AI Hub and agents
Purpose and map of the hub
Section titled “Purpose and map of the hub”The AI Hub brings together the platform’s artificial intelligence capabilities: the agents (governance, orchestration, monitoring, RBAC, tracing), the domain experts (RCA, security, HR), RAG (retrieval), AI governance (Judge/Sentinels), the process optimiser and the AI panels on tickets.
Consolidation (important). The agent pages have been consolidated: Agent Registry, Agent Command Center and Monitoring Agents have merged into two main consoles — the Agentic AI Control Tower (Agentic AI Control Tower) and the Agentic AI Operations Console (Agentic AI Operations Console). The number of agents has been rationalised (consolidation 218→53).
AI Hub landing page
Section titled “AI Hub landing page”What it is for. It is the entry point to the AI functions: it collects the tiles for Control Tower, Operations Console, agent RBAC, workflow trace, experts, RAG, Sentinels, Service Quality Analyzer, Ticket AI Panel, Process Optimizer and Flow Engine.
When to use it. At the start of your AI session, to choose the tool.
How to open it. From the Core Home (AI Hub tile).
What it shows and how to use it. The tiles for: Agent Control Tower, Agent RBAC Config, Workflow Trace, Agentic Ops Console, Flow Engine, HR Expert, Process Optimizer, RAG Admin, RCA Expert, Sec Expert, Sentinels, Service Quality Analyzer, Ticket AI Panel.
What happens next. The chosen tool opens.
Notes. AI visibility and actions are subject to governance (Sentinels / Judge Governance).
Agentic AI Control Tower
Section titled “Agentic AI Control Tower”What it is for. It is the agent control tower: registry, enablement, operational supervision and monitoring (it consolidates Registry and Monitoring).
When to use it. To see which agents exist, their status and their activity.
How to open it. From the AI Hub, Control Tower tile.
What it shows and how to use it. The registry of available agents with status and configuration, and monitoring (runs, health, errors). From here you enable/configure agents and oversee their operation.
What happens next. Enabled agents operate according to their RBAC (Agent RBAC Config).
Notes. It can be opened with a hub focus, for example on security.
Agentic AI Operations Console
Section titled “Agentic AI Operations Console”What it is for. It is the agentic operations console: runtime governance of executions, with queues, timeline, traces, live monitoring, proactive actions, human-in-the-loop, safety and costs (it consolidates the Command Center).
When to use it. To orchestrate and oversee in real time what the agents are doing.
How to open it. From the AI Hub, Agentic Ops Console tile.
What it shows. A console in tabs:
| Tab | Content |
|---|---|
| Queue | The queue of agentic requests |
| Timeline | The history of runs |
| Trace | Step-by-step tracing of a run |
| Mon live | Real-time monitoring |
| Proactive | Proactive actions suggested/started by the agents |
| HITL | Human-in-the-loop: the human interventions required |
| Safety | Safety checks/guardrails |
| Cost | The costs of AI runs |
| Ask/Route | Routing requests to the right agent/tool |
How to use it. You monitor the queue and the runs, step in at the human-in-the-loop points, and check safety guardrails and costs.
What happens next. Runs produce their effects; HITL interventions unblock or correct executions.
Notes. The console is the operational core of agentic ops.
DAG Editor
Section titled “DAG Editor”What it is for. It is the editor of agentic workflows as a DAG (directed acyclic graph).
When to use it. To compose multi-step agent flows.
How to open it. From the AI Hub / Control Tower.
What it shows and how to use it. The DAG canvas with nodes (steps/agents) and edges (dependencies); you compose and connect the steps.
What happens next. The DAG you define can be run and traced (Workflow Analyzer / Trace).
Notes. It defines multi-agent orchestration.
Workflow Analyzer / Trace
Section titled “Workflow Analyzer / Trace”What it is for. It is the run analyser for agentic workflows: the trace of steps and decisions.
When to use it. To understand what a workflow did and why (debug/audit).
How to open it. From the AI Hub / from the Operations Console (Trace tab).
What it shows and how to use it. The sequence of steps of a run with input/output and decisions.
What happens next. The analysis guides the correction of flows.
Notes. Essential for explainability.
Agent RBAC Config
Section titled “Agent RBAC Config”What it is for. It configures agent RBAC: what each agent can do (principle of least privilege).
When to use it. To define and limit agent capabilities.
How to open it. From the AI Hub, Agent RBAC Config tile.
What it shows and how to use it. The agents with their allowed permissions/actions; you enable/deny capabilities.
What happens next. Agents operate within the boundaries you define.
Notes. Complements user RBAC (The RBAC model), but specific to agents.
Supervisor
Section titled “Supervisor”What it is for. It is the supervisor that governs agent decisions (human approval/override of critical actions).
When to use it. To oversee and, if needed, correct automated decisions.
How to open it. From the AI Hub / Operations Console.
