The AI Operating System for Managed IT and Business Operations

AI Command Center

Core plan

Kellions AI Elevate 2.0

AI Command Center

One governed orchestrator routes work to specialist agents, prepares tenant-safe context, and requests approval before sensitive action.

Specialist agents

Selected by one intent classifier.

Governed tools

4 read-only, 1 approval-gated.

Closed-loop stage

Knowledge review

KB-8328

Orchestrator

Agent routing and policy enforcement

Tenant-safe

Kellions Orchestrator

customer support

Classify intent · Prepare tenant context · Select governed tools

Service Desk Agent

service desk

Classify tickets · Detect duplicates · Recommend escalation

Knowledge Agent

knowledge

Search knowledge · Draft articles · Improve runbooks

IT Operations Agent

operations

Analyse alerts · Inspect assets · Recommend remediation

Security Agent

security

Explain vulnerabilities · Prioritise exposure · Prepare remediation

Automation Agent

automation

Suggest triggers · Build workflow drafts · Route approvals

Closed-loop support

Customer question to reviewed knowledge

First-class workflow

AI searches authorised knowledge
AI · semantic_knowledge_search
completed
AI searches resolved tickets
AI · search_resolved_tickets
completed
AI proposes troubleshooting
AI
completed
Unresolved issue becomes a support ticket
AI · TKT-4908
completed
Technician resolves ticket with notes
Technician · TKT-4908
completed
Knowledge Agent drafts article candidate
AI · KB-8328
completed
Knowledge Manager reviews and publishes
Knowledge Manager · KB-8328
active

Governed AI tool registry

Every tool declares permission, tenant scope, risk, timeout and audit behaviour.

Tool	Permission	Risk	Approval	Audit
semantic_knowledge_search Search authorised articles, runbooks, resolved tickets and troubleshooting guides.	knowledge.read	read	Automatic	read audit
search_resolved_tickets Retrieve similar resolved tickets within the current tenant for support guidance.	support.read	read	Automatic	read audit
create_support_ticket Create a support ticket from an unresolved AI support conversation.	support.manage	low	Automatic	execution audit
draft_knowledge_article Create a draft knowledge article from a resolved ticket and technician notes.	knowledge.write	low	Automatic	execution audit
security_risk_prioritisation Prioritise findings using exploitability, exposure, asset value and controls.	security.read	read	Automatic	read audit
workflow_recommendation Recommend a governed workflow with triggers, conditions, actions and approval gates.	workflows.read	read	Automatic	read audit
create_ai_action Queue a governed AI action for approval and immutable audit before execution.	workflows.manage	high	Required	approval audit

Infrastructure Agent

Low-risk remediation

Pending approval

Status

Active

Confidence

82%

Mode

Explainable

Latest finding

NLC-FS-02 has a repeat CPU pattern after backup verification.

Restart the indexing service and validate CPU recovery for 15 minutes.

Investigated logs
Linked service timeout
Prepared rollback step

Security Agent

High-risk actions require human approval

Suggested

Status

Requires approval

Confidence

76%

Mode

Explainable

Latest finding

Aurora Logistics has elevated identity and endpoint exposure.

Contain two risky identities, rotate stale credentials, and prioritise missing patches.

Mapped alerts
Ranked exposure
Prepared incident summary

Support Agent

Draft and classify tickets

Running

Status

Active

Confidence

88%

Mode

Explainable

Latest finding

13 tickets match known low-risk runbooks.

Auto-draft customer replies and queue technician approval for remediation.

Classified queue
Matched knowledge
Drafted updates

Compliance Agent

Evidence suggestions only

Suggested

Status

Learning

Confidence

71%

Mode

Explainable

Latest finding

Cyber Essentials evidence is complete for 78% of mapped controls.

Request MFA screenshots and device encryption exports from two customers.

Mapped controls
Checked evidence age
Flagged gaps

Finance Agent

Read-only financial analysis

Completed

Status

Active

Confidence

84%

Mode

Explainable

Latest finding

£3,420 monthly software waste identified across inactive users.

Create licence optimisation proposals for Meridian and Aurora.

Analysed licences
Calculated savings
Generated proposal draft

Sales Agent

Pipeline analysis and proposal drafts

Suggested

Status

Active

Confidence

79%

Mode

Explainable

Latest finding

Brightstone renewal risk and hardware refresh create a combined opportunity.

Prepare a renewal brief with a staged endpoint refresh proposal.

Scored renewal risk
Linked lifecycle signal
Drafted next step

Procurement Agent

Supplier comparison and purchase recommendations

Suggested

Status

Learning

Confidence

73%

Mode

Explainable

Latest finding

Backup capacity and laptop warranties need supplier review this quarter.

Request quotes and route purchase decisions through approval policy.

Checked warranties
Matched supplier category
Calculated budget impact

Executive Agent

Board and KPI report drafts

Completed

Status

Active

Confidence

81%

Mode

Explainable

Latest finding

Service margin is falling where ageing endpoints drive repeat support work.

Add lifecycle investment and automation approvals to the weekly operating brief.

Combined service metrics
Ranked business risk
Prepared executive summary