IT Director Guide: AI Infrastructure Management

As an IT Director overseeing AI infrastructure, you manage the operational reality of running AI services at scale — capacity planning, vendor relationships, budgets, SLAs, and change processes. Keeptrusts gives you the operational data and controls needed to manage AI as a governed enterprise service rather than a collection of ungoverned experiments.

Use this page when

You are planning AI infrastructure capacity (users, requests/day, gateway sizing)
You need to manage multiple LLM vendor relationships and track provider performance
You are forecasting AI service budgets and negotiating SLAs
You want to implement change management processes for AI governance configurations
You need to run AI as a governed enterprise service rather than ad hoc experiments

Primary audience

Primary: Technical Leaders (IT Directors, Heads of IT Operations)
Secondary: Procurement, CTOs, DevOps Engineers, Finance

Capacity Planning

Current State Assessment

Before planning capacity, baseline your current AI usage through the Console and API:

# Get current usage volume
curl -H "Authorization: Bearer $API_TOKEN" \
  "https://api.keeptrusts.com/v1/events?since=30d&group_by=gateway"

# Identify peak usage patterns
curl -H "Authorization: Bearer $API_TOKEN" \
  "https://api.keeptrusts.com/v1/events?since=7d"

The Console Dashboard provides visual trends for request volumes, active users, and policy trigger rates.

Capacity Planning Model

Dimension	Current	6-Month Projection	Planning Action
Active users	Measure via events	+50-100% with AI adoption wave	Provision gateway capacity
Requests/day	Measure via events	2-3x growth expected	Load test gateway cluster
Teams onboarded	Count gateways	All engineering + business teams	Template and onboarding plan
LLM providers	Count in config	Add 1-2 for redundancy	Provider evaluation
Storage (events)	Measure via API	Linear growth with usage	Event retention policy

Gateway Infrastructure Sizing

Deployment Size	Users	Requests/Day	Gateway Instances	Resources
Small	1-50	Under 5K	1	2 CPU, 4GB RAM
Medium	50-500	5K-50K	2-3	4 CPU, 8GB RAM each
Large	500+	50K+	5+	8 CPU, 16GB RAM each

Vendor Management

LLM Provider Portfolio

Manage multiple LLM providers as a portfolio with Keeptrusts providing the governance abstraction:

pack:
  name: it-director-providers-1
  version: 1.0.0
  enabled: true
providers:
  targets:
  - id: openai
    provider: 
    secret_key_ref:
      env: OPENAI_API_KEY
  - id: anthropic
    provider: 
    secret_key_ref:
      env: ANTHROPIC_API_KEY
  - id: azure-openai
    provider: 
    base_url: https://your-instance.openai.azure.com
    secret_key_ref:
      env: AZURE_OPENAI_API_KEY
policies:
  chain:
  - audit-logger
policy:
  audit-logger:
    immutable: true
    retention_days: 365
    log_all_access: true

Vendor Performance Tracking

Use Keeptrusts event data to track provider performance:

# Provider comparison — usage volume and policy triggers
curl -H "Authorization: Bearer $API_TOKEN" \
  "https://api.keeptrusts.com/v1/events?since=30d&group_by=provider"

# Export provider performance data for vendor review
kt export create \
  --type events \
  --format csv \
  --since 90d \
  --description "Quarterly vendor performance review"

Vendor Review Checklist

Review Item	Source	Frequency
Usage volume per provider	Console Events + API	Monthly
Cost per provider	Console Usage	Monthly
Policy trigger rate by provider	Event analytics	Quarterly
Provider uptime and reliability	Gateway health logs	Monthly
Contract renewal timeline	Vendor management tool	Quarterly
Security and compliance certifications	Vendor documentation	Annually

Budget Forecasting

AI Spend Components

Cost Category	Tracking Method	Forecast Input
LLM API usage	Console Cost Center	Historical trend + growth rate
Gateway infrastructure	Infrastructure monitoring	Capacity plan
Keeptrusts platform	License management	User/team count
Internal team costs	HR / Finance	Headcount plan
Training and onboarding	Project tracking	Adoption roadmap

Pulling Cost Data for Forecasting

# Monthly cost breakdown by team
curl -H "Authorization: Bearer $API_TOKEN" \
  "https://api.keeptrusts.com/v1/events?since=30d&group_by=gateway"

# Quarterly cost trend data
kt export create \
  --type events \
  --format csv \
  --since 90d \
  --description "Quarterly cost trend for budget forecast"

Budget Governance Through Policy

Enforce per-team spending limits to prevent budget overruns:

policies:
  - name: team-cost-cap
    type: cost_limit
    monthly_limit: 5000
    action: block
    enabled: true

The Console Cost Center shows real-time spend against budget for each team and provides alerts when teams approach their limits.

SLA Management

Internal AI Service SLAs

Define SLAs for the AI governance platform as an internal IT service:

SLA Component	Target	Measurement
Gateway availability	99.9% uptime	Health check monitoring
Policy enforcement latency	Under 50ms added latency	Gateway performance logs
Event ingestion	Under 5 seconds end-to-end	Event timestamp analysis
Escalation response	Under 30 min for P1	Console Escalation queue
Configuration changes	Under 15 min to deploy	Config change audit log
Export completion	Under 30 min for standard exports	Export job tracking

Monitoring SLA Compliance

# Verify gateway health (available as a health check endpoint)
kt doctor

# Check event pipeline latency
kt events list --since 1h --limit 10

The Console Dashboard provides real-time indicators for gateway health and event processing.

