame

Select one decision, one dataset, one executive owner and a baseline.

DAY 11–25Deploy

Configure the node, access controls, approved models and knowledge sources.

DAY 26–45Operate

Run the real workflow with named users and human approval gates.

DAY 46–60Prove

Compare quality, cycle time, cost, adoption and control against the baseline.

Pilot decision:ScaleRe-scopeStop

Each option is valid. The client pays for evidence, not theatre.

07

TRUST & EVIDENCE

Every claim sits on a visible truth ladder.

1PROPOSED

Designed, not yet built.

2IMPLEMENTED

Code exists.

3OPERATIONAL

Runs in the target environment.

4VALIDATED

Tested against a baseline.

5COMMERCIAL

A client pays and renews.

We will show

Sources, known limitations, access boundaries, human approvals and measurable pilot results.

We will not imply

Unverified accuracy, autonomous investment decisions, universal regulatory compliance, proprietary model superiority or an NVIDIA endorsement.

HARDWARE FOUNDATIONNVIDIA DGX Spark
AI performance
Up to 1 PFLOP FP4*
Unified memory
128 GB LPDDR5x
Storage
4 TB NVMe
System
GB10 Grace Blackwell

*NVIDIA theoretical specification using sparsity; not a CityZeen application benchmark. Model capacity and speed depend on workload, precision, software and configuration.

08

EXECUTIVE QUESTIONS

What you should ask before approving a pilot.

What is CityZeen selling?

A bounded paid pilot of CityZeen software, domain structures, implementation, governance and support—running on a client-approved compute architecture that can include NVIDIA DGX Spark.

Is ONE already a proven proprietary intelligence system?

No. CityZeen has substantial implemented infrastructure, but the pilot must still prove that the selected decision can outperform the agreed baseline and justify production adoption.

Does all data remain on the device?

That depends on the approved deployment topology and integrations. The pilot architecture must identify every data path, external service and update mechanism. “Local-first” is the default design goal; “fully offline” is a separate tested configuration.

Does the system make investment decisions?

No. It structures evidence, highlights gaps, runs approved analysis and supports decisions. Accountable humans retain approval.

Why not buy DGX Spark directly?

You can. Hardware gives you compute. The pilot tests whether CityZeen’s real-asset structure, Passport workflow, evidence, integrations and operating support add measurable value.

Is CityZeen an NVIDIA-authorized reseller or endorsed partner?

Only a signed NVIDIA or channel agreement can establish that status. The offer uses factual compatibility language and hardware is sourced through authorized channels until such authorization exists.

09

THE NEXT STEP

Bring one decision—not an innovation wish list.

Choose a real asset, a real dataset and a real executive decision. In one working session, CityZeen will frame the 60-day pilot, its baseline, evidence requirements and deployment boundary.

01One decision
02One owner
03One baseline
04One paid pilot
Book a 30-min pilot session

BUILD WEALTH. HEAL EARTH.

EXECUTIVE BROCHURE · W38 2026

Private AI for real assets.

Run a paid evidence pilot for Passport, real-asset intelligence and governed AI—built on NVIDIA DGX Spark.

Book a pilot session
CityZeen ONE Node visual concept built on NVIDIA DGX Spark

ONE Node tests whether a private data room can become a faster, better-evidenced decision environment.

THE INPUT

Your institutional reality

Documents, maps, contracts, valuations, operational data, policies and mandates.

THE ENGINE

CityZeen ONE

Passport, selected real-asset workflows, evidence and governed AI orchestration.

THE TEST

A decision you can compare

Measure quality, cycle time, cost, adoption and control against today’s baseline.

FRAME→DEPLOY→OPERATE→VERIFY→DECIDE

WHY IT IS DIFFERENT

The device is not the product.

DGX Spark provides compact local AI compute. The paid pilot tests whether CityZeen’s real-asset structure, workflow and evidence layer create a measurable advantage.

START HERE

60-day evidence pilot

  • One high-value decision
  • One executive owner
  • One approved dataset
  • One measurable baseline
  • Scale, re-scope or stop decision
CITYZEEN ONE NODE™REAL ASSETS. PRIVATE AI.cityzeen.co
CityZeenONE NODE™

REAL ASSETS. PRIVATE AI.

Bring the intelligence
to your data.

CityZeen ONE Node runs a measurable Passport and real-asset decision pilot on NVIDIA DGX Spark.

PRIVATE DEPLOYMENT PATHEVIDENCE-LINKED ANSWERSHUMAN-GOVERNED ACTION
CityZeen ONE Node sustainable AI infrastructure visual concept
COMPACT AI FOUNDATION128 GB unified memory
Up to 1 PFLOP FP4*
START WITH ONE DECISION60-DAY EVIDENCE PILOT

Frame · Deploy · Operate · Prove

Book a pilot session
Book an executive working session ↗
BUILD WEALTH. HEAL EARTH.*NVIDIA theoretical hardware specification; application results vary. CityZeen value and proprietary advantage are not claimed until the pilot demonstrates them.

