Industrial AI Integration Platform

Analyze Your Splunk Data
Without Leaving Your Network

Splunk Hosted AI (| ai) keeps your data inside Splunk — no external egress. Hybrid PII masking for Claude, OpenAI, Gemini, and Ollama. Zero-egress by design for OT/IIoT environments — KVKK & GDPR privacy-by-design.

Discrete Manufacturing
Process Manufacturing
Energy Management
Utilities Management
Machine Builders (OEM)
Supported AI Providers
Anthropic Claude OpenAI Google Gemini Ollama NVIDIA NIM
| ai
Splunk Hosted AI — Zero Egress
100%
Data Never Leaves Network
6+
LLM Providers
1254
Automated Tests
How It Works

Secure AI Integration Flow

Every request passes through the masking engine — sensitive data never reaches the AI API.

1
Your App
ERP, SCADA,
Splunk, ThingWorx
→
2
Cedrix Gateway
Central control
point
→
3
Mask PII
Sensitive fields
anonymized
→
4
AI Model
Claude · OpenAI · Gemini
Ollama · NVIDIA NIM
→
5
Unmask
Original context
restored
Live Demo

Data Masking — Try It

Edit the text below and press Mask.

Input — Raw Text
Output — Text Sent to API
Masking result will appear here...
Features

Built for Every Need

🛡 Hybrid Data Masking
Operator name, badge no, supplier code, OT asset tag
OT tag mapping engine
In-memory processing — never written to disk
Regex + NLP hybrid engine
Session-based de-anonymization
autoMaskEnabled flag support
⚡ Model Catalog
Anthropic Claude Opus / Sonnet / Haiku
OpenAI GPT-4o / GPT-4o Mini
Google Gemini Pro / Flash
Ollama — local / air-gap models
NVIDIA NIM — Llama, Mixtral
Single interface — all providers
Cost-based intelligent routing
runStream · extractToolCalls API
📊 Splunk Integration
Every prompt & response logged
Token & cost tracking
Anomaly detection & alerting
Per-user reporting
Integrates with existing infrastructure
Native SPL query support
🔒 Splunk Hosted AI
Data stays inside Splunk — zero egress
Foundation-Sec-1.1-8B (cybersecurity)
OpenAI GPT-OSS 20B / 120B models
Runs inside the | ai pipeline
Requires Splunk AI Toolkit 5.7+
Splunk Cloud & On-Prem support
📋 Template Analysis
Save an SPL template, ask natural language
2-step pipeline: fetch rows → embed in AI
Real row values delivered to the model
Inline SPL editor with live preview
Successful queries auto-saved as templates
Turkish / English natural language support
Use Cases

Industry-Specific AI Solutions

🏭 Discrete Manufacturing
OEE analysis — customer code masked, machine data sent to AI
PLC/SCADA log interpretation — ThingWorx & Ignition integration
Quality control anomaly detection — product ID anonymized
Work order optimization — customer info masked before AI processing
🧪 Process Manufacturing
Reactor parameter optimization — process secrets protected
Safety incident analysis — confidential process data removed
Batch recipe analysis — supplier codes masked for AI recommendations
Process efficiency — DCS/SCADA data processed via Cedrix
⚡ Energy Management
Fault prediction — account numbers masked, equipment data processed
Demand forecasting — anonymous regional consumption analysis
Grid anomaly detection — Splunk log data processed via Cedrix
Renewable integration — generation forecast modeling
🏢 Utilities Management
Compressor, generator, cooling tower efficiency — operator identity masked
Predictive maintenance — IoT sensor data routed to AI via Cedrix gateway
Energy consumption optimization — consumption point info anonymized
Alarm management — alert text analyzed without sensitive location data
⚙️ Machine Builders (OEM)
Field telemetry analysis — customer site info masked before AI
Remote diagnostics assistant — serial number and customer info stripped
Warranty analysis — failed component data processed without customer code
Product development — field feedback anonymized for AI analysis
Power Features

Three Capabilities That Set Cedrix Apart

From setup to analysis, from scheduling to memory — all in one platform.