SLA Reporting

# Generate SLA compliance report data
kt export create \
  --type events \
  --format csv \
  --since 30d \
  --description "Monthly SLA compliance data"

Change Management

Change Categories for AI Governance

Change Type	Risk Level	Approval	Process
Policy threshold adjustment	Low	Team lead	Config change + validate
New policy addition	Medium	IT Director	Review + test + deploy
Provider addition/removal	Medium	IT Director + Security	Vendor review + config change
Gateway infrastructure change	High	Change Advisory Board	Full change process
Major version upgrade	High	Change Advisory Board	Staging test + phased rollout

Change Process with Keeptrusts

Request: Document the change and its business justification
Validate: Test the configuration change before deployment:

kt policy lint --file updated-policy.yaml

Approve: Route through appropriate approval chain
Deploy: Apply the change through Console or Git-linked configuration
Verify: Confirm the change is active and functioning:

kt events list --since 1h --limit 10

Audit: The Console Audit Log automatically records all changes with who, what, and when

Rollback Procedures

If a change causes issues, roll back using Git-linked configurations (revert the commit) or through the Console by reverting to the previous policy configuration.

Change Audit Trail

The Console Audit Log captures:

Policy configuration changes with before/after comparison
User access modifications
Gateway key rotations
Provider configuration updates
Export and administrative actions

# Pull recent changes for change review meetings
curl -H "Authorization: Bearer $API_TOKEN" \
  "https://api.keeptrusts.com/v1/events?since=7d"

Operational Runbook

Daily Operations

Task	Tool	Owner
Check gateway health	`kt doctor` + Console Dashboard	Operations team
Review escalation queue	Console Escalations	SOC / governance team
Monitor cost trends	Console Cost Center	FinOps

Weekly Operations

Task	Tool	Owner
Review SLA metrics	Export reports	IT Director
Analyze usage trends	Console Dashboard	IT Director
Review pending changes	Change management tool	IT Director

Monthly Operations

Task	Tool	Owner
Vendor performance review	Export data	IT Director
Budget reconciliation	Cost Center exports	Finance + IT Director
Capacity review	Event volume trends	IT Director
Policy effectiveness review	Event analytics	Governance team

IT Director Workflow with Keeptrusts

Task	Frequency	Tool
Infrastructure health review	Daily	Console Dashboard + `kt doctor`
Budget tracking	Weekly	Console Cost Center
Vendor performance review	Monthly	Event exports by provider
Capacity planning	Quarterly	Event volume analysis
SLA reporting	Monthly	Export reports
Change Advisory Board prep	Per change	Console Audit Log

Success Metrics for IT Management

Metric	Target	Source
Service availability	99.9% uptime	Gateway health monitoring
Budget variance	Within 5% of forecast	Cost Center vs. budget
Change success rate	> 95% without rollback	Change records
SLA compliance	> 99% of targets met	SLA tracking reports
Team onboarding time	Under 2 business days	Onboarding tracker
Vendor satisfaction	All contracts renewed on time	Vendor management

For AI systems

Canonical terms: Keeptrusts, capacity planning, vendor management, budget forecasting, SLA management, change management, infrastructure sizing
Key surfaces: Console Dashboard, Console Usage, Events API (group_by=gateway, group_by=provider), Console Settings, Console Configurations
Commands: kt export create, kt events list, kt doctor
Sizing guide: Small (1-50 users, under 5K req/day, 1 instance), Medium (50-500 users, 2-3 instances), Large (500+ users, 5+ instances)
Config: multi-provider providers block, cost_limit policy, Git-linked config sync for change management
Best next pages: Architecture Overview, DevOps Guide, Procurement Guide

For engineers

Baseline current usage: GET /v1/events?since=30d&group_by=gateway for volume assessment
Track provider performance: GET /v1/events?since=30d&group_by=provider for vendor review data
Export performance data: kt export create --type events --format csv --since 90d --description "Quarterly vendor review"
Gateway sizing: 2 CPU/4GB for under 5K req/day, 4 CPU/8GB for 5K-50K, and 8 CPU/16GB for 50K+
Verify health: kt doctor before any capacity change

For leaders

Capacity planning model projects 2-3x request growth as AI adoption accelerates — plan gateway infrastructure ahead of demand
Multi-provider portfolio managed through Keeptrusts gives negotiating leverage and eliminates single-vendor risk
Usage reporting provides actual spend data (not estimates) for accurate budget forecasting by team, provider, and model
Change management for AI configurations uses Git-linked sync — policy changes follow PR review, approval, and audit trail
SLA tracking through event data (latency, error rates, uptime) enables evidence-based vendor accountability

Next steps

Deploy infrastructure: DevOps Guide
Vendor evaluation framework: Procurement Guide
Architecture patterns: Architecture Overview
Configuration management: Configuration Management

Use this page when​

Primary audience​

Capacity Planning​

Current State Assessment​

Capacity Planning Model​

Gateway Infrastructure Sizing​

Vendor Management​

LLM Provider Portfolio​

Vendor Performance Tracking​

Vendor Review Checklist​

Budget Forecasting​

AI Spend Components​

Pulling Cost Data for Forecasting​

Budget Governance Through Policy​

SLA Management​

Internal AI Service SLAs​

Monitoring SLA Compliance​

SLA Reporting​

Change Management​

Change Categories for AI Governance​

Change Process with Keeptrusts​

Rollback Procedures​

Change Audit Trail​

Operational Runbook​

Daily Operations​

Weekly Operations​

Monthly Operations​

IT Director Workflow with Keeptrusts​

Success Metrics for IT Management​

For AI systems​

For engineers​

For leaders​

Next steps​