DNYANA.CLOUD

Enterprise-grade GenAI platform

Governed, performant, deployable anywhere

🔒 Security-first 📊 Full observability 🌐 Any deployment

Executive Summary

🎯

Total Control

Deploy in your environment, manage policies centrally, audit every request. No shadow AI, no data leakage.

100% visibility into LLM usage

Compliance Ready

SOC2 Type II, HIPAA-ready architecture, GDPR DPA. Evidence export for audits, BAA available.

4 frameworks supported out-of-box
📈

Measurable Performance

Reduce LLM costs 30–50%, improve p95 latency 60%, track spend per department/project in real-time.

ROI visible within 30 days

Current Risks: Unmanaged LLM Usage

HIGH RISK

Data Exposure

  • Employees paste proprietary data into ChatGPT, Claude
  • Customer PII sent to third-party LLM APIs
  • No DLP controls, no audit trail
  • Regulatory exposure (GDPR, HIPAA, PCI)
MEDIUM RISK

Cost Drift

  • Decentralized API keys → no budget control
  • Teams pick expensive models by default
  • No caching, redundant calls
  • Finance discovers $50K/month AI bill
MEDIUM RISK

Shadow Tools

  • Engineering uses OpenAI, marketing uses Anthropic
  • Inconsistent vendor SLAs and compliance terms
  • Fragmented monitoring and support
  • Procurement nightmare, contract sprawl
HIGH RISK

Audit Gaps

  • No immutable logs of LLM interactions
  • Cannot prove data handling for SOC2/ISO
  • No model-level access controls
  • Audit findings = delayed certification

Platform Overview: Day 1 Capabilities

🔗 Access Layer

  • Unified API: OpenAI-compatible, all major LLMs
  • Web UI: White-label playground + chat
  • SDK Support: Python, Node, Go, Java
  • Streaming: SSE, WebSocket

🛡️ Security & Identity

  • SSO/SAML: Okta, Azure AD, Google Workspace
  • RBAC: Org/project/user roles
  • API Keys: Scoped, time-limited, revocable
  • MFA: TOTP, push, biometric

⚡ Intelligent Routing

  • SmartRoute: Cost/latency/quality optimizer
  • Fallbacks: Auto-retry with backup models
  • Rate Limiting: Per-user, per-project quotas
  • Caching: Semantic dedup, 30–40% savings

🔍 Observability

  • Request Traces: Full latency breakdown
  • Cost Tracking: Real-time per-org/project
  • Audit Logs: Immutable, tamper-evident
  • Alerts: Slack, email, PagerDuty, webhooks

📜 Governance

  • Model Catalog: Approved models only
  • PII Redaction: Auto-scrub sensitive data
  • Retention Policies: 0/30/90/365 days
  • Exception Workflow: Request high-risk models

📊 Compliance

  • Evidence Export: CSV, JSON for auditors
  • DPA/BAA: Templates included
  • Data Residency: US, EU, APAC regions
  • Encryption: TLS 1.3, AES-256, BYOK

Deployment Choices

Criterion SaaS (Dnyana-managed) Private VPC (Your AWS/Azure/GCP) On-Premises (Your datacenter)
Time to Deploy < 1 day (API keys ready) 3–5 days (Terraform/Helm) 2–4 weeks (dependencies, airgap)
Data Residency US/EU/APAC regions Your chosen region Your datacenter
Network Isolation Shared tenancy, encrypted Private VPC, your firewall rules Air-gapped / private network
Ops Responsibility Dnyana (24/7 SRE) Shared (Dnyana support + your ops) Your team (Dnyana assists)
Compliance Posture SOC2, GDPR-ready, BAA available Inherit your VPC compliance + ours Full control for HIPAA, PCI, FedRAMP
Upgrade Cadence Weekly (automated) Monthly (coordinated) Quarterly (scheduled with you)
Cost Model Token-based + SaaS fee Token-based + infra passthrough Annual license + token-based

Recommendation: Start with SaaS for pilot, migrate to Private VPC or On-prem for production if required by security/compliance posture.

Security Architecture

Your Application TLS 1.3 Encrypted API Gateway SSO/SAML, MFA Policy Engine RBAC, PII Redaction SmartRoute Model Selection + Governance Check OpenAI Encrypted API Anthropic Encrypted API Others Encrypted API Data Layer AES-256 at-rest BYOK/KMS Support Observability Immutable Logs Audit Trail Secrets Mgmt Vault, AWS Secrets Mgr Rotate LLM API keys Network Isolation Private Link/VPN Zero internet egress Security Features: ✓ TLS 1.3 encryption in transit (all connections) ✓ AES-256 encryption at rest (logs, config, secrets) ✓ BYOK/KMS support (AWS KMS, Azure Key Vault, HashiCorp Vault) ✓ Network isolation (Private Link, VPN, air-gapped) ✓ Secrets rotation (automated LLM API key rotation) ✓ Zero-trust architecture (verify every request) ✓ DDoS protection (Cloudflare, rate limiting, backpressure)