⚙️ Smart Setup Wizard
Production-ready in 5 minutes

Step-by-step guidance on first launch. API keys, Splunk connection, masking rules, and your first test query are configured automatically. Re-runnable — updating settings is just as easy.

1 API Keys — Claude, OpenAI, Gemini, Ollama, NVIDIA NIM — all providers entered and connection-tested from one screen.
2 Splunk Connection — Host, port, and token auto-validated. Existing index list discovered; relevant indexes selected.
3 Masking Rules — Line code, product code, equipment ID, supplier code — quick setup via template picker.
4 Test & Validate — First query runs, masking preview is shown on screen, and confirmed.
5 Go Live — Configuration saved bound to your HWID. Ready on next launch — no setup from scratch.
⏱️ Scheduler
Let analyses wait for you

Schedule your Splunk query and AI analysis prompt. A ready report on your desk every morning — before you finish your coffee. Independent schedulers can be defined for different lines.

Cron Scheduling — Daily, weekly, or custom cron expression. Different schedules per equipment group.
Query + Prompt — Splunk SPL and AI analysis prompt defined and executed together on one screen.
Auto Delivery — Analysis output automatically sent via email or Splunk alert. Format customizable.
Line-based Tracking — Independent schedulers for different production lines, equipment groups, and shifts.
Run History — Status, token cost, and error logs stored for every run.
Schedulers + New
LINE-A Motor Report Daily 07:00 Active
Energy Analysis Mon 08:30 Active
Weekly OEE Summary Fri 17:00 Pending
🧠 Memory
AI that remembers the past

Cedrix remembers previous analyses, anomaly findings, and user preferences. Start with "M-19 since last week..." Memory is HWID-bound — cannot be transferred to another machine.

Result Memory — Past analysis outputs stored. Automatic context built for comparative queries.
User Preferences — Model selection, report format, and frequently used queries remembered per user.
Equipment History — Anomaly history kept per motor and line. "Has this error occurred before?" answered instantly.
Change Analysis — Automatic comparison with previous analysis for "What changed this week?"
HWID Security — Memory stored only on the licensed device. Cannot be moved or copied to another machine.
You: Show me last week's analysis for M-19
Cedrix: In the April 21 analysis, a bearing temperature trend anomaly of +280°C was detected for M-19. This week's value: +15°C — a significant drop. Maintenance action appears effective.
Security

Data Sovereignty — by Design

Masking runs entirely in your local environment, in memory. Never written to disk, never transmitted over the network.

KVKK by Design GDPR by Design Air-Gap Support HWID Licensing Offline Operation MCP Tool Support Splunk AI Toolkit 5.7+ CDTSM Forecasting
Frequently Asked Questions

Industrial AI Platform — FAQ

Most common technical and general questions about Cedrix. Full FAQ →

What is Cedrix?
An AI platform that connects to Splunk. Facility managers ask questions in natural language and the platform generates SPL automatically. Splunk + Cedrix deployed as one bundle — no other infrastructure needed.
How does an Industrial AI Platform work?
Cedrix reads sensor data from Splunk, converts the user's question to an SPL query, and delivers results via real-time streaming. With Splunk Hosted AI, data never leaves the network.
Is it GDPR / KVKK compliant?
Yes. With On-Board AI, data never leaves Splunk. For external providers, the hybrid masking engine clears PII in memory before any API call — nothing written to disk.
Which AI providers are supported?
Splunk Hosted AI (Foundation-Sec, GPT-OSS), Claude, OpenAI, Gemini, NVIDIA NIM, and Ollama (local, no internet). All providers managed through the same interface.
Does it work without Splunk?
Cedrix and Splunk are deployed as a single bundle. PostgreSQL can be added as an optional source. Without Splunk, core AI analysis features do not function.
How long does installation take?
Pilot installation ~30 minutes: Splunk AI Toolkit, Cedrix-Setup.exe, MCP connection, first template. Requirements: Splunk 10.x+, AITK 5.7+, Windows 10/11.
Licensing

