05
SYSTEM 05|GENESIS Architecture

Action Fabric

From advisory to autonomous — the execution layer with guardrails

5
Autonomy Levels
200+
Action Types
50+
Playbooks
30d
Rollback Window
The Problem

The Insight-to-Action Gap

Most cloud management tools stop at telling you what is wrong. They generate dashboards full of recommendations that pile up in backlogs, growing stale before anyone acts. The result? Organizations leave millions on the table — not because they lack insight, but because they lack a safe, graduated path from knowledge to action. Action Fabric closes that gap with five levels of autonomy, each with its own guardrails, approval workflows, and rollback capabilities.

73%
of recommendations go un-actioned
Industry average for cloud optimization tools — recommendations expire before teams can implement them.
6.2 weeks
average time to implement a recommendation
From identification to execution, most organizations take over a month to act on optimization opportunities.
$2.4M
average annual waste from inaction
The cost of delayed action on optimization recommendations for a typical enterprise cloud environment.
Graduated Autonomy

Five Levels of Autonomy

Every organization has a different comfort level with automation. Action Fabric meets you where you are and grows with you — from purely advisory to fully autonomous, with safety guardrails at every level.

L0
📋 Advisory Only
Recommendations with full context, zero execution

The system observes, analyzes, and recommends — but never touches infrastructure. Every suggestion includes complete context: impact analysis, risk assessment, estimated savings, and implementation steps. Perfect for organizations beginning their automation journey or for high-risk environments where human control is paramount.

Capabilities
Generate detailed optimization recommendations with ROI projections
Provide step-by-step implementation guides for each recommendation
Calculate blast radius and risk scores for proposed changes
Prioritize recommendations by savings potential and implementation effort
Create executive-ready reports with visualization of opportunities
Map recommendations to organizational policies and compliance requirements
Track recommendation acceptance rates and time-to-implement metrics
Benchmark current state against industry best practices and peer organizations
Generate what-if scenarios showing outcomes of different action paths
Maintain a recommendation backlog with aging and staleness tracking
L1
🔧 Guided Execution
Step-by-step instructions, human confirms each step

The system prepares every action, pre-validates it, and presents it for human approval one step at a time. Think of it as an expert co-pilot who has already done the research, written the commands, and checked the safety — you just confirm each step. Ideal for teams building confidence in automation or for actions that cross compliance boundaries.

Capabilities
Pre-validate each action step against current infrastructure state
Generate exact CLI commands, API calls, or console steps for each action
Provide rollback commands alongside each execution step
Show real-time preview of expected infrastructure changes before execution
Pause and wait for explicit human confirmation at every decision point
Maintain execution context across multi-step workflows with session persistence
Offer alternative approaches when a step encounters unexpected conditions
Log every human decision for audit trail and compliance documentation
Estimate time and cost impact for each individual step before execution
Support collaborative execution where multiple team members approve different steps
L2
Assisted Workflow
Automated sequences with approval gates at key decisions

The system executes routine steps automatically but pauses at critical junctures for human approval. Approval gates are placed at points where business judgment is needed: before committing spend, before modifying production resources, or before crossing organizational boundaries. This level dramatically reduces toil while maintaining human oversight where it matters most.

Capabilities
Execute multi-step workflows with configurable approval gate placement
Batch related actions for efficient bulk approval by authorized personnel
Automatically handle prerequisite checks and dependency resolution
Implement parallel execution paths for independent action sequences
Provide rich context at each approval gate including impact summary and risk score
Support delegation rules — route approvals to appropriate team based on resource type
Auto-skip approval gates for actions below configurable risk and cost thresholds
Maintain workflow state across approval delays with timeout and escalation policies
Generate approval request notifications via Slack, Teams, email, or PagerDuty
Track approval SLAs and automatically escalate stale approval requests
L3
🤖 Supervised Autonomous
Self-executing within policy bounds, human oversight maintained

The system executes actions autonomously within defined policy boundaries. Humans set the rules — budget limits, resource scopes, change windows, risk tolerances — and the system operates freely within those constraints. Every action is logged, monitored, and reversible. Humans receive real-time feeds and can intervene or pause at any time. Ideal for mature organizations with well-defined cloud governance.

