Order Fulfillment and Customer Service
Operations
March 26, 2026
Documents the six-stage digital order fulfillment pipeline, ownership matrix, SLA framework, service programs, and feedback channels for GPU compute provisioning.
Fulfillment Workflow
Six-Stage Digital Provisioning Pipeline
1
Inquiry & Qualification
1–5 days / instant
2
Contract & Agreement
2–12 weeks / instant
3
Infrastructure Provisioning
5 min – 2 weeks
4
Connectivity & Access
Network activation
5
Go-Live Verification
Joint benchmark
6
Ongoing Delivery
SLA 99.9%+ uptime
Track 1
Pay-per-Use
API self-service portal. Automated provisioning.
<5 min provisioning
Track 2
Reserved Capacity
Bilateral contract. Dedicated partition.
24–72 hr provisioning
Track 3
Private AI Factory
Full 288-GPU NVL72 cluster. Isolated network.
1–2 wk provisioning
Order Handling
Ownership Matrix
| Stage | Owner | Accountability |
|---|---|---|
| Inquiry & Qualification | Founder / Head of Sales | Client vetting, technical fit |
| Contract & Agreement | Founder + Legal | Negotiation, compliance docs |
| Provisioning | CTO / Head of Ops | GPU allocation, network config |
| Connectivity | Head of Ops / DevOps | Credentials, endpoint activation |
| Go-Live | CTO + Client Lead | Joint benchmark, SLA start |
| Ongoing Delivery | Head of Operations | Monitoring, incidents, reporting |
| API Layer (PPU) | Automated Platform | Self-service, metering, billing |
Accuracy & Response
Incident Response SLA
P1
Service down — complete outage
≤ 15 min
P2
Degraded — performance below SLA
≤ 1 hour
P3
Non-critical — advisory or minor issue
≤ 4 hours
Fulfillment Quality
Target Error Rates (Pre-Commercial)
SLA Breach (downtime)
<0.1%
Billing Discrepancy
<0.5%
Config Mismatch
<1%
Connectivity Failure
1–3%
Security Isolation Gap
<0.01%
Overall target:
<2% exception rate
Service Delivery
Network Connectivity Stack
Fiber Optic
Primary — dedicated fiber to Tbilisi IX and international transit. Included in CAPEX.
Starlink LEO
Backup — satellite redundancy for connectivity resilience. ~$500/mo per terminal.
VPN / Encrypted Tunnel
Secure overlay for all client access. Negligible cost (software).
Cross-Connect
Direct physical interconnect for colocation scenarios. Negotiated per contract.
Network transit cost:
$2K–5K/mo per facility
(included in OPEX)
Emergency Handling
Burst & Emergency Provisioning
Burst Capacity
Additional GPUs from 20% PPU reserve pool. Provisioned within hours. Billed at PPU rate ($5–7/GPU-hr) regardless of contracted track.
Hardware Failure Swap
Hot-spare activation from on-site reserve. Target: ≤4 hours GPU, ≤2 hours network. No cost to client — covered under SLA.
Maintenance & Service
Included Service Programs
24/7 Monitoring & Incident Response
Continuous automated monitoring: GPU, network, cooling, power. On-call 24/7. P1 response ≤15 min. All tiers.
Proactive Maintenance
Monthly scheduled windows (7-day notice). Firmware, security, cooling, network. Zero-downtime rolling updates.
Hardware Replacement
On-site spares. 2–5% annual GPU failure rate (6–15 units at 288 GPUs). NVIDIA warranty covers 3 years.
Performance Advisory
Quarterly workload review for Private/Reserved clients. Utilization optimization recommendations. CTO-delivered.
Returns & SLA Credits
Contract Protection Framework
99.9%
Uptime Target
1.5x
Credit Multiplier (P1)
<10%
Annual Churn Target
≤0.5%
SLA Credit Exposure
Private AI Factory
12-month minimum. Early exit: remaining-term payment obligation.
Reserved Capacity
1–3 year term. 3-month notice + forfeited prepaid balance.
Pay-per-Use
No minimum term. Immediate cancellation. Final invoice only.
Complaints & Feedback
Escalation Path
Operations Team
First response ≤15 min
→
CTO
Unresolved within SLA
→
Founder
Commercial / contractual
Feedback Collection Channels
Quarterly
Business Review (QBR)
Structured performance & satisfaction meeting with client leads
Monthly
Performance Report
Automated uptime, utilization, SLA compliance with feedback form
Continuous
Direct Executive Access
Client contacts CTO or Founder directly for any concern
Per Incident
Post-Incident Review
Root cause analysis shared after every P1/P2 event
Annual
NPS Survey
Formal satisfaction measurement across all client segments
Continuous
API Layer Feedback
In-platform mechanism for PPU self-service users
