ITFOS · IT Foundry OS
ITFOS overview
An AI Native IT operations system that ties service management, infrastructure operations, asset management and security operations into one flow.
AI Native IT operations system
Four domains look at one incident together
ITFOS uses configuration data (CMDB), observability data and reasoning rules in one place. When an incident occurs, the related services, devices, software and security events are gathered under the same trace number, and AI agents rank the root cause candidates.
- 44 AI agents divide the work by domain
- Every tool run passes permission, approval and audit records in the same order
- Shows whether the data behind a decision is live, imported history or synthetic

Suited to these organizations
- Operations teams that switch between many tools to piece an incident together
- Organizations that spend a lot of time on root cause analysis and reports
- Organizations that want automated actions but need approvals and records
- Institutions that need air-gapped or on-premises installation
Common situations before adoption
- Each tool sees a different incidentMonitoring, tickets, asset registers and security devices live on separate screens, so matching one incident takes time.
- Symptoms show before causesThe loudest alert is mistaken for the cause, and the device that actually changed first is easy to miss.
- Automation versus controlThe more actions are automated, the harder it is to explain what ran and when.
ITFOS · IT Foundry OS
Key features
Reasoning on configuration data
163 configuration item classes and 48 relation types describe how services and devices connect, and impact and causes are traced along those links.
Ranked root cause candidates
It measures when each device started to drift from its usual range and puts the ones that changed before the symptom first.
AI agents with clear roles
Detection, correlation, root cause, reporting, asset and security agents each do their part and hand work to each other.
Approval connected to execution
Once approved, the action runs and the result stays on the same incident. There are 47 execution handlers.
Security operations built in
43 detection rules and 14 threat behavior sequences put security events in the same flow as operational incidents.
Imply Lumi observability data
Logs, metrics and traces are collected in Imply Lumi and queried under one consistent schema through 41 logical tables.
Tamper-evident audit records
Who, which agent, what was run and on what evidence is kept in a hash-chained record.
Connects to existing systems
35 external integrations, 16 MCP servers and 13 trigger connectors can be connected.
From detection to action
The path one incident takes. Each step has an AI agent in charge, and a person approves before any irreversible action.
Detect
Find devices outside their usual range in observability data.
Analyze cause
Walk the links upstream and put the device that changed first as a candidate.
Check impact
Gather the affected services, assets and security events.
Approve
One or two people approve, depending on the risk of the action.
Act and record
Run the action and record the result and the reasoning.
StructureThe same data is used from the data layer at the bottom to the work layer at the top.
What AI agents and people do
AI agents research, organize the evidence for decisions and prepare actions. People approve and make the final call.
| Step | What the AI agents do | What people do |
|---|---|---|
| Detect | Compare against the usual range, find anomalies and group alerts | Set alert policies and severity levels |
| Root cause | Rank root cause candidates and write down why | Confirm the cause or ask for more investigation |
| Action | Prepare the action, its blast radius and how to roll it back | Approve or reject the action |
| Report | Summarize the timeline and results into a report draft | Review and distribute the report |
Verified figures
Counted directly from the product code and test results.
44
AI agents
Source Agent registry
32
Built-in skills
Source Skill registry
190
AI agent tools
Source Tool list
95
Top-1 root cause accuracy (%)
Source Test on 60 synthetic incidents
AI agents by domainCounted by domain from the agent registry.
Delivery options
Ways to use ITFOS. Each option lists what is included, who operates what and the steps to start.
SaaS
Shared itfos.ai- What is included
- Get a workspace on itfos.ai and start right away.
- Operation
- Operated by AIONIS.
- Getting started
- Request a pilot → connect observability data → import configuration data
Installed SaaS
Installed in your environment and run by you- What is included
- Installed in your environment and connected to your company accounts.
- Operation
- You operate it, supported by AIONIS.
- Getting started
- Check the environment → install → activate the license → connect data
RolloutStart small and expand.
Connect
Connect configuration and observability data and start from an industry template.
Output Topology and data connection checklist
Observe
Run with AI agents reasoning but not executing, and watch their accuracy.
Output Root cause results and an accuracy report
Expand
Set approval procedures, turn on execution and widen the scope.
Output Approval policies and an action list
Fit ITFOS to your operations
We look at your configuration data and observability setup together and propose where to start and how to adopt.