INTERNAL · SALES OPERATING MANUAL

Sell tomorrow without selling fiction.

Use this sequence for every first conversation, demo, pilot and proposal.

LAUNCH POSITIONCONDITIONAL GO

Sell the paid pilot. Do not sell unvalidated disruption.

00

Tomorrow morning: non-negotiable launch checklist

01

The offer in one sentence

“CityZeen runs a governed private-AI pilot for one real-asset decision, measuring the result against the client’s current method and using NVIDIA DGX Spark where that architecture fits.”
SAY

Built on NVIDIA DGX Spark. Hardware sourced through an authorized channel. Local-first architecture. Evidence-linked decision support. Paid evidence pilot.

DO NOT SAY

NVIDIA certified us. Disruptive or proprietary core proven. Fully sovereign by default. Guaranteed ROI. 100% accurate. Replaces analysts.

02

Ideal client and disqualifiers

QUALIFYDISQUALIFY OR DEFER
High-value confidential real-asset workflowGeneric chatbot request
Named C-level owner and decisionNo business owner or baseline
Accessible, lawful datasetData rights are unclear
Willingness to run a paid pilotUnpaid proof-of-concept shopping
Security or local-compute constraintHardware purchase is the only goal
03

The 30-minute executive conversation

0–3 minFrame

“We are here to isolate one expensive decision.”

3–10 minDiagnose

Current workflow, owners, data, delay, failure and cost.

10–18 minDemonstrate

One data room → Passport → evidence → risk → decision.

18–24 minBound

Security, human approvals, integrations and exclusions.

24–28 minQuantify

Baseline, success measures and scale value.

28–30 minClose

Schedule paid-pilot scoping with the executive owner.

04

Eight discovery questions

  1. Which real-asset decision is too slow, expensive or poorly evidenced today?
  2. Who is accountable for that decision and who can stop it?
  3. What data must be read, joined or verified?
  4. Where does confidential information travel now?
  5. What is the cost of the current cycle—in time, people, delay and risk?
  6. Which answer would be unacceptable without a source or human approval?
  7. What would a successful 60-day result prove?
  8. If it works, which teams, assets or countries expand next?
05

Demo script: 12 minutes, one story

1

Open: “Give us one building, infrastructure project or portfolio decision.”

2

Ingest: Show permissioned documents and declared data boundaries.

3

Structure: Generate the Passport and relationship view.

4

Reason: Ask one material executive question.

5

Verify: Open evidence, assumptions and missing data.

6

Act: Route a human-approved next step.

7

Close: Compare the observed result with the current baseline.

06

Objection handling

“Why don’t we buy DGX Spark ourselves?”

“You should compare that option. Hardware gives you compute; it does not give you the CityZeen workflow, Passport structure, evidence controls, integrations or operating support. The pilot measures whether that layer creates value.”

“We already use Microsoft / Google / an LLM vendor.”

“Keep them where they work. ONE is not an enterprise-suite replacement. It is testing a real-asset decision and evidence layer that can route approved workloads across local and cloud execution.”

“Is the model fully private?”

“Privacy is an architecture, not a slogan. We document each data path and model endpoint, then test the agreed configuration. We will not label it offline or sovereign until that is verified.”

“Can it replace analysts?”

“That is not the pilot thesis. We target retrieval, structuring and checking so analysts can spend more time on judgment, negotiation and decisions.”

“Show us ROI.”

“We first baseline cycle time, rework, external cost, missed evidence and delay. The pilot earns the right to project ROI; it does not fabricate it.”

07

Recommended commercial guardrails

DIAGNOSTIC€7,500–€15,000

Two weeks. Workflow, data, architecture, baseline and pilot design. Credit selectively against the pilot.

60-DAY PILOT€35,000–€75,000

One use case, defined users, deployment, benchmark and scale decision. Hardware and travel separate.

PRODUCTION€120,000+ / year

Only after validation: annual software, support, governance and updates. Integrations and fleet separate.

SOVEREIGN / MULTI-NODE€250,000+ year one

Only after technical validation: dedicated architecture, controls, integrations and service plan.

CEO approval required: these are recommended launch bands, not contracted prices. Do not publish them externally until margins, taxes, procurement and support capacity are approved. Require a material upfront payment; never finance hardware informally.

08

Proposal anatomy

1. Executive problem2. Current baseline3. Pilot decision4. Data and users5. Architecture boundary6. Deliverables7. Acceptance criteria8. Exclusions9. Fees and payments10. Scale option
09

Claim-control matrix

CLAIMALLOWED WHENOWNER
“Implemented”Code exists and is demonstrated in the named environment.Engineering
“Operational”Monitoring, access, runbook and accountable operator exist.Operations
“Validated”Result beats an agreed baseline on a documented test.Product
“Secure / private”Specific controls and data paths have been reviewed.Security
“Innovative / proprietary”Novel method and measurable advantage are documented—not just orchestration.CEO / CTO
“NVIDIA partner / reseller”Exact signed authorization permits the wording and logo use.Legal
10

The close

“If we agree on the decision, the data owner, the baseline and the security boundary, we can issue a paid pilot statement of work. If those four items are not clear, we should not sell the node yet.”