Capabilities
Execute actions autonomously within administrator-defined policy boundaries
Enforce budget guardrails with automatic throttling as limits approach
Respect change window schedules and freeze periods across all actions
Implement progressive rollout strategies for high-impact changes
Maintain continuous health monitoring during and after action execution
Auto-rollback changes that trigger degradation in monitored health metrics
Generate real-time execution feeds for human oversight dashboards
Support exception handling with automatic escalation for edge cases
Coordinate actions across multiple cloud accounts and regions safely
Implement circuit breaker patterns to halt execution chains on repeated failures
L4
🧠 Fully Autonomous
Self-directing within risk and value thresholds, audit-logged

The highest level of automation. The system not only executes but decides what to execute based on continuous analysis of the environment. It identifies optimization opportunities, evaluates risk and reward, and acts — all within strict value and risk thresholds set by the organization. Every decision and action is comprehensively audit-logged. Human oversight shifts from approval to governance: setting policies, reviewing outcomes, and adjusting boundaries.

Capabilities
Continuously scan infrastructure for optimization opportunities without prompting
Self-prioritize actions based on ROI, risk, urgency, and organizational goals
Make autonomous commitment decisions (RI/SP purchases) within approved budgets
Dynamically adjust resource allocations based on predictive demand models
Negotiate with cloud provider APIs for spot pricing and reserved capacity
Implement self-healing workflows that detect and remediate drift automatically
Generate comprehensive audit trails meeting SOC 2, ISO 27001, and FedRAMP requirements
Provide explainable AI reasoning for every autonomous decision made
Implement multi-objective optimization balancing cost, performance, and reliability
Self-tune policy boundaries based on historical action outcomes and organizational feedback
Action Categories

What Action Fabric Can Do

Over 200 pre-built action types organized into categories. Each action type includes validation logic, rollback procedures, blast radius calculations, and audit trail integration out of the box.

📐
Medium25-40%

Resource Right-Sizing

Continuously analyze and adjust compute, storage, and memory allocations to match actual workload requirements.

Downsize over-provisioned EC2, GCE, and Azure VM instances
Right-size RDS, Cloud SQL, and Azure SQL database instances
Optimize EBS, Persistent Disk, and Managed Disk volumes
Adjust container resource requests and limits in Kubernetes
Resize Lambda, Cloud Functions, and Azure Functions memory allocations
Optimize ElastiCache and Memorystore node types
▼ View actions
📊
Low30-60%

Commitment Management

Intelligently purchase, exchange, and manage Reserved Instances, Savings Plans, and committed-use discounts.

Purchase Reserved Instances matching stable workload patterns
Buy Savings Plans optimized for compute or specific service families
Exchange underutilized reservations for better-fitting options
Consolidate reservations across linked accounts for maximum coverage
Schedule commitment purchases to align with renewal cycles
Monitor and alert on expiring commitments before renewal deadlines
▼ View actions
🏷️
Low5-15%

Tagging Enforcement

Ensure every resource is properly tagged for cost allocation, ownership tracking, and compliance reporting.

Auto-tag newly created resources based on creator and project context
Fix missing or incorrect cost-allocation tags across all accounts
Enforce mandatory tag schemas before resource provisioning
Propagate tags from parent resources to child and dependent resources
Generate tag compliance reports by team, project, and environment
Quarantine untagged resources with escalating enforcement actions
▼ View actions
🛡️
Medium10-25%

Policy Enforcement

Automatically enforce organizational standards for resource lifecycle, access patterns, and operational hygiene.

Shutdown idle and unused resources based on utilization thresholds
Enforce instance type restrictions by environment and team
Apply storage lifecycle policies for aging and archival data
Terminate long-running spot instances exceeding cost thresholds
Enforce encryption-at-rest and in-transit requirements
Block deployment of non-approved instance families or regions
▼ View actions
💰
LowAccuracy

Cost Reallocation

Move costs to the correct business units, projects, and cost centers to ensure accurate financial reporting.

Reallocate shared service costs using configurable allocation models
Split multi-tenant resource costs by actual usage metrics
Correct historical cost misallocations with adjustment entries
Automate chargeback and showback calculations monthly
Map cloud costs to internal project codes and P&L structures
Generate departmental cost reports aligned with financial calendars
▼ View actions
🤝
Low8-20%

Vendor Negotiation Support

Generate data-driven negotiation briefs and track vendor commitments for enterprise discount programs.