Model Governance

Model Risk Matrix

Model Risk Tier Use Cases Restrictions Approval
GPT-4o-mini, Claude 3.5 Haiku Low General Q&A, summaries, tagging No PII, 30-day retention Auto-approved
GPT-4o, Claude 3.5 Sonnet Medium Analysis, code review, reports PII redaction required Manager approval
o1, o1-pro (reasoning) High Strategic planning, research Zero retention, audit all CISO approval
Fine-tuned / custom models Restricted Special projects only Case-by-case review Security committee

Exception Workflow

1

Request

User requests access to restricted model via UI

2

Review

Manager/CISO receives notification with context

3

Approve/Deny

Decision logged, access granted or denied

4

Audit

All usage tracked in immutable log

Compliance Posture

SOC2 Type II

✓ Certified
  • Security, availability, confidentiality controls
  • Annual audit by Big 4 firm
  • Report available under NDA
  • Continuous monitoring for control drift

HIPAA

Ready
  • Architecture supports HIPAA requirements
  • BAA (Business Associate Agreement) available
  • PHI encryption, access controls, audit logs
  • Breach notification procedures in place

GDPR

✓ Compliant
  • DPA (Data Processing Agreement) standard
  • EU data residency (Frankfurt, Dublin)
  • Right to deletion, data portability
  • Privacy by design, minimal data retention

ISO 27001

In Progress
  • Information security management system
  • Certification expected Q3 2025
  • Gap analysis complete, controls implemented
  • Audit scheduled with accredited body

Compliance Tooling

Evidence Export: One-click export of logs, policies, access records for auditors (CSV, JSON, PDF)
DPA/BAA Templates: Pre-negotiated agreements ready for your legal review
Control Matrix: Map our controls to your framework (SOC2, NIST, ISO, CIS)
Audit Support: Dedicated compliance engineer for audit prep and questionnaires

Observability & Auditing

🔍 Request Tracing

{
  "request_id": "req_7f3a9b2c",
  "timestamp": "2025-10-27T14:32:18Z",
  "user": "alice@company.com",
  "org": "acme-corp",
  "model": "gpt-4o",
  "input_tokens": 245,
  "output_tokens": 512,
  "cost_usd": 0.0189,
  "latency_ms": 1240,
  "status": "success",
  "pii_redacted": true,
  "policy_checks": ["approved_model", "no_phi"]
}

📊 Real-time Dashboards

  • Cost per org/project/user
  • Latency p50/p95/p99
  • Model usage distribution
  • Error rates and throttling

🚨 SLO Alerts

  • Budget threshold exceeded
  • Latency SLA breach
  • Unusual usage patterns
  • Policy violations

🔗 SIEM Integration

  • Splunk, Datadog, New Relic
  • Webhook connectors
  • Syslog export
  • Custom dashboards

🔒 Tamper-evident Logs

  • Cryptographic hashing
  • Append-only storage
  • Audit trail integrity
  • Compliance-grade logging

Performance & Cost

Latency (p50)
Direct: 180ms
Dnyana: 110ms
39% faster
Latency (p95)
Direct: 850ms
Dnyana: 320ms
62% faster
Cost per 1K tokens
Direct: $0.0250
Dnyana: $0.0165
34% savings
Cache Hit Rate
Direct: 0%
Dnyana: 38%
38% fewer API calls

Monthly Cost Comparison (10M tokens)

Scenario Direct LLM API Dnyana.cloud Savings
Mixed workload (GPT-4o + mini) $2,500 $1,650 -$850 (34%)
+ Dnyana platform fee $500
Total $2,500 $2,150 -$350 (14%)

*Savings increase with volume; typical enterprise customers save 30–50% at scale due to caching + smart routing.

Integration & Identity

🔐 Identity Providers

Okta
Azure AD
Google Workspace
Auth0
OneLogin
PingIdentity

SAML 2.0, OAuth2, OIDC support. SCIM provisioning for automatic user sync.

👥 Roles & Permissions

Role Permissions Typical User
Viewer Read logs, view dashboards Auditors, analysts
User Call API, use approved models Developers, end users
Manager Approve exceptions, manage quotas Team leads, PMs
Admin Configure policies, manage users Platform team, SRE
Owner All permissions, billing CTO, CISO

🔌 API & Webhooks

  • REST API: Full management API for automation
  • GraphQL: Flexible queries for observability data
  • Webhooks: Real-time events (budget alerts, policy violations)
  • Terraform Provider: Infrastructure-as-code for deployment
  • Kubernetes Operator: GitOps-friendly on-prem deployment

Data Residency & Retention

🌍 Available Regions

🇺🇸 United States

us-east-1 (Virginia), us-west-2 (Oregon)

🇪🇺 Europe

eu-central-1 (Frankfurt), eu-west-1 (Dublin)

🌏 Asia Pacific

ap-southeast-1 (Singapore), ap-northeast-1 (Tokyo)

🏢 On-Premises

Your datacenter (air-gapped or VPN)

Data sovereignty: All request data, logs, and configs stay in your chosen region. No cross-border transfers without explicit consent.