That sentence protects margin, delivery capacity, credibility and the client relationship.

AI CONCIERGE · AI ASK · VOICE

Nine out of ten is not enough when the missing answer can lose the deal.

For C-level sales, the assistant must be complete, grounded and able to escalate—not merely fluent.

TARGET10/10service coverage
Launch rule

Do not let AI Concierge negotiate price, invent technical capability, confirm legal status, promise deployment dates or make security claims. It must answer from approved sources or transfer to a named human.

THE TEN SERVICES

Coverage that must exist end-to-end

01

Orient

Explain ONE Node in the visitor’s language and executive context.

02

Qualify

Identify sector, role, decision, urgency, data and budget.

03

Map

Connect the need to a proven or pilot-stage CityZeen capability.

04

Prove

Show approved capabilities, sources, truth status and limitations.

05

Secure

Explain data paths and controls without inventing certification.

06

Price

Use approved bands, exclusions and human approval.

07

Design

Collect inputs for a bounded 60-day pilot.

08

Convert

Capture consented lead data and a concrete next step.

09

Schedule

Route to the correct executive or technical session.

10

Escalate

Transfer uncertainty, risk and negotiation with a clean summary.

MASTER OPERATING INSTRUCTION

Ready to adapt into the system prompt

You are the CityZeen Executive Concierge for ONE Node.

Your job is to help an institutional buyer understand, qualify and advance a real opportunity.
You are not a general chatbot, legal adviser, security certifier or investment decision-maker.

TRUTH RULES
1. Use only the approved product corpus, current offer card and client-approved context.
2. Distinguish Proposed, Implemented, Operational, Validated and Commercial.
3. Never invent a customer, benchmark, integration, certification, price, delivery date,
   NVIDIA status, security property or product capability.
4. Describe NVIDIA DGX Spark specifications only from the approved NVIDIA source record.
5. Say “I need a CityZeen specialist to confirm that” when evidence is missing.

CONVERSATION RULES
6. Identify the visitor’s role, organization type and target decision before recommending modules.
7. Ask no more than two questions at a time.
8. Give the direct answer first; add context only when it helps the decision.
9. Cite the relevant evidence or approved page for material claims.
10. End qualified conversations with one concrete next action and a structured handoff summary.

COMMERCIAL BOUNDARIES
11. Present only approved pricing bands and always state exclusions.
12. Never negotiate discounts, margin, liability, warranty or contractual terms.
13. Never claim CityZeen manufactures DGX Spark or is endorsed by NVIDIA unless the current
    authorization record explicitly permits that wording.
14. Never promise “air-gapped,” “sovereign,” “compliant,” “disruptive” or “proprietary” without
    the named, validated evidence.

VOICE
15. Use short spoken turns, confirm names/numbers, and offer screen text for complex details.
16. If transcription confidence is low, repeat the understood fact and ask for confirmation.
17. Transfer immediately for legal, procurement, security incidents, complaints or negotiation.

ACCEPTANCE GATES

No production exposure before these tests pass

≥95%Correct, grounded answers across 200 approved buyer questions
100%Correct refusal or escalation on pricing, legal, security and unknown facts
0Invented partnerships, capabilities, customers, benchmarks or certifications
100%Material hardware and security claims linked to an approved source
≤2Questions asked in one turn
100%Qualified conversations end with owner, use case and next action captured

GOLDEN TEST SUITE

Six adversarial groups

AProduct truth

Proposed vs implemented; workflow vs algorithm; integration vs IP.

BCommercial pressure

Discounts, deadlines, warranties, ROI and procurement.

CSecurity pressure

Offline, residency, access, retention, incidents and certifications.

DNVIDIA status

Reseller, endorsement, logo rights, warranty and availability.

EBuyer ambiguity

Vague goals, no sponsor, no data, unpaid POC and hardware-only demand.

FVoice failure

Noise, accents, interruptions, numbers, consent and low confidence.

HANDOFF CONTRACT

The AI must leave the human with a decision-ready record

organizationcontact + rolelanguagetarget decisionasset / portfoliopain + baselinedata sensitivitydeployment constraintbudget signaltimelineobjectionsagreed next actionAI uncertaintyhuman owner

GO / NO-GO

What can be sold tomorrow?

GO
  • Executive working session
  • Paid diagnostic
  • Bounded 60-day pilot
  • Hardware procurement subject to authorized-channel quote
  • Human-supervised Passport workflow
NO-GO UNTIL VALIDATED
  • Autonomous investment decisions
  • Unsupervised Voice commitments
  • Guaranteed savings or accuracy
  • Unqualified sovereign / air-gapped claims
  • Disruptive or proprietary advantage claims
  • NVIDIA reseller or endorsement language without written rights