Generate spend analysis reports for vendor negotiation preparation
Calculate leverage metrics based on growth trajectory and commitment potential
Track EDP/PPA commitment attainment against contractual thresholds
Model scenarios for different discount structures and commitment levels
Monitor vendor compliance with negotiated terms and SLAs
Benchmark pricing against market rates and peer organizations
▼ View actions
🏗️
High20-50%

Architecture Optimization

Recommend and implement architectural changes that fundamentally reduce cost while maintaining or improving performance.

Migrate workloads from VMs to containers or serverless where appropriate
Implement auto-scaling policies based on demand pattern analysis
Consolidate underutilized clusters and reduce infrastructure sprawl
Recommend multi-region optimization to reduce data transfer costs
Identify and eliminate redundant load balancers and NAT gateways
Optimize database architectures with read replicas and caching layers
▼ View actions
🚨
HighProtection

Incident Response

Automatically detect and remediate cost anomalies, spending spikes, and runaway resources before they become budget emergencies.

Auto-detect and alert on abnormal spending patterns in real-time
Quarantine runaway resources that exceed cost velocity thresholds
Automatically terminate cryptocurrency mining and unauthorized workloads
Implement emergency budget caps with automatic resource throttling
Generate incident reports with root cause analysis and prevention plans
Coordinate with security teams when cost anomalies indicate breaches
▼ View actions
MediumCompliance

Compliance Remediation

Automatically detect and fix resources that violate organizational, regulatory, or industry compliance standards.

Remediate resources violating CIS benchmark configurations
Fix non-compliant IAM policies and overly permissive access rules
Enforce data residency requirements by region and jurisdiction
Apply required security group and firewall rule configurations
Ensure logging and monitoring is enabled on all applicable resources
Generate compliance attestation reports for auditor review
▼ View actions
🚧
LowPrevention

Budget Guard Rails

Proactively prevent overspend with intelligent alerting, throttling, and automatic enforcement of budget boundaries.

Set dynamic budget thresholds that adjust with organizational growth
Implement progressive alerting at 50%, 75%, 90%, and 100% of budget
Auto-throttle non-critical workloads when budgets approach limits
Block new resource provisioning when department budgets are exhausted
Generate forecast-based early warnings before budget overruns occur
Implement approval workflows for any spend exceeding threshold amounts
▼ View actions
Low30-65%

Scheduling & Lifecycle

Automate resource scheduling and lifecycle management to eliminate waste from resources running when not needed.

Schedule dev/test environments to run only during business hours
Implement automated weekend and holiday shutdown policies
Manage ephemeral environment lifecycles with TTL-based expiration
Coordinate cross-timezone scheduling for global development teams
Optimize batch processing schedules for spot instance availability
Automate AMI and snapshot lifecycle with retention policy enforcement
▼ View actions
Execution Pipeline

Every Action, Seven Stages

No action bypasses the pipeline. Whether advisory or fully autonomous, every proposed change flows through the same rigorous stages — the only difference is where human approval is required.

1Recommendation💡2Validation🔍3Risk Assessment⚖️4Approval Gate🚪5Execution6Verification7Rollback Plan↩️─── Autonomy-Dependent Routing ───L0-L1: Full Human ControlL2-L3: Selective Approval GatesL4: Auto-Proceed (Policy-Bounded)
💡1. Recommendation

System identifies an optimization opportunity based on continuous analysis of signals and patterns.

🔍2. Validation

Proposed action is validated against current infrastructure state, dependencies, and prerequisites.

⚖️3. Risk Assessment

Blast radius calculation, impact analysis, and risk scoring determine the safety profile of the action.

🚪4. Approval Gate

Based on autonomy level, the action either auto-proceeds or waits for human approval at the configured gate.

5. Execution

The action is executed with real-time monitoring, progress tracking, and automatic health checks throughout.

6. Verification

Post-execution verification confirms the action achieved its intended outcome without negative side effects.

↩️7. Rollback Plan

A tested rollback plan remains active for the configured window, ready for instant activation if issues arise.

Safety & Guardrails

Automation Without Fear

The difference between automation and dangerous automation is guardrails. Action Fabric implements six layers of safety that ensure no action can cause unrecoverable harm — even at the highest autonomy levels.

💥

Blast Radius Calculation

Before every action, the system calculates the potential impact across all dependent services, users, and business processes.