📅 Retention Policies

Data Type Default Retention Options Deletion
Request prompts 30 days 0 / 7 / 30 / 90 / 365 days Auto-purge + manual
LLM responses 30 days 0 / 7 / 30 / 90 / 365 days Auto-purge + manual
Audit logs 365 days 90 / 365 / 2555 days (7 yrs) Manual only (compliance)
Metrics / analytics 90 days 30 / 90 / 365 days Auto-aggregate
User PII Until deleted GDPR: right to deletion Immediate on request

🗑️ Deletion Workflow

User-initiated: Self-service deletion via UI/API → data purged within 24 hours → confirmation email

GDPR request: Submit request → identity verification → deletion within 30 days → certificate of deletion

Automated: Retention policy expires → data auto-purged → logged in audit trail

Customer Success: 4-Week Pilot Plan

Week 1: Foundation
  • Kickoff meeting: Align on goals, success criteria, stakeholders
  • Environment setup: SaaS sandbox provisioned, SSO configured
  • Access granted: 5–10 pilot users onboarded
  • Training: 2-hour workshop on platform capabilities
Week 2: Integration
  • Use case selection: Pick 1–2 high-value use cases
  • API integration: Dev team integrates SDK into app
  • Policy setup: Configure model catalog, retention, redaction
  • Monitoring: Dashboards configured, alerts enabled
Week 3: Testing
  • Functional testing: Verify use cases work end-to-end
  • Performance testing: Measure latency, cost, cache hit rate
  • Security review: CISO reviews logs, policies, architecture
  • Feedback session: Gather user feedback, iterate
Week 4: Evaluation
  • Results review: Cost savings, latency improvements, user satisfaction
  • Security sign-off: CISO approves for production
  • Procurement: Commercial terms finalized
  • Production plan: Timeline for VPC deployment if needed

Ongoing Support

Standard: Email support, 24-hour response, community Slack
Premium: Dedicated CSM, 4-hour response, monthly QBR
Enterprise: Dedicated Slack channel, 1-hour response, on-call support, TAM

Commercials: Tiers, SLAs, Pricing

Pricing Tiers

Tier Base Fee Token Pricing Support SLA
Pilot $500/month 20% markup on LLM costs Email, 24-hour response 99.5% uptime
Professional $2,500/month 15% markup on LLM costs Dedicated CSM, 4-hour response 99.9% uptime
Enterprise Custom (typically $10K+/month) 10% markup + volume discounts TAM, 1-hour response, on-call 99.95% uptime + credits

Enterprise Add-ons

  • Private VPC deployment: +$5K/month (infra passthrough)
  • On-premises license: $100K/year (includes support)
  • Professional services: $250/hour (integration, training, custom dev)
  • Extended retention: +$1K/month per TB (beyond 365 days)
  • Custom model integration: $10K one-time + $1K/month hosting

SLA Commitments

Metric Professional Enterprise
Uptime 99.9% (43 min/month) 99.95% (22 min/month)
API Latency (p95) < 500ms < 350ms
Support Response 4 hours (business) 1 hour (24/7)
Credits (breach) 10% monthly fee 25% monthly fee

Case Study: Global Financial Services Company

[Company Name Redacted]

Industry: Financial Services | Size: 15,000 employees | Region: Global

Challenge

  • Legal and compliance teams using ChatGPT for document review → data exposure risk
  • $75K/month in untracked LLM spending across 12 departments
  • No audit trail for SOC2 / ISO 27001 compliance
  • CISO blocked all public LLM access, productivity dropped 40%

Solution

  • Deployed Dnyana.cloud in Private VPC (AWS us-east-1)
  • SSO integration with Okta, RBAC policies for 200 users
  • PII redaction enforced, zero-retention for sensitive docs
  • Approved model catalog: GPT-4o-mini (general), Claude 3.5 Sonnet (legal review)

Results (90 days)

$32K
Monthly cost (down from $75K)
-57% cost reduction
180ms
p95 latency (vs 820ms direct)
-78% latency improvement
100%
Audit coverage
SOC2 audit passed
200
Active users
Productivity restored

"Dnyana.cloud gave us the control and visibility our CISO needed, while unblocking our teams. We passed our SOC2 audit with zero findings related to LLM usage." — VP of Information Security

Next Steps

🔒

Security Review

Share architecture docs, SOC2 report (under NDA), complete security questionnaire

Timeline: 1–2 weeks
🧪

Sandbox Access

Provision SaaS sandbox, onboard 5–10 users, provide API keys and documentation

Timeline: 1 day
🚀

4-Week Pilot

Integrate use cases, configure policies, measure results, get CISO sign-off

Timeline: 4 weeks
📋

Procurement

Finalize commercial terms, execute MSA/DPA/BAA, plan production deployment

Timeline: 2–4 weeks

Contact Information

Sales: enterprise@dnyana.cloud
Security: security@dnyana.cloud
Support: support@dnyana.cloud
Documentation: docs.dnyana.cloud
QR: dnyana.cloud