Flexible License Model

Partner Edition
Splunk Partner / Consultant
All LLM providers
HWID-based offline license
Model Catalog access
Splunk integration
Hybrid masking engine
Partner-priority support
Site Edition RECOMMENDED
Enterprise / On-Site Deployment
All Partner features
Multi-HWID support
Enterprise SLA
On-premises deployment support
Custom model configuration
Reseller / channel support
Sample Outputs

Analyses Generated by Cedrix

Sample reports generated from anonymized data collected in real customer environments.

Cedrix
⌘K MCP ✓ 9 tools ● Connected
S
Analyze my Splunk indexes, show storage sizes and event counts
TOOL | rest /services/data/indexes | fields title disabled currentDBSizeMB totalEventCount
C
📊 SPLUNK INDEXES — ANALYSIS
18
Active Indexes
61.1 GB
Total Size
311.5M
Total Events
7
Empty Indexes
#IndexEventsSizeStatusNote
1_audit163,090,94421.6 GB✅🔴 HIGH — Audit logs
2_internal56,061,9676.4 GB✅Splunk system logs
3_introspection8,426,2049.8 GB✅🟡 System diagnostics
4gmdevices8,407,6460.9 GB✅⭐ MAIN DATA — Engine sensors
5gmsensors9,978,6291.1 GB✅⭐ IIoT Sensor data
6gmstatus334,4620.2 GB✅Engine status data
7main1,644,0900.1 GB✅Default index
+ 11 more indexes (7 empty)…
STORAGE DISTRIBUTION
_audit
35.4%
_introspection
16.1%
_internal
10.5%
gmsensors
1.7%
gmdevices
1.5%
Others
34.8%
CRITICAL FINDINGS
🔴 _audit (21.6 GB) — 35% of total index storage is audit logs. Retention policy must be applied → Estimated savings: 10–15 GB
🟡 _introspection (9.8 GB) — Diagnostics over-verbose. Reduce logging level + 30-day retention recommended.
✅ gmdevices / gmsensors — Current sizes optimal. Continue monitoring.
ACTION RECOMMENDATIONS
PriorityIndexActionSavings
🔴 Critical_auditRetention policy (30 days)10–15 GB
🔴 Critical_introspectionLogging level ↓ + retention5–7 GB
🟡 ImportantEmpty indexesDelete/disable 7 empty indexes~7 MB
🟢 Infogmdevices, gmsensorsContinue monitoring—
🛡 Data sovereignty: Company name, IP addresses, and user information were anonymized by the Cedrix masking engine before reaching the API.
Viewing chat history — click "Return to Today" to type
↵
Full Reports

Browse Sample AI Analysis Reports

Full-page reports produced through the Cedrix masking engine using anonymized industrial data. View in browser or download as PDF.

Report 01 · 2 pages
Gas Consumption Analysis (Nm³/kWh)
24h gas-consumption performance across 28 motors; critical / efficient grouping, variance analysis, action recommendations.
Splunk 28 motors GDPR Masked
View →
Report 02 · 2 pages
24-Hour Performance Analysis
Motor power output, bearing & oil-temperature monitoring, urgent-action list, Splunk infrastructure health.
Performance Temperature Urgent action
View →
Report 03 · 2 pages
Failure Prediction — Machine Learning
Linear-trend model on 30 days of history; 24h failure-risk forecast scores and prioritized action list per motor.
ML 88% confidence Forecasting
View →
Report 04 · 2 pages
Splunk Health Assessment
Splunk infrastructure version, license, indexing rate, data volume and capacity projection — health score GREEN.
Infrastructure Health Capacity
View →

Transform Your Industry with AI

Stay ahead of the competition while keeping your sensitive data protected.

Request Demo → info@cedrix.io