Dependency graph analysis identifies all upstream and downstream services affected
User impact estimation calculates how many end-users could be affected by the change
Revenue impact modeling estimates potential business impact during the change window
Cascading failure simulation tests whether the change could trigger chain reactions
Historical incident correlation checks if similar changes have caused issues before
Multi-dimensional risk scoring combines technical, business, and compliance risk factors
↩️

Rollback Capability

Every action executed by the system is reversible within a configurable time window, ensuring no change is permanent until verified.

Pre-action snapshots capture complete resource state for point-in-time recovery
Automated rollback triggers activate on health check failures or metric degradation
Rollback testing runs in parallel to verify the rollback plan works before execution
Graduated rollback supports partial reversal for multi-step workflow failures
Rollback window is configurable from 1 hour to 30 days based on action criticality
Post-rollback verification ensures the system returns to its exact pre-action state
🚫

Policy Boundary Enforcement

The system enforces hard boundaries on what it can and cannot do, regardless of autonomy level or optimization potential.

Production database modifications require Level 1 (guided) regardless of policy
Cross-account resource deletion is permanently restricted to human-only execution
Network topology changes require explicit approval even at Level 4 autonomy
IAM and security group modifications are restricted to compliance remediation only
Data retention and deletion actions enforce legal hold and regulatory requirements
Cost threshold breakers prevent any single action from exceeding configured spend limits
👤

Human Escalation Triggers

Specific conditions automatically pause automation and escalate to human decision-makers regardless of autonomy level.

Risk score exceeding organizational threshold triggers immediate human review
Unusual patterns in action outcomes trigger anomaly-based escalation
First-time actions on previously untouched resource types require human validation
Actions affecting resources tagged as critical or sensitive always escalate
Conflicting recommendations from different analysis modules trigger human arbitration
Budget impact exceeding per-action thresholds requires financial approver sign-off
🚦

Rate Limiting & Throttling

The system enforces rate limits on action execution to prevent automation runaway and ensure controlled change velocity.

Maximum concurrent actions per account, region, and resource type are enforced
Cool-down periods between related actions prevent cascading automation issues
Daily action budgets limit the total number and scope of changes per 24-hour period
Progressive throttling reduces action rate as daily limits approach
Emergency stop capability halts all automation globally with a single command
Rate limit exceptions require explicit approval with documented justification
🕐

Change Window Awareness

The system respects organizational change management schedules, freeze periods, and maintenance windows.

Integration with change management systems (ServiceNow, Jira) for window coordination
Automatic detection and respect of organizational freeze periods and code freezes
Time-zone aware scheduling ensures actions execute during appropriate local hours
Pre-event freezes automatically pause automation before known traffic spikes
Gradual ramp-up after freeze periods to prevent action queue flooding
Calendar integration for holiday awareness across global team schedules
Live Action Feed

Real-Time Execution Monitor

Watch actions flow through the system in real-time. Every pending approval, in-progress execution, completed optimization, and rolled-back change is visible in one unified feed.

Pending(2)
In Progress(2)
Completed(3)
Rolled Back(1)
completed
2 minutes ago
Right-sized 12 EC2 instances from m5.2xlarge to m5.xlarge
Target: prod-api-cluster (us-east-1)L3🤖 Agent
$4,320/mo
savings
in-progress
5 minutes ago
Purchasing 3-year Compute Savings Plan — $50,000 commitment
Target: Organization (all accounts)L2🤖 Agent
$18,200/yr
savings
pending
8 minutes ago
Schedule weekend shutdown for 47 dev instances
Target: dev-environment (us-west-2)L2🤖 Agent
Awaiting approval from: Platform Team Lead
$6,100/mo
savings
completed
12 minutes ago
Applied missing cost-allocation tags to 234 resources
Target: All accounts (3 regions)L4🤖 Agent
Accuracy
savings
rolled-back
18 minutes ago
Attempted resize of RDS instance db-analytics-prod
Target: analytics-prod (eu-west-1)L3🤖 Agent
Reason: Connection count exceeded safe threshold during resize window
$890/mo
savings
completed
25 minutes ago
Terminated 8 orphaned EBS volumes (0% attachment rate, 90+ days)
Target: staging-account (us-east-1)L4🤖 Agent
$1,240/mo
savings
in-progress
30 minutes ago
Migrating 3 NAT Gateways to NAT Instances for dev environments
Target: dev-networking (us-west-2)L2🤖 Agent
$2,100/mo
savings
pending
35 minutes ago
Convert 15 on-demand instances to spot with fallback
Target: batch-processing (us-east-1)L2👤 Human
Awaiting approval from: FinOps Manager
$3,800/mo
savings
Action Playbooks