What it shows and how to use it. The agent decisions awaiting supervision and the override tools.
What happens next. Approved decisions proceed; blocked ones are corrected.
Notes. Part of human-in-the-loop governance.
Sentinels / Judge Governance
Section titled “Sentinels / Judge Governance”What it is for. It is AI governance: the «sentinels» (Judge) that assess the explainability and bias of responses/decisions.
When to use it. To check the quality and fairness of AI decisions.
How to open it. From the AI Hub, Sentinels tile.
What it shows and how to use it. The Explainability and Bias assessments (Judge Governance) of agent responses/decisions.
What happens next. Negative assessments trigger checks/blocks.
Notes. Sentinel→Judge with Explainability + Bias.
RAG Management
Section titled “RAG Management”What it is for. It administers RAG (Retrieval-Augmented Generation): sources, ingestion, embeddings, search and retrieval quality.
When to use it. To manage the vector knowledge base that feeds AI responses.
How to open it. From the AI Hub, RAG Admin tile.
What it shows. A console in tabs:
| Tab | Content |
|---|---|
| Overview | RAG summary |
| Sources | The knowledge sources |
| Ingest | Document ingestion |
| Embed | Embedding generation |
| Search | Semantic search testing |
| Quality | Retrieval quality |
| Tenants | Per-tenant configuration |
How to use it. You configure sources, ingest documents, generate embeddings and check search quality.
What happens next. A well-fed RAG improves the responses of experts and agents.
Notes. Vectorize engine; per-tenant configuration.
RCA Expert
Section titled “RCA Expert”What it is for. It is the expert assistant for Root Cause Analysis.
When to use it. While analysing a problem/incident, to get help with the RCA.
How to open it. From the AI Hub, RCA Expert tile.
What it shows and how to use it. The conversational interface that, given a case, proposes root cause hypotheses and analysis steps.
What happens next. Assisted analysis speeds up the RCA (Problem Management Console).
Notes. Responses are subject to AI governance.
Security Expert
Section titled “Security Expert”What it is for. It is the security expert assistant.
When to use it. For security analysis and questions (threats, remediation, best practice).
How to open it. From the AI Hub, Sec Expert tile.
What it shows and how to use it. The conversational interface on security topics, with access to the relevant context.
What happens next. It supports SOC/XDR work (Security Hub).
Notes. Subject to AI governance.
HR Expert
Section titled “HR Expert”What it is for. It is the HR domain expert assistant (policies, processes, data).
When to use it. For HR questions and analysis.
How to open it. From the AI Hub or the HR Hub (AI Analyzer, AI Insights, HR Expert).
What it shows and how to use it. The HR conversational interface with anti-hallucination measures.
What happens next. It supports HR operations.
Notes. See AI Analyzer, AI Insights, HR Expert.
Service Quality Analyzer
Section titled “Service Quality Analyzer”What it is for. It analyses service quality with AI (patterns, anomalies, areas for improvement), with anti-hallucination measures.
When to use it. To assess the quality of the service delivered and find improvements.
How to open it. From the AI Hub, Service Quality Analyzer tile.
What it shows and how to use it. Quality analyses on service data, with verifiable evidence (anti-hallucination: statements are anchored to the data).
What happens next. The insights guide continuous improvement actions.
Notes. Anti-hallucination binding to real data.
Ticket AI Panel
Section titled “Ticket AI Panel”What it is for. It is the AI panel that supports ticket work: suggestions, categorisation, summaries.
When to use it. While handling tickets, to speed up diagnosis and response.
How to open it. From the AI Hub, Ticket AI Panel tile (and built into the Composer, Composer, ticket creation).
What it shows and how to use it. AI suggestions on the ticket (title, priority, category, duplicates, KB, runbook) and summaries.
What happens next. The suggestions speed up the work (the operator decides).
Notes. Consistent with the Composer’s AI Copilot.
Process Optimizer
Section titled “Process Optimizer”What it is for. It optimises processes through process mining (bottlenecks, deviations).
When to use it. To find inefficiencies and improve processes.
How to open it. From the AI Hub, Process Optimizer tile.
What it shows and how to use it. Process mining analyses (GM-PROCESS-REVIEW: process, period, bottleneck, deviation rate, recommendations).
What happens next. The recommendations guide process improvement.
Notes. It links operational data and optimisation.
AI in daily work (summary)
Section titled “AI in daily work (summary)”Besides the dedicated pages, AI is built into day-to-day operations: the Composer’s AI Copilot (Composer, ticket creation), the Ticket AI Panel (Ticket AI Panel), the ITOM/Events AIOps (NOC, Observability and ITOM), the HR assistants (AI Analyzer, AI Insights, HR Expert) and the Flow Engine (Administration). All AI functions are subject to governance: explainability, bias control and human supervision (Supervisor/Judge).