Agent Hub
Agent Hub is the user-facing entry point for configured FactVerse AI Agents in a tenant environment. It shows which agents are available, what each agent is intended to do, whether it is ready for use, and where the user should go next.
Use this page when planning how customers, operators, engineers, or implementation teams should discover and enter Agent workflows.
What Agent Hub should answer
| Question | Agent Hub answer |
|---|---|
| Which agents are available in this tenant? | The tenant-specific agent roster. |
| What does this agent help with? | A short capability description tied to an operating task. |
| Can the user enter it now? | Status and entry point for available agents. |
| What data does it depend on? | Data-health indicators or links to source readiness pages. |
| What happens next? | A route into a module page, dashboard, assistant, workflow, or approval path. |
Agent Hub works as a live product entry point: it reflects the agents that are configured and useful in the current environment.
Agent card model
Each agent card should keep a small set of fields visible.
| Field | Purpose |
|---|---|
| Agent name | The customer-facing name users recognize. |
| Status | Whether the agent is available for use, in rollout, or unavailable. |
| Capability summary | The operating task the agent supports. |
| Entry point | Where the user goes to work with the agent. |
| Data-health signal | Whether required upstream data appears ready, delayed, missing, or under review. |
| Review boundary | Whether the agent produces read-only output, drafts, or actions that need approval. |
Keep names practical. A card title should use a business capability or operating role that users can recognize.
Status behavior
| Status | Product behavior |
|---|---|
| Available | The user can open the agent entry point and run the supported workflow. |
| Limited | The agent can be viewed, but data, scope, or workflow readiness limits what it can do. |
| Planned | The card can point users to the right product owner or rollout note when future capability is visible in the tenant. |
| Empty | If no agents are configured for a tenant, show a tenant-specific empty state. |
A tenant-specific empty state keeps the Hub aligned with the current environment.
Data health in Agent Hub
Agent usefulness depends on data readiness. Agent Hub should surface data-health context when the agent depends on live or recently synchronized sources.
| Source type | Examples |
|---|---|
| Platform records | tenant, site, asset, equipment, location, permissions |
| DFS data | connector sync status, mapped fields, quality checks, fusion task output |
| Inspector records | alarms, inspections, work orders, operator notes |
| Designer context | scene version, model asset version, SimReady package, simulation setup |
| Documents | manuals, SOPs, procedures, evidence files |
When data health is weak, the user should see Data Readiness or the relevant module documentation before running the workflow.
Tool visibility
Agent Hub is connected to runtime tool discovery. A tenant should see the tools that are available to its configured agents, enabled modules, and access scopes.
This means:
- a tool can exist in the product while remaining visible only to agents with the right binding and scope;
- a module can be enabled but still expose only the tools configured for that workflow;
- a write-capable tool can be visible but still require confirmation or human review before execution;
- missing tools should be investigated through endpoint, scope, module enablement, and agent binding.
Use MCP Tool Reference for planning and runtime discovery for the live environment.
User flow
Design checks for a new Agent Hub entry
Before adding an agent to the Hub, confirm:
- the agent has a clear operating task and owner;
- the user can understand the capability through clear names and expanded terms;
- required data sources and source owners are known;
- the entry point exists and matches the user's workflow;
- available tools match the intended risk boundary;
- write actions have a confirmation or approval path;
- completed work can be traced to an operating record.
Use Platform Overview for the overall model and Workflow Guides when designing task-specific pages.