Pre-Built Automation Recipes

Battle-tested playbooks that encode proven optimization patterns. Each playbook defines trigger conditions, a sequence of actions, safety checks, and expected outcomes — ready to deploy at any autonomy level.

Weekend Dev Shutdown

$12,400/mo
Trigger: Cron: Friday 19:00 UTC

Automatically stop all development and staging resources every Friday evening, restart them Monday morning. Excludes resources tagged as always-on.

Last run: 3 days ago
Success rate: 99.2%
Playbook Steps
01Snapshot all EBS volumes attached to dev instances for safety
02Stop all EC2 instances tagged Environment:dev or Environment:staging
03Scale down EKS dev node groups to zero nodes
04Pause non-critical RDS dev instances
05Scale down ElastiCache dev clusters to minimum nodes
06Send shutdown summary to #finops-actions Slack channel
07Monday 07:00 UTC: Restart all resources in reverse order
08Verify all services healthy post-restart, alert on failures
▼ View steps

Right-Size Queue

$28,000/mo
Trigger: Continuous (batched hourly)

Continuously monitor resource utilization and queue right-sizing recommendations. Batch similar recommendations for efficient approval and execution.

Last run: 47 minutes ago
Success rate: 97.8%
Playbook Steps
01Analyze 14-day CPU, memory, network, and disk utilization patterns
02Identify instances with sustained utilization below 40% threshold
03Generate right-sizing recommendations with projected savings
04Group recommendations by service team and approval authority
05Submit batched approval requests to designated approvers
06Execute approved right-sizing during next available change window
07Monitor resized instances for 72 hours post-change for degradation
08Auto-rollback any resize that triggers performance alerts
▼ View steps

Commitment Optimizer

$85,000/yr
Trigger: Weekly analysis, monthly execution

Analyze workload stability patterns and automatically purchase Reserved Instances and Savings Plans within approved budget boundaries.

Last run: 6 days ago
Success rate: 100%
Playbook Steps
01Analyze 90-day workload patterns for commitment-eligible stability
02Calculate optimal commitment mix across RI types and Savings Plans
03Model scenarios for 1-year vs 3-year terms with break-even analysis
04Generate purchase recommendations with confidence scores
05Submit high-confidence purchases for automated approval
06Execute commitment purchases during optimal pricing windows
07Monitor commitment utilization weekly and flag underutilized commitments
08Queue exchange recommendations for underperforming reservations
▼ View steps

Tag Compliance Sweep

$15,000/mo (allocation accuracy)
Trigger: Weekly: Sunday 02:00 UTC

Comprehensive weekly audit of all resource tags across all accounts. Auto-fix missing tags where possible, quarantine non-compliant resources.

Last run: 4 days ago
Success rate: 98.5%
Playbook Steps
01Scan all resources across all accounts and regions for tag compliance
02Identify resources missing mandatory tags (Owner, Environment, CostCenter, Project)
03Auto-apply tags where source can be determined from resource metadata
04Generate compliance exceptions for resources requiring manual tagging
05Escalate repeat offenders to team leads with compliance history
06Update tag compliance dashboard and trend metrics
07Apply quarantine tags to chronically untagged resources after 3 warnings
08Generate weekly tag compliance report for FinOps team review
▼ View steps

Anomaly Auto-Response

$40,000/yr (loss prevention)
Trigger: Real-time (event-driven)

Automatically detect and respond to cost anomalies, unauthorized resource creation, and spending spikes with graduated response actions.

Last run: 2 hours ago
Success rate: 95.1%
Playbook Steps
01Monitor real-time cost feeds for deviations exceeding 2 standard deviations
02Classify anomaly type: organic growth, misconfiguration, security, or error
03For security anomalies: immediately quarantine affected resources
04For misconfigurations: auto-remediate if pattern matches known issues
05For organic growth: alert team and update forecasts accordingly
06Generate incident ticket with full context and recommended response
07Implement temporary spending caps on affected accounts if severity is high
08Post-incident: update anomaly detection models with new pattern data
▼ View steps

Spot Instance Optimizer

$22,000/mo
Trigger: Continuous (every 15 minutes)

Continuously optimize spot instance usage by monitoring interruption rates, diversifying instance pools, and maintaining application availability.

Last run: 12 minutes ago
Success rate: 96.4%
Playbook Steps
01Monitor spot market pricing and interruption frequency across instance types
02Maintain diversified spot pools with automatic instance type substitution
03Pre-warm fallback on-demand capacity when spot interruption risk rises
04Automatically migrate workloads between spot pools based on pricing
05Implement graceful shutdown handlers for spot interruption notices
06Track spot savings vs on-demand baseline and report weekly
07Adjust bid strategies based on workload priority and deadline requirements
08Generate spot usage reports showing savings and interruption statistics
▼ View steps

Storage Tier Optimizer

$9,500/mo
Trigger: Daily: 03:00 UTC

Analyze storage access patterns and automatically migrate data to the most cost-effective storage tier based on access frequency and retrieval requirements.

Last run: 18 hours ago
Success rate: 99.7%
Playbook Steps
01Analyze S3 object access patterns over the last 30, 60, and 90 days
02Identify objects eligible for transition to Infrequent Access or Glacier tiers
03Calculate cost savings vs retrieval cost risk for each transition candidate
04Apply S3 Lifecycle policies for objects meeting automated transition criteria
05Migrate EBS volumes from gp3 to sc1/st1 for low-IOPS workloads
06Compress and deduplicate eligible datasets before archival transitions
07Monitor retrieval patterns post-transition and auto-promote if access increases
08Generate monthly storage optimization report with tier distribution analysis
▼ View steps

Network Cost Reducer

$7,200/mo
Trigger: Weekly analysis, continuous enforcement

Optimize network architecture and data transfer patterns to minimize cross-AZ, cross-region, and internet egress costs.

Last run: 2 days ago
Success rate: 98.9%
Playbook Steps
01Map data transfer flows between services, AZs, regions, and external endpoints
02Identify top data transfer cost drivers and recommend architecture changes
03Implement VPC endpoints for AWS service traffic to avoid NAT Gateway charges
04Consolidate cross-AZ traffic by co-locating communicating services
05Optimize CDN configuration to reduce origin fetches and egress costs
06Implement data compression for high-volume cross-region replication
07Review and optimize DNS query costs and Route 53 hosted zone usage
08Generate network cost heatmap showing expensive communication patterns
▼ View steps
Audit Trail

Every Action Traced

Immutable audit logs capture every action with full context: who initiated it, what autonomy level governed it, what resources were affected, and what the outcome was. Retention meets SOC 2, ISO 27001, and FedRAMP requirements.

TimestampActionLvlInitiatorTargetOutcomeSavings
2025-03-15 14:32:07 UTCInstance Right-SizeL3Agenti-0a1b2c3d4e (m5.2xl → m5.xl)Success$360/mo
2025-03-15 14:28:15 UTCTag Auto-ApplyL4Agent47 resources (us-east-1)SuccessAccuracy
2025-03-15 14:15:42 UTCRI PurchaseL2Human (J. Chen)3yr m5.xl RI x10 (us-east-1)Success$18,200/yr
2025-03-15 13:58:33 UTCDev Env ShutdownL3Agent23 instances (dev-west)Success$180/day
2025-03-15 13:45:11 UTCRDS ResizeL3Agentdb-analytics-prod (r5.2xl → r5.xl)Rolled Back-
2025-03-15 13:30:09 UTCEBS CleanupL4Agent8 orphaned volumes (1.2TB total)Success$1,240/mo
2025-03-15 13:15:55 UTCSpot MigrationL2Human (A. Patel)batch-worker fleet (15 instances)Success$3,800/mo
2025-03-15 12:58:22 UTCBudget AlertL0Agentanalytics-team budget (87% utilized)AlertedPrevention
2025-03-15 12:42:18 UTCS3 LifecycleL4Agent2.4TB → Glacier Deep ArchiveSuccess$890/mo
2025-03-15 12:30:05 UTCSecurity QuarantineL4Agenti-9z8y7x6w5 (crypto mining detected)Quarantined$2,100/day
2025-03-15 12:15:44 UTCNAT Gateway OptimizeL2Agent3 NAT GWs → NAT Instances (dev)In Progress$2,100/mo
2025-03-15 12:00:33 UTCTag Compliance SweepL4AgentAll accounts (weekly sweep)SuccessCompliance
Integration Points

Connected to Your Entire Stack

Action Fabric connects directly to cloud provider APIs, infrastructure-as-code tools, workflow platforms, and notification systems. Actions execute through native APIs — no agents, no proxies, no additional attack surface.

AWS

EC2RDSS3LambdaEKSElastiCacheCloudWatchCost ExplorerOrganizations

Azure

Virtual MachinesSQL DatabaseBlob StorageFunctionsAKSMonitorCost ManagementAdvisor

GCP

Compute EngineCloud SQLCloud StorageCloud FunctionsGKEStackdriverBillingRecommender

Kubernetes

Cluster AutoscalerVPAHPAResource QuotasLimit RangesPod Disruption Budgets

Workflow

ServiceNowJiraSlackTeamsPagerDutyOpsgenieTerraformPulumi
Technical Specifications

Built for Enterprise Scale

Action Fabric is engineered for high-throughput, low-latency execution across thousands of cloud accounts and millions of resources.

< 200ms
Action Execution Latency
From approval to API call initiation
< 30s
Rollback Time
Time to initiate and complete rollback
500+
Concurrent Actions
Parallel action execution capacity
7 years
Audit Log Retention
Immutable audit trail storage
10s
Health Check Interval
Post-action health monitoring frequency
< 50ms
Policy Evaluation
Time to evaluate action against all policies
200+
Supported Actions
Pre-built action types across all providers
< 500ms
Blast Radius Calc
Full dependency graph analysis time

Action Fabric Architecture

INPUT LAYERReasoning CorePrediction MeshSimulation LabHuman RequestScheduled TriggerACTION ENGINEPolicy EvaluatorRisk CalculatorExecution PlannerRollback ManagerEXECUTIONAWS APIsAzure APIsGCP APIsK8s API ServerTerraform/PulumiMONITORINGHealth ChecksAudit LoggerAlert SystemDashboard Feed
Why It Matters

From Knowing to Doing

Action Fabric transforms cloud management from a reactive, manual discipline into a proactive, automated one. It is the bridge between the intelligence layers of GENESIS and the real-world infrastructure they analyze.

94%
recommendation execution rate

Eliminate the Recommendation Backlog

Stop watching savings opportunities expire. Action Fabric turns recommendations into executed optimizations in minutes, not weeks.

85%
reduction in manual optimization effort

Reduce Human Toil

Free your cloud and FinOps teams from repetitive optimization tasks. Let them focus on strategy while Action Fabric handles execution.

$2.1M
average additional annual savings captured

Capture Every Dollar of Savings

With continuous, automated optimization, no savings opportunity goes unnoticed or unacted upon. The system never takes a day off.

99.4%
compliance maintenance rate

Maintain Compliance Automatically

Tag compliance, security standards, and governance policies enforced continuously — not just at audit time.

< 5 min
average anomaly response time

Prevent Cost Incidents

Real-time anomaly detection and automated response mean cost incidents are caught and remediated in minutes, not discovered in monthly reviews.

5 levels
of graduated automation maturity

Build Organizational Confidence

Graduated autonomy lets organizations build trust in automation at their own pace, with safety guardrails providing a foundation of confidence.

System Position

Where Action Fabric Sits in GENESIS

Action Fabric is System 05 of seven — the execution layer that transforms the intelligence produced by upstream systems into real-world changes, and feeds results downstream for value tracking and continuous learning.

01
Signal Fabric
02
Reasoning Core
03
Prediction Mesh
04
Simulation Lab
05
Action Fabric
06
Value Ledger
07
Trust Surface
Receives From
Reasoning Core, Prediction Mesh, Simulation Lab
05
Feeds Into
Value Ledger (outcome tracking), Trust Surface (confidence scoring)
Next in the GENESIS Architecture
06

Value Ledger

Every dollar tracked, every outcome measured — the single source of truth for cloud financial intelligence and optimization ROI.

Explore System 06 →

GENESIS System 05 — Action Fabric — AgentAAS OS